awiddersheim | or haproxy changes | 00:00 |
---|---|---|
inc0 | scaling up controllers nodes *might* work after you deploy -> reconfigure | 00:00 |
inc0 | never tested it | 00:00 |
awiddersheim | ok | 00:00 |
jascott1 | sdake: just playing soundboard | 00:00 |
sdake | jascott1 right, still recommended reading :) | 00:00 |
inc0 | 3 controllers (minimum for HA) can handle 100s of compute nodes | 00:00 |
inc0 | so well... | 00:00 |
jascott1 | will read again sir! and pay attention this time even | 00:00 |
sdake | oh if you read it already | 00:01 |
sdake | nm then :) | 00:01 |
inc0 | also scaling up rabbitmqs doens't actually make it faster | 00:01 |
inc0 | I should read it myself at some point.. | 00:01 |
*** Bico_Fino has quit IRC | 00:01 | |
sdake | its short and sweet | 00:01 |
inc0 | I know, kidding | 00:01 |
awiddersheim | inc0: great thanks for the help | 00:03 |
*** Pavo has quit IRC | 00:03 | |
inc0 | np, ping me if you need anything else | 00:03 |
*** Pavo has joined #openstack-kolla | 00:03 | |
inc0 | ...for now, I'm off | 00:03 |
sdake | well that was fun :) | 00:03 |
inc0 | have a good night ladies and gents | 00:03 |
sdake | later inc0 | 00:03 |
*** inc0 has quit IRC | 00:03 | |
sdake | v1k0d3n [16:43:23] <v1k0d3n>sdake: The request you have made requires authentication. (HTTP 401) (Request-ID: req-5742fcb9-447d-4fa1-9556-82372433b7bc) | 00:05 |
sdake | v1k0d3n this still a problem? | 00:06 |
sdake | or did you sort it out | 00:06 |
v1k0d3n | no, that is part of the problem. | 00:08 |
v1k0d3n | i have correct auth. | 00:08 |
sdake | cool | 00:08 |
sdake | lets debug that | 00:08 |
sdake | first ,what type of host are you on? | 00:08 |
sdake | i've got about 5-10 mins :) | 00:08 |
sdake | run that same commnd with --debug | 00:09 |
v1k0d3n | dont worry about it. i can get this later. today is one of those days. | 00:09 |
sdake | it should spit out a bunch of crap and at the end it will tell you what went wrong | 00:09 |
sdake | v1k0d3n today is one of thos edays every day of my life it seems :) | 00:11 |
v1k0d3n | the error debug doesnt really give anymore content | 00:11 |
sdake | your looking for something with TLS or tls in it | 00:12 |
sdake | v1k0d3n reading the googles | 00:12 |
sdake | i may ahve been wrong | 00:12 |
sdake | the correct cert command may be OS_CA_CERT | 00:12 |
sdake | rather environment | 00:13 |
sdake | v1k0d3n if tls isnt' displayed inthe debug output the environment variable is definately wrong | 00:14 |
sdake | v1k0d3n reference: https://bugs.launchpad.net/python-keystoneclient/+bug/1086278 | 00:15 |
openstack | Launchpad bug 1086278 in python-keystoneclient "OS_CA_CERT is not consistent with help message" [Low,Fix released] - Assigned to Wu Wenxiang (wu-wenxiang) | 00:15 |
sdake | v1k0d3n if you click through it explains what keystoneclient does at that partuclar bug point in time | 00:16 |
sdake | so I am pretty sure its OS_CACERT | 00:16 |
sdake | but not 100% positive | 00:16 |
sdake | wife dinner time bbl | 00:22 |
*** hyang_ has joined #openstack-kolla | 00:24 | |
*** yingjun has joined #openstack-kolla | 00:27 | |
hyang_ | Hi all, I’m trying to use Kolla for all-in-one deployment of OpenStack on CentOS 7 in virtualbox. I’m following the QuickStart doc and when it comes to set up “network_interface” variable in globals.yml, I’m confused about which network interface should I use here. | 00:42 |
hyang_ | My vm has 1 NAT interface by default, but using that will get error “Hostname has to resolve to IP address of api_interface” in kolla pre-checks. I also tried to use more network interfaces of NAT, Host-only, Bridged, but still got the same error. | 00:42 |
hyang_ | Could someone point out which network interface type is the correct choice? Appreciate for any help. | 00:42 |
*** sdake has quit IRC | 00:43 | |
*** sdake has joined #openstack-kolla | 00:45 | |
*** zhubingbing_ has joined #openstack-kolla | 00:57 | |
*** hyang_ has quit IRC | 00:59 | |
*** tovin07 has joined #openstack-kolla | 00:59 | |
*** tovin07 has quit IRC | 00:59 | |
*** tovin07 has joined #openstack-kolla | 01:00 | |
*** tovin07 has quit IRC | 01:01 | |
*** sayantani01 has quit IRC | 01:01 | |
*** tovin07 has joined #openstack-kolla | 01:01 | |
*** tovin07 has quit IRC | 01:02 | |
*** tovin07 has joined #openstack-kolla | 01:02 | |
*** f13o has joined #openstack-kolla | 01:03 | |
*** sdake has quit IRC | 01:05 | |
*** sdake has joined #openstack-kolla | 01:06 | |
*** f13o has quit IRC | 01:08 | |
*** kristian__ has joined #openstack-kolla | 01:11 | |
*** tovin07_ has joined #openstack-kolla | 01:15 | |
*** kristian__ has quit IRC | 01:15 | |
duonghq | morning guys | 01:18 |
*** zhurong has joined #openstack-kolla | 01:19 | |
*** liyifeng has quit IRC | 01:20 | |
*** newmember has quit IRC | 01:23 | |
openstackgerrit | Duong Ha-Quang proposed openstack/kolla-ansible: Specify 'become' for only neccesary tasks (all other roles) https://review.openstack.org/398685 | 01:24 |
*** sdake has quit IRC | 01:27 | |
*** sdake has joined #openstack-kolla | 01:29 | |
*** dave-mccowan has joined #openstack-kolla | 01:34 | |
*** jtriley has joined #openstack-kolla | 01:38 | |
*** diogogmt has quit IRC | 01:46 | |
*** eaguilar has joined #openstack-kolla | 01:47 | |
*** sdake has quit IRC | 01:50 | |
*** prameswar has joined #openstack-kolla | 01:52 | |
*** sdake has joined #openstack-kolla | 01:53 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Remove adding repo in magnum-base Dockerfile https://review.openstack.org/401032 | 01:59 |
*** Pavo has quit IRC | 02:03 | |
*** Pavo has joined #openstack-kolla | 02:03 | |
*** prameswar has quit IRC | 02:05 | |
openstackgerrit | Merged openstack/kolla-ansible: Fix congress broken https://review.openstack.org/400623 | 02:06 |
*** liyifeng has joined #openstack-kolla | 02:08 | |
*** DaveTurner has quit IRC | 02:10 | |
*** inc0 has joined #openstack-kolla | 02:10 | |
*** dave-mcc_ has joined #openstack-kolla | 02:11 | |
*** dave-mccowan has quit IRC | 02:12 | |
*** dave-mccowan has joined #openstack-kolla | 02:13 | |
*** sdake has quit IRC | 02:14 | |
*** sdake has joined #openstack-kolla | 02:15 | |
*** dave-mcc_ has quit IRC | 02:16 | |
*** hieulq has joined #openstack-kolla | 02:18 | |
*** dave-mccowan has quit IRC | 02:18 | |
*** hieulq has quit IRC | 02:21 | |
*** hieulq has joined #openstack-kolla | 02:23 | |
*** fguillot has quit IRC | 02:27 | |
*** jascott1 has quit IRC | 02:28 | |
*** sdake has quit IRC | 02:36 | |
*** sdake has joined #openstack-kolla | 02:37 | |
*** dave-mccowan has joined #openstack-kolla | 02:39 | |
*** inc0 has quit IRC | 02:44 | |
*** dave-mccowan has quit IRC | 02:48 | |
*** tonanhngo has quit IRC | 02:53 | |
*** fragatina has quit IRC | 02:53 | |
*** diogogmt has joined #openstack-kolla | 02:56 | |
*** sdake has quit IRC | 02:57 | |
*** sdake has joined #openstack-kolla | 02:59 | |
*** zhurong has quit IRC | 03:04 | |
*** zhurong has joined #openstack-kolla | 03:06 | |
*** kollabot has quit IRC | 03:15 | |
*** kollabot has joined #openstack-kolla | 03:16 | |
*** sdake has quit IRC | 03:19 | |
*** sdake has joined #openstack-kolla | 03:21 | |
*** unicell has quit IRC | 03:34 | |
*** zhurong has quit IRC | 03:35 | |
*** zhurong has joined #openstack-kolla | 03:36 | |
*** sdake has quit IRC | 03:41 | |
*** sdake has joined #openstack-kolla | 03:43 | |
*** fragatina has joined #openstack-kolla | 03:51 | |
*** fragatina has quit IRC | 03:52 | |
*** fragatina has joined #openstack-kolla | 03:52 | |
*** Pavo has quit IRC | 04:03 | |
*** sdake has quit IRC | 04:03 | |
*** sdake has joined #openstack-kolla | 04:04 | |
*** Pavo has joined #openstack-kolla | 04:07 | |
*** mdnadeem has joined #openstack-kolla | 04:08 | |
*** jgriffith is now known as jgriffith_away | 04:10 | |
*** sayantani01 has joined #openstack-kolla | 04:22 | |
*** sdake has quit IRC | 04:25 | |
*** sdake has joined #openstack-kolla | 04:28 | |
*** eaguilar has quit IRC | 04:28 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-ansible: Updated from global requirements https://review.openstack.org/400789 | 04:31 |
*** tonanhngo has joined #openstack-kolla | 04:32 | |
openstackgerrit | Duong Ha-Quang proposed openstack/kolla-ansible: Specify 'become' to neccesary tasks (general roles) https://review.openstack.org/398682 | 04:37 |
*** sdake has quit IRC | 04:49 | |
*** sdake has joined #openstack-kolla | 04:53 | |
*** hyang_ has joined #openstack-kolla | 04:53 | |
*** hyang_ has quit IRC | 04:54 | |
sdake | sup peeps | 05:06 |
*** Jeffrey4l has quit IRC | 05:06 | |
*** sayantani01 has quit IRC | 05:06 | |
*** sayantani01 has joined #openstack-kolla | 05:07 | |
*** newmember has joined #openstack-kolla | 05:08 | |
*** unicell has joined #openstack-kolla | 05:14 | |
*** senk has joined #openstack-kolla | 05:24 | |
*** unicell has quit IRC | 05:26 | |
*** unicell has joined #openstack-kolla | 05:26 | |
*** senk_ has joined #openstack-kolla | 05:28 | |
openstackgerrit | Steven Dake proposed openstack/kolla: Add specs are advisory readme.rst to kolla repo https://review.openstack.org/400570 | 05:28 |
*** senk has quit IRC | 05:31 | |
*** sdake_ has joined #openstack-kolla | 05:31 | |
*** sdake has quit IRC | 05:35 | |
*** senk has joined #openstack-kolla | 05:37 | |
*** senk_ has quit IRC | 05:40 | |
*** yingjun has quit IRC | 05:43 | |
*** prameswar has joined #openstack-kolla | 05:49 | |
*** skramaja has joined #openstack-kolla | 05:50 | |
*** zhangshuai has quit IRC | 06:00 | |
*** Pavo has quit IRC | 06:03 | |
*** Pavo has joined #openstack-kolla | 06:07 | |
*** zhangshuai has joined #openstack-kolla | 06:12 | |
*** yingjun has joined #openstack-kolla | 06:14 | |
*** kristian__ has joined #openstack-kolla | 06:16 | |
*** jtriley has quit IRC | 06:19 | |
*** kristian__ has quit IRC | 06:20 | |
bjolo | morning | 06:33 |
sdake_ | sup bjolo | 06:33 |
bjolo | sup dake | 06:34 |
sdake_ | not much just reviewing | 06:35 |
sdake_ | review queue long | 06:35 |
bjolo | i hear ya | 06:38 |
openstackgerrit | caoyuan proposed openstack/kolla-ansible: Update the version for hacking from global requirement https://review.openstack.org/401083 | 06:38 |
bjolo | sdake_, where can i see all new PS in gerrit? | 06:39 |
sdake_ | bjolo you can add the projects you want to watch to your watched project list | 06:39 |
sdake_ | to do so, click your username then click watched projects then type in something like openstack/kolla-ansible | 06:39 |
sdake_ | i watch kolla, kolla-ansible, and kolla-kubernetes | 06:40 |
bjolo | i want to help out more, both with reviews and code, but im sooo swamped in work right now | 06:41 |
sdake_ | bjolo there is no expectation that you do reviews and code :) | 06:41 |
bjolo | but i want to :) | 06:41 |
sdake_ | bjolo imo you help quite a bit in the community already | 06:41 |
bjolo | i want to take a crack on the db backup | 06:41 |
sdake_ | keep in mind this is a volunteer effort | 06:41 |
sdake_ | cool that would be pretty sweet | 06:42 |
bjolo | yes it is, but it is also a great learning place | 06:42 |
sdake_ | right, that is the magic of irc ;) | 06:42 |
*** kristian__ has joined #openstack-kolla | 06:42 | |
bjolo | ive learned so much about openstack and ansible just by hanging out here and do what i can to contribute | 06:42 |
openstackgerrit | caoyuan proposed openstack/kolla: Updated from global requirements https://review.openstack.org/401085 | 06:43 |
bjolo | about the backup... we really want this feature. Not to long ago we wanted to add some more compute nodes to our fuel based cloud | 06:43 |
bjolo | the deploy failed, which is kinda normal for fuel | 06:44 |
bjolo | so the normal procedure is to delete the node from fuel | 06:44 |
*** kristian__ has quit IRC | 06:44 | |
*** kristian__ has joined #openstack-kolla | 06:44 | |
bjolo | somehow fuel got this wrong and deleted every node in the cloud | 06:44 |
bjolo | basically wiped discs and rebooted every node. controllers, network, compute, storage | 06:45 |
*** msimonin has joined #openstack-kolla | 06:45 | |
*** senk has quit IRC | 06:46 | |
bjolo | we had local db backups on every controller node | 06:46 |
bjolo | but they got wiped as well | 06:46 |
bjolo | this was only a test cloud, but still... | 06:47 |
bjolo | even if we restored the database, whole ceph was gone anyway | 06:48 |
*** msimonin has quit IRC | 06:48 | |
bjolo | lesson learned. dont but all eggs in the same basket | 06:49 |
openstackgerrit | Merged openstack/kolla-ansible: Pull swift-object-expirer image https://review.openstack.org/400894 | 06:49 |
bjolo | sdake_, one of my patches are in confligt, but i dont really see where. | 06:56 |
bjolo | https://review.openstack.org/#/c/372678/2 | 06:56 |
*** kristian__ has quit IRC | 07:08 | |
*** senk has joined #openstack-kolla | 07:14 | |
*** sayantani01 has quit IRC | 07:18 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: Pull swift-object-expirer image https://review.openstack.org/401097 | 07:24 |
sdake_ | bjolo we split kolla and kolla-ansible repositories | 07:29 |
sdake_ | so you need to resubmit that change to kolla-ansible | 07:29 |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: typo error of keyring spelling https://review.openstack.org/401103 | 07:39 |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla: Fix neutron.conf.j2 metadata_workers spelling error https://review.openstack.org/401107 | 07:44 |
sdake_ | bjolo that makes me cry a little inside (your story about cloud deletion) | 07:44 |
*** jascott1 has joined #openstack-kolla | 07:47 | |
sdake_ | sup jascott1 | 07:47 |
sdake_ | i see you run my hours :) | 07:47 |
jascott1 | hey | 07:48 |
jascott1 | been a night owl since I can remember | 07:48 |
jascott1 | dont hold a candle to the hours you appear to put in tho ;) | 07:49 |
jascott1 | just got back from pre thanksgiving grocery store run | 07:50 |
jascott1 | :( | 07:50 |
*** jtriley has joined #openstack-kolla | 07:53 | |
sdake_ | just got back from dinner with wife | 07:53 |
jascott1 | nice. good meal? | 07:55 |
sdake_ | ya melting pot | 07:55 |
jascott1 | ive never been. theres one downtown but idk just never have gotten by there (or the other million places around here) | 07:56 |
jascott1 | big fan of cheese tho :D | 07:56 |
*** matrohon has joined #openstack-kolla | 07:56 | |
sdake_ | ya i dont roll wit h chese often - just pizza and cheese crisp | 07:57 |
*** tonanhngo has quit IRC | 07:57 | |
*** jtriley has quit IRC | 07:58 | |
jascott1 | if I ever give up on talking to these robots and was insane enough to start a food releated biz, i would consider a place that sells fried cheese | 07:58 |
*** jtriley has joined #openstack-kolla | 07:58 | |
jascott1 | one cool thing about portland is, if you are into something weird there is a good chance others are as well ;) | 07:59 |
sdake_ | i ahve a friend in hit ech that opened a food truck | 07:59 |
jascott1 | hows it going? | 07:59 |
sdake_ | he obviously doesn't make the kind of money he did prior | 07:59 |
sdake_ | but some times career changes are good for people | 08:00 |
sdake_ | and in this case that is true | 08:00 |
jascott1 | no doubt' | 08:00 |
jascott1 | between my fam and wifes fam and friends we have about 20 people here from our hometown in MS | 08:01 |
jascott1 | down there we have 'slug burgers' | 08:01 |
*** tonanhngo has joined #openstack-kolla | 08:01 | |
jascott1 | which we all think would be a hit ina food cart | 08:01 |
sdake_ | say jascott1 michal said your working on helm related things? | 08:02 |
jascott1 | i am | 08:02 |
jascott1 | trying to figure it all out | 08:02 |
jascott1 | whats up? | 08:02 |
*** Pavo has quit IRC | 08:03 | |
*** jtriley has quit IRC | 08:03 | |
*** tonanhngo has quit IRC | 08:03 | |
jascott1 | im supposed to be xpert for our team on k8s/helm | 08:04 |
sdake_ | jascott1 nice | 08:05 |
sdake_ | jascott1 since your an expert and kevin is an expert, it would be nice to get you two working together | 08:05 |
sdake_ | i use the term expert here loosly :) | 08:05 |
jascott1 | we have definitely had some good conversations | 08:05 |
jascott1 | me too! | 08:05 |
sdake_ | since this is all new territory | 08:05 |
sdake_ | would you mind reviewing his helm implementation | 08:05 |
sdake_ | https://review.openstack.org/#/c/396296/ | 08:05 |
jascott1 | no doubt and changing every few mins | 08:05 |
jascott1 | yes I have been looking at it but not since last week | 08:06 |
jascott1 | i do need to catch him cause I have a couple questions | 08:06 |
sdake_ | the way to do that is via gerrit :) | 08:07 |
*** Pavo has joined #openstack-kolla | 08:07 | |
sdake_ | are you familiar with the tool? | 08:07 |
jascott1 | true | 08:07 |
jascott1 | yes and have been using but dont have a ton of experience with it yet | 08:07 |
sdake_ | cool | 08:08 |
sdake_ | it takes a little bit to get the hang of it | 08:08 |
jascott1 | i heard bad things but so far its been pretty cool | 08:08 |
sdake_ | but once you do, its pretty straightforward | 08:08 |
jascott1 | especially seeing the PS revisions | 08:08 |
jascott1 | ya takes a minute to remember the git workflow with the ammends | 08:09 |
jascott1 | so used to regular git feature branch approach | 08:09 |
sdake_ | yup - to work successfully in openstack key is to learn the workflow | 08:10 |
sdake_ | which is pretty straightforwrd | 08:10 |
jascott1 | yes | 08:10 |
sdake_ | have prior experience in openstck workflow? | 08:10 |
jascott1 | no but have been trained at Rackspace | 08:10 |
jascott1 | i 'get it' but need more experience actually doing the things | 08:11 |
sdake_ | ok - i dont know what trined at rax means :) | 08:11 |
jascott1 | haha | 08:11 |
*** berendt has joined #openstack-kolla | 08:12 | |
berendt | morning guys | 08:12 |
jascott1 | so Intel and Rackspace have a partnership, OSIC and when I joined they sent me Rackspace for 3 weeks to train | 08:12 |
sdake_ | sup berendt | 08:12 |
jascott1 | good morning | 08:12 |
sdake_ | experience > training | 08:12 |
jascott1 | thats the usual for me | 08:13 |
sdake_ | i see 27 patch reviews on that helm chart | 08:13 |
jascott1 | RTFM -> bang head on keyboard -> repeat | 08:13 |
sdake_ | roger that :) | 08:13 |
sdake_ | it would be helpful to see your reviews | 08:13 |
jascott1 | will do | 08:14 |
sdake_ | more minds working together > less minds working together | 08:14 |
jascott1 | i wanted to talk w him again to get more of a sense of where he is going with the layout in wrt to installing things | 08:14 |
jascott1 | ya weve had some good talks so far on it | 08:15 |
sdake_ | ya the downside of that is all that context is lost to the other reviewers | 08:15 |
jascott1 | good point | 08:15 |
sdake_ | so other reviewer dont know where you think rpoblems are in the implementation | 08:16 |
sdake_ | reviewers | 08:16 |
sdake_ | you see sbezverk and kfox1111 in that review have alot of conversation going on | 08:16 |
sdake_ | I can spend 30 minutes reading the review log | 08:16 |
sdake_ | and then make a decision whether to +2 the patch or not | 08:16 |
sdake_ | otherwise if it needs further revision leave a -1 on the patch | 08:17 |
jascott1 | right | 08:17 |
sdake_ | -1 = signal that further revision is required | 08:17 |
sdake_ | if the core reviewers can't see that work happening in the review queue, we can't actually make a determination if the work is solid enough to merge | 08:18 |
jascott1 | sure | 08:18 |
sdake_ | cool if i'm repating stuff you alredy know just ignore me :) | 08:19 |
sdake_ | you know how to use the comment feature? | 08:19 |
jascott1 | all good man. thanks for the nudge! | 08:19 |
jascott1 | yeah the inline one? | 08:19 |
sdake_ | yup | 08:19 |
jascott1 | funny I did that on the helm spec but it didnt make it to the next revision | 08:19 |
jascott1 | wasnt a big deal at all tho | 08:20 |
*** senk has quit IRC | 08:20 | |
sdake_ | ya specs are an odd thing | 08:20 |
sdake_ | different then implementation | 08:20 |
sdake_ | specs = design | 08:20 |
sdake_ | lots of commentary on design = hard to merge everyones thoughts into one coherent document | 08:20 |
*** derekjhyang has quit IRC | 08:21 | |
*** senk has joined #openstack-kolla | 08:21 | |
sdake_ | sometimes comments are missed | 08:21 |
jascott1 | ya gotta admit feeling new to openstack and kolla (and somewhat gerrit) has made me more hesitant to jump in | 08:21 |
jascott1 | BUT | 08:21 |
sdake_ | The way I deal with this is to always use "reply done" when i've addressed a comment | 08:21 |
jascott1 | pretty soon you will be saying 'oh that guy again' | 08:22 |
jascott1 | ah | 08:22 |
sdake_ | doubtful :) | 08:22 |
sdake_ | if you are an expert on helm an k8s, your input would be much valued in that review | 08:23 |
jascott1 | will do | 08:23 |
jascott1 | think I will rock that first thing in the morning :) | 08:24 |
sdake_ | sounds good | 08:24 |
sdake_ | i'm about beat for the day too | 08:24 |
jascott1 | hey whats your opinion on our test 'coverage'? | 08:24 |
jascott1 | could we be doing more or its alright or? | 08:24 |
*** tonanhngo has joined #openstack-kolla | 08:24 | |
sdake_ | gating is a continous improvement process | 08:25 |
jascott1 | sorry thats kinda dumb question since we could always do more | 08:25 |
*** tonanhngo has quit IRC | 08:25 | |
*** senk_ has joined #openstack-kolla | 08:25 | |
jascott1 | yeah ive learned alot from setup_gate.sh and friends | 08:26 |
sdake_ | pragmatically speaking writing gating is the hardest part of development | 08:26 |
sdake_ | especially functional tests | 08:26 |
jascott1 | still trying to grok enough ceph for kk8s purposes | 08:26 |
sdake_ | we struggle to have the minimal gates we have in place operational | 08:27 |
*** Serlex has joined #openstack-kolla | 08:28 | |
*** Satya_ has joined #openstack-kolla | 08:28 | |
*** senk has quit IRC | 08:28 | |
jascott1 | yeah id like to be able to help out there. ill spend some more time on it | 08:28 |
sdake_ | imo not the place to start ;) | 08:29 |
jascott1 | something todo in parallel tho right | 08:29 |
jascott1 | ive done previous CI pipeline work with docker | 08:30 |
sdake_ | i dont think gating is our most pressing issue atm | 08:30 |
jascott1 | internally and at AWS | 08:30 |
*** shardy has joined #openstack-kolla | 08:30 | |
jascott1 | imean not *for* AWS | 08:30 |
jascott1 | not moving to SEA | 08:30 |
*** msimonin has joined #openstack-kolla | 08:31 | |
openstackgerrit | Surya Prakash Singh proposed openstack/kolla-ansible: use uuidutils instead of uuid.uuid4() https://review.openstack.org/399847 | 08:31 |
jascott1 | its was very helpful to study the gate, helped me to start on getting a minimal ceph going | 08:31 |
sdake_ | yup, our development environment is not ideal | 08:32 |
jascott1 | i did goto an AWS mixer a couple months ago when intel was whacking ppl left and right | 08:32 |
jascott1 | any huge gaps with the env that jump out? | 08:33 |
jascott1 | oh wow that was 6 months ago | 08:33 |
sdake_ | jascott1 an ideal dev environment would be something that launches multiple vms on a machine in an automated way and deploys kubernetes within with whatever nodecount specified | 08:35 |
jascott1 | multinode | 08:37 |
sdake_ | jascott1 v1k0d3n mentioned earlier today he has a dev environment almost ready to go with deploys ceph for persistent volumes for k8s | 08:37 |
sdake_ | he mentioned that PVC is broken in the implementation | 08:37 |
sdake_ | duonghq pingola | 08:38 |
jascott1 | yes my team is trying to help out so Ive been following. I started glance chart but the PVC has hindered some | 08:38 |
duonghq | hi sdake_ | 08:38 |
sdake_ | sup duonghq | 08:38 |
jascott1 | kollabot hi | 08:39 |
kollabot | Hey | 08:39 |
sdake_ | jascott1 cool, so my team yoru team its team - we are the kolla team :) | 08:39 |
jascott1 | yes! | 08:39 |
jascott1 | btw really digging our little community here | 08:39 |
jascott1 | glad to help out with whatever so please ask if I can help with something | 08:39 |
duonghq | I'm looking forward to kolla-k8s base stuff and doc, guess that today irc meeting is a very interesting meeting | 08:40 |
jascott1 | like the helm review :) | 08:40 |
sdake_ | jascott1 i did and you mentioend you would review | 08:40 |
jascott1 | i shall | 08:40 |
sdake_ | duonghq right, in the meantime mind taking a look at https://review.openstack.org/#/c/396296/ | 08:41 |
duonghq | sure | 08:41 |
sdake_ | jascott1 if you have a glance chart, another way you can help is to submit it to gerrit | 08:42 |
duonghq | btw, you know Brazilian? http://www.dicionarioinformal.com.br/pingola/ (j/k) | 08:42 |
sdake_ | jascott1 typically we use blueprints to track thi stuff | 08:42 |
jascott1 | it was for v1k0d3ns repo, i was checking out their implementation | 08:42 |
sdake_ | duonghq dont knwo brazillian | 08:42 |
jascott1 | but its all for kolla-k8s | 08:42 |
jascott1 | they are just poc'ing over there and have a deadline | 08:42 |
jascott1 | everyone is templating differently tho | 08:43 |
jascott1 | do we need to have a come to jeebus meeting | 08:43 |
sdake_ | pretty much - in a review :) | 08:43 |
sdake_ | jascott1 what other deltas do you notice | 08:43 |
sdake_ | i'm not particularly up to speed on helm | 08:44 |
jascott1 | ya I was thinking about carving up a gist for the meeting or post meeting | 08:44 |
jascott1 | should do that right quick | 08:44 |
sdake_ | a gist containing which? | 08:44 |
jascott1 | opens around different approaches | 08:44 |
sdake_ | best way to do that is in the review process | 08:45 |
sdake_ | if you could submit your glance implementation for helm that would be helpful | 08:45 |
sdake_ | the outcome of that is that templating is done in one way | 08:47 |
sdake_ | whichever way is sorted in the review queue for the kolla-kubernetes repository | 08:47 |
duonghq | sdake_, sorry I cannot recall atm, but what is our etherpad to keeptrack who do which on kolla-k8s? | 08:48 |
sdake_ | duonghq thats what launchpad is for dude :) | 08:48 |
sdake_ | duonghq https://launchpad.net/kolla-kubernetes/+milestone/occata-1 | 08:49 |
duonghq | so, as I can see, we do not have work items for kolla-k8s bp: helm-packaging? | 08:49 |
sdake_ | duonghq can you fix that pls | 08:49 |
*** jmccarthy has quit IRC | 08:49 | |
jascott1 | sdake_ yes all good w all of that. | 08:49 |
*** jmccarthy has joined #openstack-kolla | 08:50 | |
*** jmccarthy has left #openstack-kolla | 08:50 | |
duonghq | as I understand, from now and then, we use operator for manage our cluster? | 08:50 |
*** jmccarthy has joined #openstack-kolla | 08:50 | |
duonghq | sdake_, I think helm-packaging should be targeted to o-1? | 08:51 |
*** liyifeng has quit IRC | 08:51 | |
duonghq | or we need more PoC? | 08:52 |
sdake_ | duonghq well we ae actually in o2 atm | 08:52 |
duonghq | ah, sorry | 08:52 |
sdake_ | but launchpad is not really up to date | 08:52 |
sdake_ | we need to fix that | 08:52 |
jascott1 | where does "Mongodb Integration" come in? | 08:53 |
sdake_ | jascott1 who knows :) | 08:53 |
sdake_ | are you working on that part related to glance? | 08:53 |
jascott1 | or fit in rather | 08:53 |
sdake_ | mongodb is a dep for a bunch of services unrelated to compute kit | 08:53 |
jascott1 | no i just noticed it on the blueprint list | 08:53 |
jascott1 | oh ha kevin beat me to it "Why? The only thing I know of that uses mongodb is ceilometer, and I believe they deprecated it? - kfox1111" | 08:54 |
sdake_ | ya they moved to gnocchi | 08:54 |
sdake_ | we adapt to our upstreams in kolla | 08:54 |
sdake_ | that blueprint can likely be deprecated | 08:54 |
*** tonanhngo has joined #openstack-kolla | 08:54 | |
duonghq | do we follow this: https://releases.openstack.org/ocata/schedule.html | 08:55 |
sdake_ | jeffrey4l is the foremost expert in how all that stuff works - so i'd rather let him deprecate it :) | 08:55 |
sdake_ | duonghq thats the idea | 08:55 |
sdake_ | duonghq since o1 has shipped | 08:55 |
duonghq | so, the helm-packaging should be the same milestone as operator-base? | 08:55 |
Serlex | Morning, I see nova_compute is not started if ironic is enabled. why so? | 08:56 |
sdake_ | duonghq the way we typically handle launchpad is we put everything in o2 | 08:56 |
duonghq | ah, now I can recall that the 4.0.0 is tagged :) | 08:56 |
sdake_ | 4.0.0.0b1 of kolla-ansible | 08:56 |
sdake_ | we are working on 1.0.0.0b2 of kolla-kubernetes | 08:56 |
*** tonanhngo has quit IRC | 08:57 | |
sdake_ | 4.0.0.0b1 of kolla repo | 08:57 |
duonghq | can we add o-2 to our projects? | 08:58 |
*** jtriley has joined #openstack-kolla | 08:59 | |
sdake_ | duonghq sure what date you got for it | 09:00 |
sdake_ | Serlex no clue | 09:01 |
sdake_ | Serlex file a bug maybe someone will triage it :) | 09:01 |
sdake_ | duonghq i can add o2 - what is the schedule date | 09:02 |
duonghq | I think if we want to keep up with OpenStack schedule, it's Dec 12-16? | 09:02 |
sdake_ | cool | 09:02 |
sdake_ | so thats the 15th - wednesday | 09:02 |
sdake_ | release team likes releases on the 15th | 09:02 |
sdake_ | rather on wednesdays :) | 09:02 |
duonghq | roger :) | 09:03 |
sdake_ | duonghq your a member of kolla-drivers correct? | 09:04 |
duonghq | yeah | 09:04 |
sdake_ | jascott1 your a member of kolla-drivers? | 09:04 |
jascott1 | yes | 09:04 |
sdake_ | cool | 09:04 |
sdake_ | i could use a bit of help moving stuff from o1 to o2 that is unfinished | 09:05 |
*** jtriley has quit IRC | 09:05 | |
sdake_ | we file a bunch of blueprints, target them for a release, and if they miss, we kick em down the road | 09:05 |
sdake_ | o2 is created | 09:05 |
sdake_ | can both of you help me move the unfinished stuff to o2? | 09:06 |
jascott1 | sure | 09:06 |
sdake_ | i grow weary of pushing launchpd around | 09:06 |
sdake_ | thanks | 09:06 |
sdake_ | we can do it all together :) | 09:06 |
jascott1 | 10 each? | 09:07 |
sdake_ | i'll start at bottom 3rd, jascott top 3rd, duong somewhere in middle | 09:07 |
jascott1 | ha ok | 09:07 |
sdake_ | dont move the ones that are implemented | 09:07 |
duonghq | sdake_, roger, but the ocata is ocata, not occata | 09:07 |
*** rmart04 has joined #openstack-kolla | 09:07 | |
sdake_ | duonghq moment i'll fi that typo | 09:07 |
duonghq | roger | 09:07 |
*** Satya_ has quit IRC | 09:08 | |
jascott1 | im looking at how to do it :) | 09:08 |
jascott1 | found it | 09:08 |
jascott1 | uh | 09:09 |
jascott1 | i changed 2 and now I get 'lost something' | 09:09 |
Serlex | ok sdake | 09:09 |
duonghq | jascott1, the milestone's name is changed | 09:11 |
jascott1 | hmm ok im just working off of 'all blueprints' now | 09:12 |
sdake_ | ya sorry about that - i marked the o1 inactive accidentally | 09:13 |
sdake_ | launchpad takes some experience to manage properly | 09:13 |
sdake_ | stil make mistakes 5 years later :( | 09:13 |
*** tonanhngo has joined #openstack-kolla | 09:15 | |
*** tonanhngo has quit IRC | 09:15 | |
sdake_ | nice faster with 3 then with 1 :) | 09:16 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-ansible: Updated from global requirements https://review.openstack.org/400789 | 09:16 |
jascott1 | fan out! | 09:16 |
openstackgerrit | Zhigang Li proposed openstack/kolla-kubernetes: Replace basestring with six.string_types https://review.openstack.org/401144 | 09:16 |
sdake_ | winning :) | 09:16 |
duonghq | not much for o-1 | 09:16 |
jascott1 | are done? | 09:16 |
duonghq | jascott1, done in my PoV | 09:16 |
sdake_ | done with that step | 09:16 |
sdake_ | depends how much launchpad torture you want tonight :) | 09:17 |
jascott1 | heheh | 09:17 |
duonghq | guys should update current status | 09:17 |
openstackgerrit | Jack Ning proposed openstack/kolla: Add customized plugin to keystone image https://review.openstack.org/401145 | 09:17 |
sdake_ | yup lets do that | 09:17 |
jascott1 | sorry what exactly? | 09:18 |
duonghq | sdake_, iirc, we have compute-kit from newton? | 09:19 |
duonghq | do we need to move it to operator pattern? | 09:19 |
sdake_ | duonghq we can implement any way we like | 09:20 |
sdake_ | as long as we meet the spec ;) | 09:20 |
*** fragatina has quit IRC | 09:20 | |
duonghq | roger | 09:20 |
*** fragatina has joined #openstack-kolla | 09:20 | |
jascott1 | brb | 09:21 |
*** egonzalez90 has joined #openstack-kolla | 09:21 | |
egonzalez90 | morning | 09:21 |
sdake_ | sup egonzalez90 | 09:23 |
sdake_ | duonghq jascott lets work on breaking this blueprint into parts | 09:23 |
sdake_ | https://blueprints.launchpad.net/kolla-kubernetes/+spec/helm-packaging | 09:23 |
sdake_ | rather jascott1 | 09:24 |
duonghq | sdake_, you mean break it into "sub" bp? | 09:24 |
sdake_ | right | 09:24 |
sdake_ | helm for ceph | 09:24 |
egonzalez90 | sdake_: read the other day that we need to abandon changes pointing to stable/liberty to tag liberty as eol | 09:24 |
egonzalez90 | these are the changes ATM https://review.openstack.org/#/q/status:open+project:openstack/kolla+branch:stable/liberty | 09:25 |
sdake_ | egonzalez90 inc0 is your man :) | 09:25 |
sdake_ | egonzalez90 ryan wont be able to continue on kolla-kubernetes it appears | 09:25 |
sdake_ | so in the mean time i'll fill in for him | 09:26 |
sdake_ | egonzalez90 so really have no bandwidth to deal with wider project issues that are the ptl's job :) | 09:26 |
duonghq | sdake_, why rhallisey do not continue with kolla-k8s? | 09:26 |
sdake_ | duonghq its personal in nature and i don't discuss other people's personal business | 09:27 |
egonzalez90 | roger sdake_ , i just was wondering if we force abandons or request the author to abandon the change | 09:27 |
duonghq | roger | 09:27 |
sdake_ | he did a fantastic job getting us rolling | 09:27 |
egonzalez90 | yeah, a great job! | 09:27 |
sdake_ | egonzalez90 force abandons are rude however if this is an opentack process i'd typicllly consult ttx on the matter | 09:28 |
sdake_ | egonzalez90 but ttx can't have a million people consulting with him, so inc0 cna do that :) | 09:28 |
sdake_ | so scope of kolla-kubernetes is compute kit | 09:29 |
sdake_ | inc0 set context he would like to see it done by ocata | 09:30 |
sdake_ | so thats what we are doing now | 09:30 |
sdake_ | duonghq the scope answer was for you | 09:31 |
duonghq | thank you sdake_ | 09:31 |
sdake_ | i think we need ceph in the mix too from what i gather :) | 09:31 |
egonzalez90 | sdake_: i will join k8s for christmas, i dont have enough free time to be in ansible and k8s at the same time now. | 09:31 |
sdake_ | egonzalez90 roger volunteer project - work on what you want :) | 09:32 |
jascott1 | wasnt ryan working on operator? | 09:33 |
sdake_ | jascott1 yup | 09:33 |
egonzalez90 | yup i want to work on k8s too, but need more time | 09:33 |
sdake_ | egonzalez90 cool - would be good to see everyone work on all the projects :) | 09:33 |
sdake_ | rather deliverables | 09:33 |
sdake_ | duonghq you got a date for o3? | 09:34 |
duonghq | sdake_, Jan 23-27 | 09:34 |
sdake_ | thanks so jan 25 then | 09:34 |
duonghq | ya, | 09:35 |
duonghq | I must go to my class in about 20mins, see you guys in today meeting | 09:38 |
*** yingjun has quit IRC | 09:38 | |
*** yingjun has joined #openstack-kolla | 09:38 | |
duonghq | move through Hanoi traffic jam | 09:38 |
*** strigazi_AFK is now known as strigazi | 09:39 | |
jascott1 | good luck! | 09:39 |
sdake_ | duonghq enjoy | 09:40 |
*** gfidente has joined #openstack-kolla | 09:40 | |
sdake_ | jascott1 willing to pitch in and finish the job on organizing launchpad appropriately | 09:41 |
jascott1 | sure | 09:41 |
sdake_ | cool | 09:41 |
jascott1 | can we do in ~20 mins? | 09:41 |
jascott1 | prob force myself to get some sleep | 09:41 |
jascott1 | but yea lets get er done | 09:41 |
duonghq | jascott1, I've daily report to be done on my deck, sorry | 09:42 |
jascott1 | np we will knock it out | 09:42 |
sdake_ | i think it will take more then 20 minutes | 09:42 |
jascott1 | 40? | 09:42 |
jascott1 | np lets rock it | 09:42 |
sdake_ | but if people have to drop off thats cool | 09:42 |
*** yingjun has quit IRC | 09:43 | |
jascott1 | what are we doing? | 09:43 |
sdake_ | so the trick here is figuring out what is essential | 09:43 |
sdake_ | and what isn't | 09:43 |
sdake_ | to deliver compute kit | 09:44 |
sdake_ | according to the specification | 09:44 |
jascott1 | werd | 09:44 |
sdake_ | the stuff that isn't essential we can put in discussion state | 09:44 |
jascott1 | spec for compute kit? I havent seen it | 09:44 |
*** tonanhngo has joined #openstack-kolla | 09:44 | |
sdake_ | the kolla-kubernetes architecture specification | 09:44 |
jascott1 | oh ok | 09:44 |
sdake_ | scope limited to compute dkit | 09:45 |
jascott1 | ok | 09:45 |
*** tonanhngo has quit IRC | 09:46 | |
sdake_ | swift - not part of compute kit | 09:46 |
jascott1 | the ones listed here? https://blueprints.launchpad.net/kolla-kubernetes/+spec/helm-packaging | 09:47 |
jascott1 | thats what I have been thinking compute kit means | 09:47 |
sdake_ | yes those | 09:48 |
sdake_ | pretty much :) | 09:48 |
jascott1 | cool | 09:48 |
sdake_ | that is a meta-blueprint | 09:48 |
sdake_ | duonghq do you mind dropping work on swift to focus on compute kit ? | 09:48 |
duonghq | sdake_, np | 09:49 |
openstackgerrit | Zhigang Li proposed openstack/kolla-kubernetes: Replace basestring with six.string_types https://review.openstack.org/401144 | 09:49 |
duonghq | sdake_, I'll join the compute-kit force | 09:49 |
sdake_ | duonghq cool | 09:49 |
jascott1 | with helm and operators? | 09:50 |
sdake_ | yup and init containers with entrypoint | 09:50 |
jascott1 | got it | 09:50 |
sdake_ | thats the architecture | 09:50 |
sdake_ | first order of business is to get launchpad organized | 09:51 |
sdake_ | because atm, it isn't | 09:51 |
sdake_ | duonghq can you untarget swift in launchpad | 09:52 |
sdake_ | from ocata | 09:52 |
sdake_ | we are in essence kicking it out of ocata :) | 09:52 |
duonghq | sdake_, sure | 09:53 |
sdake_ | lights out recovery | 09:53 |
sdake_ | soudns nice | 09:53 |
sdake_ | not part of our objective | 09:53 |
*** jtriley has joined #openstack-kolla | 09:53 | |
duonghq | sdake_, done :) | 09:53 |
sdake_ | i'll untarget that one | 09:53 |
sdake_ | mongodb jascott1 you mentioned you took a look at the contents, can you untarget | 09:54 |
jascott1 | sure | 09:55 |
sdake_ | ssl support sounds nice | 09:56 |
sdake_ | we didn't have it for kolla 1.0.0 | 09:56 |
jascott1 | we are setting 'Definition' to 'Discussion' correct? | 09:56 |
sdake_ | i think we can untarget that for now and revisit | 09:56 |
sdake_ | yup set definition to discussion | 09:56 |
jascott1 | not actual Re-target | 09:56 |
jascott1 | ok thanks | 09:56 |
sdake_ | ya and remove from ocata release | 09:57 |
jascott1 | "Run ceilometer on Kubernetes"? | 09:57 |
sdake_ | so thats the retarget part | 09:57 |
sdake_ | ya nix that :) | 09:57 |
jascott1 | Re-target blueprint and leave blank? | 09:57 |
sdake_ | we can always add em back in | 09:57 |
sdake_ | do your thing on that blueprint and i'll have a look | 09:57 |
sdake_ | and see if its correct :) | 09:57 |
duonghq | cya in today meeting | 09:58 |
*** duonghq has quit IRC | 09:58 | |
*** jtriley has quit IRC | 09:58 | |
jascott1 | oh its doesnt accept blank | 09:58 |
*** jtriley has joined #openstack-kolla | 09:59 | |
jascott1 | is it Seried Goal? | 09:59 |
sdake_ | jascott1 not a series goal | 09:59 |
jascott1 | sorry, how to un-target ? | 09:59 |
sdake_ | jascott1 can you link - i dont' see the blueprint in launchpad | 09:59 |
jascott1 | https://blueprints.launchpad.net/kolla-kubernetes/+spec/ceilometer-kubernetes | 09:59 |
sdake_ | i'll teach you | 10:00 |
sdake_ | see accepted for ocata | 10:00 |
jascott1 | yes | 10:00 |
sdake_ | see yellow button next to it | 10:00 |
jascott1 | under 'Series Goal' | 10:00 |
sdake_ | click that one | 10:00 |
sdake_ | the yellow button | 10:00 |
jascott1 | ok check it now | 10:01 |
sdake_ | looks good - one note - all launchpad blueprints shoudl be set to not started | 10:01 |
*** zhurong has quit IRC | 10:01 | |
jascott1 | ok | 10:01 |
jascott1 | lemme fix mongo one up | 10:01 |
sdake_ | that are in the state that particular one is in | 10:01 |
sdake_ | dunn owhy launchpad has that set the way it does | 10:02 |
sdake_ | oh set priority to undefined as well :) | 10:02 |
jascott1 | oh what about milestone target? | 10:02 |
sdake_ | since its not accepted for ocata, it can't hav ea priority | 10:02 |
sdake_ | milestone target should be untargeted too | 10:02 |
jascott1 | ok | 10:02 |
*** Pavo has quit IRC | 10:03 | |
*** jtriley has quit IRC | 10:03 | |
sdake_ | https://blueprints.launchpad.net/kolla-kubernetes/+spec/ceilometer-kubernetes looks pretty good but currently set to implementation unknown | 10:03 |
sdake_ | we need to get rid of the noise | 10:04 |
jascott1 | what should implementation be set to? | 10:05 |
*** tonanhngo has joined #openstack-kolla | 10:05 | |
*** YuYangWang has joined #openstack-kolla | 10:05 | |
*** tovin07_ has quit IRC | 10:06 | |
*** tonanhngo has quit IRC | 10:07 | |
jascott1 | "Lights off recovery" is out correct? | 10:07 |
sdake_ | ya | 10:07 |
sdake_ | in the meeting today people can pull this stuff back in if they want | 10:07 |
*** Pavo has joined #openstack-kolla | 10:08 | |
sdake_ | need to get things rolling - lots of blockages | 10:08 |
sdake_ | general rule of thumb stuff in good progress state stays in until deadlines are hit :) | 10:09 |
sdake_ | even if its extra | 10:09 |
jascott1 | sounds like nice to have: https://blueprints.launchpad.net/kolla-kubernetes/+spec/cli-resource-types | 10:10 |
sdake_ | people always add extra stuff - which is cool - atm i see a sea of blueprints and as a developer its har to know where to start | 10:10 |
jascott1 | but not sure | 10:10 |
jascott1 | i will take it out and they can pull it back liek you say | 10:10 |
sdake_ | no implementation | 10:10 |
sdake_ | medium priority | 10:10 |
sdake_ | not part of objective or spec | 10:11 |
sdake_ | again ppl can always pull stuff back in if they implmeent it | 10:11 |
sdake_ | jascott1 re https://blueprints.launchpad.net/kolla-kubernetes/+spec/init-container-fail | 10:14 |
sdake_ | does that make any sense to you :) | 10:14 |
jascott1 | yes | 10:15 |
jascott1 | started, i would say keep | 10:15 |
jascott1 | init container runs a script but only fails out if last command in script fails | 10:16 |
jascott1 | needs to fail out if any command fails | 10:16 |
jascott1 | but | 10:16 |
jascott1 | are we using init containers for that anymore | 10:17 |
*** zhangshuai has quit IRC | 10:17 | |
jascott1 | since it goes to consistency/reliability of compute kit I would keep it | 10:17 |
jascott1 | sdake should we be setting ocata-2 if its something we are keeping? | 10:18 |
jascott1 | this needs more detail imo, as its written this is covered by entrypoint https://blueprints.launchpad.net/kolla-kubernetes/+spec/health-checks | 10:22 |
sdake_ | jascott1 yup ocata-2 if keeping | 10:22 |
*** zhangshuai has joined #openstack-kolla | 10:23 | |
sdake_ | jascott1 do containers need healthchecks built in? | 10:23 |
sdake_ | our architcture has entrypoint in an init container | 10:23 |
jascott1 | right | 10:23 |
sdake_ | so init-container-fail stays? | 10:24 |
jascott1 | entrypoint handles those deps | 10:24 |
*** tonanhngo has joined #openstack-kolla | 10:24 | |
sdake_ | ok sounds like we have some container work to do as well | 10:24 |
sdake_ | we can tackle that later ;) | 10:24 |
jascott1 | yes | 10:24 |
jascott1 | yes init-container-fail stays | 10:24 |
sdake_ | cool | 10:24 |
jascott1 | its about correctnes | 10:25 |
sdake_ | wfm | 10:25 |
*** tonanhngo has quit IRC | 10:25 | |
sdake_ | i just dont understnd what the content is in the blueprint :) | 10:25 |
*** zhubingbing_ has quit IRC | 10:26 | |
jascott1 | i hear ya. thats how I usually feel around here :) | 10:26 |
jascott1 | but have sense of k8s/helm and have looked at several of the new implementations going on | 10:26 |
jascott1 | but yeah this guys is covered by entrypoint unless they are intending more advanced logic https://blueprints.launchpad.net/kolla-kubernetes/+spec/health-checks | 10:27 |
jascott1 | and if its advanced logic then they should update it to say 'that entrypoint doesnt provide' | 10:27 |
jascott1 | or similar | 10:27 |
jascott1 | one prob for me is sometimes ppl say init container and they mean the stackanets entrypoint that simply waits for deps and sometimes they mean a Job container to 'init' the service | 10:30 |
jascott1 | like this one https://blueprints.launchpad.net/kolla-kubernetes/+spec/dependency-init-container | 10:30 |
jascott1 | that one is entrypoint afaik | 10:31 |
sdake_ | jascott1 sorry was working in launchpad give mea moment to read | 10:33 |
sdake_ | ok so health checks sounds like something for kolla repo | 10:34 |
*** jascott1_ has joined #openstack-kolla | 10:34 | |
sdake_ | where as dependency-init-contianer sounds like something more for kolla-kubernetes | 10:34 |
sdake_ | jascott1_ [03:34:34] <sdake_>ok so health checks sounds like something for kolla repo | 10:35 |
sdake_ | [03:34:50] jascott1_ (jascott1@nat/intel/x-hqtginzhsxxzhsii) joined the channel | 10:35 |
sdake_ | [03:34:52] <sdake_>where as dependency-init-contianer sounds like something more for kolla-kubernetes | 10:35 |
*** jascott1 has quit IRC | 10:35 | |
jascott1_ | well health check is part of k8s container def | 10:35 |
sdake_ | if we really aren't certain lets punt and keep it in and we can hash it out in todays meeting | 10:35 |
jascott1_ | cool | 10:35 |
*** ppalacios has joined #openstack-kolla | 10:36 | |
sdake_ | probably those blueprints need to be merged | 10:36 |
jascott1_ | ya i was going to leave it also | 10:36 |
jascott1_ | yea | 10:36 |
sdake_ | or one needs to go to kolla to implement the healthchecking | 10:36 |
*** ppalacios has left #openstack-kolla | 10:36 | |
jascott1_ | going to change my avatar https://lh3.googleusercontent.com/_5CNFBJhHzCw/S0VHrZYKESI/AAAAAAAApNE/0bkOKq1nb5Q/Futurama-Hermes-Conrad.jpg | 10:37 |
*** newmember has quit IRC | 10:37 | |
jascott1_ | :) | 10:37 |
sdake_ | zomg | 10:37 |
sdake_ | that guy rocks :) | 10:37 |
sdake_ | fwiw we dont rubber stamp reviews unless they need rubber stamping | 10:37 |
jascott1_ | sounds legit | 10:38 |
sdake_ | and there is still an expectation of core reviewers that htey should check out the code | 10:38 |
sdake_ | and make sure it is right | 10:38 |
sdake_ | rubber stamping is rare | 10:38 |
jascott1_ | oh this one https://blueprints.launchpad.net/kolla-kubernetes/+spec/kolla-kubernetes-tools | 10:39 |
jascott1_ | just says 'tools' | 10:39 |
jascott1_ | paraphrasing | 10:39 |
jascott1_ | im sure it was a placeholder | 10:39 |
*** jascott1 has joined #openstack-kolla | 10:41 | |
jascott1 | vpn choked or something idk | 10:41 |
jascott1 | thoughts on tools? (sorry if ye replied already, bouncer is on the fritz today) | 10:42 |
jascott1 | ha we just edited same | 10:44 |
jascott1 | password-secrets | 10:44 |
*** mmoster_ has quit IRC | 10:44 | |
*** jascott1_ has quit IRC | 10:45 | |
*** liyifeng has joined #openstack-kolla | 10:45 | |
jascott1 | sdake ? | 10:45 |
jascott1 | imean sdake_ | 10:46 |
sdake_ | sorry | 10:47 |
sdake_ | was working :) | 10:47 |
sdake_ | its a kolla-in-a-box | 10:47 |
sdake_ | thsi hsa been discussed multiple times | 10:47 |
sdake_ | people want to work on it | 10:47 |
sdake_ | is it a priority for spec + compute kit? | 10:47 |
sdake_ | think not | 10:47 |
sdake_ | if someone comes along and wants to work on it, more power to em | 10:48 |
jascott1 | nuking in 3 . 2 | 10:48 |
sdake_ | we dont force people to work on things or kick out work if its mergable | 10:48 |
berendt | anybody knows why it could happen that rabbitmq does not startup ? | 10:48 |
sdake_ | berendt no idea | 10:48 |
berendt | process in the container is running (/bin/sh -e /usr/lib/rabbitmq/bin/rabbitmq-server), but nothing happens | 10:49 |
sdake_ | berendt no time to debug atm | 10:49 |
sdake_ | rabbitmq is a pile of shit | 10:49 |
sdake_ | what more can be said | 10:49 |
jascott1 | similar https://blueprints.launchpad.net/kolla-kubernetes/+spec/pre-checks | 10:49 |
sdake_ | ya that can go | 10:49 |
sdake_ | its a feature we have in kolla | 10:50 |
sdake_ | what ryan did wit hteh blueprints is wrote a whole slew of em to give the community work to focus on | 10:50 |
sdake_ | the problem was they are all approved for ocata | 10:50 |
sdake_ | and there is not enough capacity to execute for ocata | 10:50 |
sdake_ | with all the stuff that is in there | 10:51 |
sdake_ | kolla-kubernetes 1.0.0 will not be equivalent feature wise to kolla-ansible | 10:51 |
sdake_ | it will probably take some time to catch up | 10:51 |
sdake_ | we are not doing a poc, we are doing a 1.0.0 project | 10:51 |
sdake_ | ryan was doing a 4.0.0 project ;) | 10:51 |
*** athomas has quit IRC | 10:52 | |
*** jtriley has joined #openstack-kolla | 10:53 | |
*** tonanhngo has joined #openstack-kolla | 10:54 | |
jascott1 | yeah its such a moving target, helms 'about to' have plugins which would open up flexibility for us wrt to sub chart releases and what not | 10:54 |
jascott1 | sorry blabbering a bit there | 10:55 |
berendt | looks like it is an issue with the rabbitmq-clusterer plugin, single node deployment works | 10:56 |
jascott1 | sdake know any details on this one? https://blueprints.launchpad.net/kolla-kubernetes/+spec/ssl-support | 10:57 |
*** jtriley has quit IRC | 10:57 | |
*** jtriley has joined #openstack-kolla | 10:58 | |
jascott1 | sounds like a bug? | 10:58 |
*** tonanhngo has quit IRC | 10:59 | |
sdake_ | no details known | 10:59 |
sdake_ | work ittems not filled out | 10:59 |
sdake_ | i am kind of a security nut | 10:59 |
sdake_ | however kolla-ansible lacked a tls implementation | 11:00 |
sdake_ | adding tls is fairly easy | 11:00 |
sdake_ | if we have time we can do it | 11:00 |
sdake_ | or if someone wants to tackle it cool :) | 11:00 |
*** athomas has joined #openstack-kolla | 11:00 | |
jascott1 | leave? | 11:01 |
jascott1 | nuking kuryr | 11:01 |
*** Bico_Fino has joined #openstack-kolla | 11:02 | |
jascott1 | this looks like re-target project to kolla-ansible https://blueprints.launchpad.net/kolla-kubernetes/+spec/bool-filter | 11:02 |
*** jtriley has quit IRC | 11:03 | |
sdake_ | check it out - something achievable :) https://launchpad.net/kolla-kubernetes/+milestone/ocata-2 | 11:03 |
sdake_ | nice work jascott :) | 11:03 |
sdake_ | not sure if we move that or not - lets keep where it is for the moment | 11:04 |
*** tonanhngo has joined #openstack-kolla | 11:04 | |
sdake_ | we have this funky genconfig thing at present | 11:04 |
sdake_ | which basically generates configuration from the defaults | 11:04 |
sdake_ | the defaults are jinja2 | 11:05 |
*** tonanhngo has quit IRC | 11:05 | |
sdake_ | ansible runs the genconfig | 11:05 |
sdake_ | it was a short term hack to get dev moving | 11:05 |
sdake_ | intent there is to move the config into docker repo | 11:06 |
sdake_ | rather then kolla-kubernetes repo | 11:06 |
sdake_ | inc0 said he would write a spec on it | 11:06 |
*** msimonin has quit IRC | 11:06 | |
sdake_ | i suggested blueprint but he wants to write a spec, god help him :) | 11:06 |
sdake_ | ok - since i perceive alot of people working on helm | 11:08 |
sdake_ | the way we handle that is with one blueprint | 11:08 |
sdake_ | with work items | 11:08 |
sdake_ | jascott1 mind learning how to do that | 11:09 |
sdake_ | jascott1 i already know how :0 | 11:09 |
*** prameswar has quit IRC | 11:09 | |
sdake_ | will walk yo uthrough it | 11:09 |
sdake_ | important thing to know | 11:09 |
jascott1 | creating blueprint with workitems? | 11:10 |
sdake_ | we already have the bleuprint | 11:11 |
sdake_ | not sure how familiar you are with launchpad | 11:11 |
jascott1 | oh spec right | 11:11 |
jascott1 | increasingly | 11:11 |
sdake_ | ;) | 11:11 |
jascott1 | i havent written a spec or authored much in general in our systems | 11:12 |
jascott1 | but am familiar | 11:12 |
sdake_ | nah thats not what i mean | 11:13 |
sdake_ | work items in launchpad - let me show you | 11:13 |
*** liyifeng has quit IRC | 11:13 | |
*** Jeffrey4l has joined #openstack-kolla | 11:14 | |
*** igordcard has quit IRC | 11:14 | |
jascott1 | ok | 11:14 |
*** msimonin has joined #openstack-kolla | 11:15 | |
*** igordcard has joined #openstack-kolla | 11:15 | |
sdake_ | https://blueprints.launchpad.net/kolla-kubernetes/+spec/helm-packaging | 11:15 |
jascott1 | o | 11:15 |
jascott1 | ok | 11:15 |
sdake_ | you see the field that says whiteboard? | 11:15 |
jascott1 | yes | 11:16 |
sdake_ | see how it has all that stuff in there | 11:16 |
jascott1 | i do | 11:16 |
sdake_ | that needs to go into the workitems section | 11:16 |
sdake_ | this is how we divide up work so people dont do the same thing | 11:16 |
jascott1 | ok | 11:17 |
jascott1 | so just copy pasta | 11:17 |
jascott1 | ? | 11:17 |
sdake_ | sometimes that happens, but typically not becaue the blueprints are wrong :) | 11:17 |
sdake_ | nah the work items have a specific format | 11:17 |
jascott1 | so go ahead and create them with these titles | 11:17 |
jascott1 | ? | 11:17 |
sdake_ | dont create blueprintgs | 11:18 |
sdake_ | create work items with the stuff in the whiteboard | 11:18 |
jascott1 | i get that but I hit the pencil and its just a text field | 11:18 |
jascott1 | i can look at another | 11:19 |
jascott1 | oh none of them have them it seems | 11:19 |
sdake_ | there is one | 11:20 |
sdake_ | the syntax is (unassigned)ceph: TODO | 11:20 |
sdake_ | i believe | 11:20 |
jascott1 | found some here but idk correctness https://blueprints.launchpad.net/kolla-kubernetes/+spec/swift-kubernetes | 11:20 |
sdake_ | yup that is right | 11:21 |
sdake_ | INPROGRESS = in progress | 11:21 |
sdake_ | unassigned = unassigned | 11:21 |
sdake_ | rather TODO = undone | 11:22 |
sdake_ | (unassigned)ceph: TODO | 11:22 |
sdake_ | give that a go | 11:22 |
jascott1 | gotcha. so do them all that format with TODO | 11:23 |
sdake_ | ya | 11:23 |
sdake_ | although one isn't todo | 11:23 |
*** tonanhngo has joined #openstack-kolla | 11:24 | |
*** prameswar has joined #openstack-kolla | 11:25 | |
jascott1 | which? | 11:25 |
sdake_ | not sure | 11:25 |
sdake_ | mark em all todo atm | 11:25 |
sdake_ | let me hunt down that review | 11:25 |
jascott1 | ok | 11:25 |
*** tonanhngo has quit IRC | 11:25 | |
jascott1 | https://blueprints.launchpad.net/kolla-kubernetes/+spec/helm-packaging | 11:26 |
sdake_ | neutron is in progress | 11:28 |
*** liyifeng has joined #openstack-kolla | 11:29 | |
Jeffrey4l | sdake_, what that means in https://review.openstack.org/401032 ? conversion to proper j2 macro? we do not have j2 macro in mitaka branch. | 11:31 |
mliima_ | morning guys | 11:33 |
jascott1 | good morning! | 11:35 |
Jeffrey4l | hi mliima_ | 11:35 |
jascott1 | sdake_ done | 11:35 |
mliima_ | hey Jeffrey4l \o | 11:35 |
sdake_ | Jeffrey4l sorry dude | 11:35 |
sdake_ | didn't catch that was mitaka | 11:35 |
Jeffrey4l | could u review this https://review.openstack.org/#/c/400922/ and all the following patches. it unblock the mitaka branch | 11:36 |
Jeffrey4l | sdake_, ;) ^^ | 11:36 |
sdake_ | Jeffrey4l can you get another core to review pls | 11:36 |
openstackgerrit | Merged openstack/kolla-ansible: Remove docker folder from kolla-ansible https://review.openstack.org/398320 | 11:36 |
mliima_ | \o/ | 11:36 |
mliima_ | i'm here Jeffrey4l sdake_ | 11:37 |
Jeffrey4l | all patches are: https://review.openstack.org/400922 https://review.openstack.org/400892 https://review.openstack.org/401032 | 11:37 |
Jeffrey4l | sdake_, ok. np. | 11:37 |
sdake_ | jascott1 isn't glance in progress too? | 11:44 |
*** tonanhngo has joined #openstack-kolla | 11:44 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Added support for a stop playbook in kolla https://review.openstack.org/399289 | 11:45 |
*** Bico_Fino has quit IRC | 11:46 | |
jascott1 | yes but i started it in v1k0d3n's format | 11:46 |
*** tonanhngo has quit IRC | 11:46 | |
jascott1 | in the same manner in which their poc is implemented | 11:47 |
sdake_ | jascott1 cool thanks for your help - that got things moving in the right direction | 11:47 |
jascott1 | I appreciate the lesson! | 11:47 |
sdake_ | jascott1 ok - so thats still TODO then | 11:47 |
jascott1 | ya but I will convert it and update the work item :) | 11:48 |
sdake_ | sounds good :) | 11:48 |
jascott1 | gonna catch some z's. tty in the meeting | 11:48 |
sdake_ | enjoy | 11:48 |
jascott1 | |) | 11:48 |
Jeffrey4l | klindgren, around? | 11:51 |
*** Bico_Fino has joined #openstack-kolla | 11:55 | |
*** msimonin has quit IRC | 11:55 | |
openstackgerrit | Mauricio Lima proposed openstack/kolla-ansible: Remove fragments of docker files https://review.openstack.org/401222 | 11:55 |
sdake_ | ok Jeffrey4l still need reviews or all set? | 11:56 |
*** jtriley has joined #openstack-kolla | 11:59 | |
*** mkoderer has joined #openstack-kolla | 11:59 | |
openstackgerrit | Mauricio Lima proposed openstack/kolla-ansible: Remove fragments of docker files https://review.openstack.org/401222 | 11:59 |
*** Pavo has quit IRC | 12:03 | |
Jeffrey4l | sdake_, yep. https://review.openstack.org/400922 and its following https://review.openstack.org/400892 and last one you reviewed https://review.openstack.org/401032 . | 12:03 |
Jeffrey4l | the last one get green gate. | 12:03 |
*** openstackgerrit has quit IRC | 12:03 | |
sdake_ | Jeffrey4l you got it looking | 12:03 |
Jeffrey4l | thanks. | 12:03 |
*** openstackgerrit has joined #openstack-kolla | 12:03 | |
*** jtriley has quit IRC | 12:04 | |
*** jascott1 has quit IRC | 12:04 | |
sdake_ | Jeffrey4l this patch is -1 | 12:04 |
sdake_ | https://review.openstack.org/#/c/400922/ | 12:04 |
sdake_ | from jenkins | 12:04 |
sdake_ | not sure why | 12:04 |
sdake_ | your patch stream wont merge unless you fix that problem | 12:04 |
Jeffrey4l | because percona repo issue see the logs http://logs.openstack.org/22/400922/1/check/gate-kolla-dsvm-build-oraclelinux-source-centos-7/3117f6f/console.html#_2016-11-22_18_33_11_144565 | 12:05 |
Jeffrey4l | the second patch https://review.openstack.org/400892 fix this | 12:06 |
Jeffrey4l | sdake_, ^^ | 12:06 |
sdake_ | you will have to collapse the patches tehn | 12:06 |
sdake_ | rather squash | 12:06 |
sdake_ | to get past the gate | 12:06 |
sdake_ | are they dependent? | 12:07 |
Jeffrey4l | collapse means squash? | 12:07 |
Jeffrey4l | is there any other workaround for such case? | 12:07 |
sdake_ | you could reorder the patches | 12:07 |
sdake_ | welcome to voting gates ;) | 12:08 |
*** Pavo has joined #openstack-kolla | 12:08 | |
mliima_ | sdake_, Jeffrey4l should i remove https://github.com/openstack/kolla-ansible/blob/master/tests/test_kolla_docker.py and https://github.com/openstack/kolla-ansible/blob/master/tests/test_set_config.py ? | 12:08 |
openstackgerrit | Merged openstack/kolla: Pull swift-object-expirer image https://review.openstack.org/401097 | 12:08 |
Jeffrey4l | sdake_, no. if put the percona first won't work, because disable ipv6 bust the gate. | 12:08 |
sdake_ | i see | 12:08 |
sdake_ | yup squash is only answer | 12:08 |
sdake_ | actually there is another answer | 12:09 |
sdake_ | make the gates nonvoting that are cusing jenkisn to fail | 12:09 |
Jeffrey4l | mliima_, test_kolla_docker no. test_set_config yes. the former is used by ansible, the latter is used by docker images. | 12:09 |
sdake_ | but that answer isn't what inc0 wants | 12:09 |
mliima_ | ok | 12:09 |
sdake_ | so squash is what needs to happen when gate bustages happen | 12:10 |
*** schwicht has joined #openstack-kolla | 12:10 | |
Jeffrey4l | sdake_, OK. i will squash there two patches into one. | 12:10 |
openstackgerrit | Mauricio Lima proposed openstack/kolla-ansible: Remove fragments of docker files https://review.openstack.org/401222 | 12:11 |
openstackgerrit | Merged openstack/kolla: typo error of keyring spelling https://review.openstack.org/401103 | 12:12 |
mliima_ | sdake_, Jeffrey4l i need some review here https://review.openstack.org/#/c/400734/ and https://review.openstack.org/#/c/399715/ | 12:14 |
Jeffrey4l | roger, reviewing | 12:15 |
*** tonanhngo has joined #openstack-kolla | 12:15 | |
Jeffrey4l | mliima_, should we backport the sahara nfs issue to newton? | 12:15 |
*** tonanhngo has quit IRC | 12:16 | |
mliima_ | no | 12:16 |
Jeffrey4l | because manila-data is added in O, right? | 12:16 |
Jeffrey4l | mliima_, ^ | 12:16 |
mliima_ | yes Jeffrey4l | 12:16 |
Jeffrey4l | mliima_, does manila support ceph-fs right now? is it recommended to use in production? | 12:18 |
mliima_ | ceph-fs? | 12:20 |
Jeffrey4l | yup. | 12:20 |
mliima_ | no | 12:20 |
mliima_ | well | 12:20 |
mliima_ | i tested manila generic backend using ceph | 12:20 |
mliima_ | because cinder + lvm doesn't works in ubuntu 16.04 | 12:21 |
mliima_ | so i tested using cinder + ceph, cinder is required to use manila + generic backend | 12:21 |
mliima_ | but ceph-fs, btw ceph-fs = ceph-filesystem? | 12:22 |
openstackgerrit | Merged openstack/kolla: Fix neutron.conf.j2 metadata_workers spelling error https://review.openstack.org/401107 | 12:22 |
Jeffrey4l | mliima_, yep http://docs.ceph.com/docs/jewel/cephfs/ | 12:22 |
openstackgerrit | Merged openstack/kolla: Fix Designate source image container assets https://review.openstack.org/400730 | 12:22 |
openstackgerrit | Merged openstack/kolla: Add specs are advisory readme.rst to kolla repo https://review.openstack.org/400570 | 12:22 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Fix the gate https://review.openstack.org/400922 | 12:23 |
mliima_ | manila doesn't support ceph-fs Jeffrey4l , but I can do. | 12:23 |
mliima_ | http://docs.openstack.org/developer/manila/devref/cephfs_native_driver.html | 12:23 |
mliima_ | basically Manila is ready to support any backend, you just need set the configs inside manila.conf | 12:25 |
Jeffrey4l | mliima_, never tried manila before ;). but cephfs native driver is stable and ready for production use, right? | 12:25 |
mliima_ | yes | 12:26 |
Jeffrey4l | sound cool. | 12:26 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Move glance precheck into its own role https://review.openstack.org/399354 | 12:28 |
*** mdnadeem has quit IRC | 12:33 | |
*** schwicht has quit IRC | 12:33 | |
*** alteriks has joined #openstack-kolla | 12:34 | |
*** tonanhngo has joined #openstack-kolla | 12:34 | |
*** tonanhngo has quit IRC | 12:35 | |
*** msimonin has joined #openstack-kolla | 12:36 | |
*** shardy is now known as shardy_lunch | 12:39 | |
*** mmoster_ has joined #openstack-kolla | 12:40 | |
openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla-ansible: Add missing pull task for neutron-vpnaas-agent https://review.openstack.org/401240 | 12:41 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add socat into haproxy image https://review.openstack.org/401242 | 12:45 |
*** liyifeng has quit IRC | 12:47 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Remove adding repo in magnum-base Dockerfile https://review.openstack.org/401032 | 12:52 |
sean-k-mooney | speaking of ceph has anyone used the rbd volumn driver for docker? | 12:57 |
sean-k-mooney | i need to redeploy my ceph cluster this week. | 12:58 |
sean-k-mooney | sorry kolla cluster | 12:58 |
sean-k-mooney | has anyone used the rbd storage driver for docker before? | 12:58 |
sean-k-mooney | http://ceph.com/planet/getting-started-with-the-docker-rbd-volume-plugin/ | 12:59 |
*** Bico_Fino has quit IRC | 12:59 | |
*** liyifeng has joined #openstack-kolla | 12:59 | |
pbourke | Jeffrey4l: Im still getting denied by rootwrap in the vpnaas container :( | 13:03 |
Jeffrey4l | pbourke, enable debug log and post the log file out in neutron-server. | 13:04 |
*** msimonin has quit IRC | 13:05 | |
*** tonanhngo has joined #openstack-kolla | 13:05 | |
*** prameswar has quit IRC | 13:06 | |
*** tonanhngo has quit IRC | 13:06 | |
*** Bico_Fino has joined #openstack-kolla | 13:06 | |
*** schwicht has joined #openstack-kolla | 13:17 | |
openstackgerrit | satya proposed openstack/kolla-ansible: added separate interface for rabbitmq https://review.openstack.org/401256 | 13:21 |
*** Satya_ has joined #openstack-kolla | 13:22 | |
Satya_ | hi sdake | 13:22 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Bump OpenStack service version https://review.openstack.org/394271 | 13:23 |
Satya_ | hi jeffrey | 13:23 |
*** Bico_Fino has quit IRC | 13:23 | |
*** jtriley has joined #openstack-kolla | 13:23 | |
Satya_ | Please review https://review.openstack.org/#/c/401256/ | 13:24 |
*** tonanhngo has joined #openstack-kolla | 13:25 | |
Jeffrey4l | roger. | 13:25 |
*** tonanhngo has quit IRC | 13:26 | |
*** newmember has joined #openstack-kolla | 13:26 | |
*** shardy_lunch is now known as shardy | 13:28 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Bump OpenStack service version https://review.openstack.org/394271 | 13:29 |
*** sp__ has joined #openstack-kolla | 13:30 | |
*** Bico_Fino has joined #openstack-kolla | 13:32 | |
*** kristian__ has joined #openstack-kolla | 13:34 | |
*** srwilkers has joined #openstack-kolla | 13:35 | |
kristian__ | Hi I have a question. Can someone make a script or add it to the configs an automatic horizon plugin installer? | 13:35 |
*** zhubingbing_ has joined #openstack-kolla | 13:35 | |
zhubingbing_ | hello guys | 13:35 |
sdake_ | sup zhubingbing_ | 13:35 |
*** saneax-_-|AFK is now known as saneax | 13:36 | |
sdake_ | sup srwilkers | 13:36 |
zhubingbing_ | hi | 13:36 |
srwilkers | hey sdake_ | 13:36 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add socat into haproxy image https://review.openstack.org/401242 | 13:36 |
zhubingbing_ | sdake, can u help me review trove role? | 13:36 |
sdake_ | srwilkers quick q | 13:36 |
sdake_ | zhubingbing_ moment | 13:36 |
srwilkers | sdake_, shoot | 13:36 |
zhubingbing_ | :) | 13:36 |
zhubingbing_ | ok thanks | 13:36 |
zhubingbing_ | https://review.openstack.org/#/c/400580/ | 13:36 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Add socat into haproxy image https://review.openstack.org/401242 | 13:36 |
*** ppalacios has joined #openstack-kolla | 13:37 | |
sdake_ | srwilkers https://launchpad.net/kolla-kubernetes/+milestone/ocata-2 | 13:38 |
sdake_ | your blueprint is marked good progress | 13:38 |
sdake_ | i think we are in general agreement kolla-kubernetes needs a dev env | 13:38 |
*** lamt has joined #openstack-kolla | 13:39 | |
*** schwicht has quit IRC | 13:39 | |
srwilkers | agreed -- when we started looking at getting involved months back, it was definitely something we saw as an opportunity | 13:39 |
sdake_ | i was speaking with v1k0d3n last night but lost laptop power | 13:40 |
sdake_ | he mentioned a problem with PVC | 13:40 |
sdake_ | 2 qs. | 13:40 |
sdake_ | actually :) | 13:40 |
sdake_ | what is the PVC problem and how can we help solve it | 13:40 |
sdake_ | and when are the docs going to be in usable shape | 13:41 |
*** jtriley has quit IRC | 13:43 | |
sdake_ | zhubingbing_ gate is green, thats a positive sign :) | 13:44 |
mliima_ | hi zhubingbing_ | 13:45 |
*** dmsimard|away is now known as dmsimard | 13:45 | |
*** zhubingbing_ has quit IRC | 13:45 | |
openstackgerrit | Merged openstack/kolla: Fix the gate https://review.openstack.org/400922 | 13:46 |
*** kristian__ has quit IRC | 13:46 | |
*** kristian__ has joined #openstack-kolla | 13:47 | |
Jeffrey4l | anyone is familiar with cross project testing? http://docs.openstack.org/infra/zuul/gating.html#cross-project-testing | 13:49 |
*** kristia__ has joined #openstack-kolla | 13:50 | |
*** zhubingbing has joined #openstack-kolla | 13:51 | |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: Move glance precheck into its own role https://review.openstack.org/399354 | 13:52 |
*** kristian__ has quit IRC | 13:52 | |
*** tonanhngo has joined #openstack-kolla | 13:55 | |
*** tonanhngo has quit IRC | 13:56 | |
*** BIOS_Hazard has joined #openstack-kolla | 13:57 | |
BIOS_Hazard | What up Kolla people! | 13:57 |
BIOS_Hazard | I have had an eye on this project for a while and finally got a chance to try it out yesterday. I got all the way to the prechecks and am unsure of how to overcome an error (though I suspect I am simply misunderstanding the proper way to configure the globals). | 13:58 |
BIOS_Hazard | Here is my issue: | 13:59 |
BIOS_Hazard | ~~~ | 13:59 |
BIOS_Hazard | TASK [prechecks : Checking if kolla_internal_vip_address is in the same network as network_interface on all nodes] *** | 13:59 |
BIOS_Hazard | fatal: [localhost]: FAILED! => {"changed": false, "cmd": ["ip", "-4", "-o", "addr", "show", "dev", "enp4s0"], "delta": "0:00:00.003768", "end": "2016-11-22 22:53:55.705335", "failed": true, "failed_when_result": true, "rc": 0, "start": "2016-11-22 22:53:55.701567", "stderr": "", "stdout": "2: enp4s0 inet 198.136.52.141/29 brd 198.136.52.143 scope global enp4s0\\ valid_lft forever preferred_lft forever", "stdout_lines" | 13:59 |
BIOS_Hazard | to retry, use: --limit @/usr/share/kolla/ansible/prechecks.retry | 13:59 |
BIOS_Hazard | ~~~ | 13:59 |
sdake_ | BIOS_Hazard please use a paste service | 13:59 |
BIOS_Hazard | that makes sense | 13:59 |
sdake_ | srwilkers ^^ not sure if you saw my msg above | 13:59 |
mliima_ | paste.openstack.org BIOS_Hazard | 13:59 |
mliima_ | :) | 13:59 |
BIOS_Hazard | http://pastebin.com/dYwaT06i | 13:59 |
BIOS_Hazard | Thanks mliima_ | 14:00 |
BIOS_Hazard | I will use that next time | 14:00 |
sdake_ | zhubingbing looking good dude ! | 14:00 |
sdake_ | zhubingbing few minor nitpicks i see in there | 14:00 |
zhubingbing | ok | 14:00 |
zhubingbing | thanks | 14:00 |
BIOS_Hazard | and here are my globals | 14:01 |
BIOS_Hazard | http://paste.openstack.org/show/590208/ | 14:01 |
sdake_ | zhubingbing unfortunately i dont know how to recommend how to fix - Jeffrey4l should provide a review i think since he touched that code last ;) | 14:01 |
BIOS_Hazard | On enp4s0 I have 198.136.52.141/29 configured | 14:01 |
Jeffrey4l | zhubingbing, sup | 14:01 |
zhubingbing | here | 14:02 |
BIOS_Hazard | I found the playbook section for this precheck, but I am unsure of what the "failed_when" logic entails | 14:02 |
*** Pavo has quit IRC | 14:02 | |
*** duonghq has joined #openstack-kolla | 14:03 | |
sdake_ | sup pavo | 14:04 |
duonghq | evening guys | 14:04 |
sdake_ | pavo any chance you can help BIOS_Hazard with his network setup :) | 14:04 |
sdake_ | oops pavo is gone | 14:04 |
BIOS_Hazard | Thanks sdake_ | 14:04 |
*** jascott1 has joined #openstack-kolla | 14:04 | |
sdake_ | duonghq sup dude | 14:04 |
BIOS_Hazard | I am very excited to pass the prechecks and get this all-in-one deployed. | 14:04 |
sdake_ | how was the ower nap jascott1 | 14:04 |
*** fguillot has joined #openstack-kolla | 14:05 | |
sdake_ | power nap | 14:05 |
BIOS_Hazard | kolla is clearly the future of openstack deployment and upkeep | 14:05 |
duonghq | thank BIOS_Hazard | 14:05 |
BIOS_Hazard | thank you guys, for putting so much work into this | 14:05 |
duonghq | cannot figure out why the h button on my keyboard is unstable hmmm | 14:06 |
*** Pavo has joined #openstack-kolla | 14:07 | |
*** jascott1 has quit IRC | 14:09 | |
BIOS_Hazard | Howdy, Pavo | 14:09 |
sdake_ | pavo can you help BIOS_Hazard out with his network problem | 14:09 |
sdake_ | Pavo i think your our foremost expert on network debug atm :) | 14:09 |
Bico_Fino | Odd behavior | 14:09 |
BIOS_Hazard | I love when the foremost expert hops on these IRC channels | 14:10 |
BIOS_Hazard | lol | 14:10 |
BIOS_Hazard | Its happened to me in almost every other channel I go to | 14:10 |
Bico_Fino | I enabled Debug on Horizon and ironic-ui started working. o_O | 14:10 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Install mkfs.vfat tool in ironic-conductor image https://review.openstack.org/400709 | 14:11 |
sean-k-mooney | Bico_Fino: could that be because fo the allowed_hosts? | 14:13 |
sean-k-mooney | Bico_Fino: didi you have an error in the apache log before? | 14:14 |
Bico_Fino | sean-k-mooney: no errors. | 14:14 |
Bico_Fino | sean-k-mooney: That’s why I enabled the Debug. | 14:14 |
sean-k-mooney | Bico_Fino: strange | 14:14 |
Bico_Fino | Also allowed hosts looks fine | 14:14 |
Bico_Fino | ALLOWED_HOSTS = ['*'] | 14:14 |
sean-k-mooney | well ALLOWED_HOSTS = ['*'] is a sledge hammer | 14:15 |
duonghq | anybody using some irc bouncer? | 14:15 |
Bico_Fino | But earlier I ran /var/lib/kolla/venv/bin/manage.py compress | 14:15 |
sean-k-mooney | it should be retricted to your loadblancer ip and i think the ip/domain name of each of the host running horizon | 14:15 |
Bico_Fino | I tried various ironic-ui versions | 14:16 |
Bico_Fino | I’m using ironic-ui==2.0.0 now | 14:16 |
egonzalez90 | BIOS_Hazard: your internal_vip_address is in a different network as your network interface | 14:16 |
egonzalez90 | vip_address 198.136.52.141 | 14:16 |
BIOS_Hazard | oh wait, should that just be my main IP then?? | 14:17 |
BIOS_Hazard | sec | 14:17 |
egonzalez90 | sorry network itnerface have that IP | 14:17 |
egonzalez90 | vip is 198.49.65.174 | 14:17 |
egonzalez90 | vip address is where services are going to communicate, and network_interface need to be in that network in order to communicate with it | 14:18 |
BIOS_Hazard | > When running an All-In-One | 14:18 |
BIOS_Hazard | # without haproxy and keepalived, this should be the first IP on your | 14:18 |
BIOS_Hazard | # 'network_interface' as set in the Networking section below. | 14:18 |
BIOS_Hazard | 2: enp4s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000 | 14:18 |
BIOS_Hazard | link/ether 00:25:90:61:cf:5e brd ff:ff:ff:ff:ff:ff | 14:18 |
BIOS_Hazard | inet 198.136.52.141/29 brd 198.136.52.143 scope global enp4s0 | 14:18 |
BIOS_Hazard | (err, sorry, should have used a paste) | 14:18 |
BIOS_Hazard | this suggests to me that I should use that one IP I have on enp4s0 | 14:19 |
sean-k-mooney | yes | 14:19 |
sean-k-mooney | it is a security thing | 14:19 |
BIOS_Hazard | doing this fails the check before: | 14:19 |
sean-k-mooney | basically that controls the cross orgin request checking as far as i know | 14:19 |
BIOS_Hazard | http://paste.openstack.org/show/590212/ | 14:19 |
BIOS_Hazard | "services are going to communicate" | 14:20 |
BIOS_Hazard | meaning with each other? | 14:20 |
BIOS_Hazard | or with me, the API end user? | 14:20 |
srwilkers | sdake_, sorry. got pulled away. pretty sure we've got it worked out. patches should be going in today | 14:20 |
sean-k-mooney | BIOS_Hazard: for allin one with out haproxy they are the same | 14:21 |
sean-k-mooney | BIOS_Hazard: you can use a different external api endpoint when you use haproxy | 14:21 |
BIOS_Hazard | That makes sense | 14:21 |
BIOS_Hazard | however, "Checking if kolla_internal_vip_address and kolla_external_vip_address are not pingable from any node" precheck fails when I use the one IP on my enp4s0 `network_interface` | 14:22 |
sean-k-mooney | for an all in one test deployment * is fine but i would tighten it up in production | 14:22 |
BIOS_Hazard | oh, so this precheck can be ignored? | 14:22 |
BIOS_Hazard | Is there a preferred means of skipping this check then? | 14:22 |
sean-k-mooney | yes the precheck dose not take acount of your haproxy setting | 14:22 |
BIOS_Hazard | ah | 14:22 |
sean-k-mooney | it a bug that should be fixed | 14:22 |
BIOS_Hazard | that makes sense then. the globals config does a good job of explaining how to do this then | 14:23 |
sean-k-mooney | but that should be skipped if not using haproxy | 14:23 |
BIOS_Hazard | and the precheck is simply trollin me | 14:23 |
sean-k-mooney | yep | 14:23 |
sdake_ | srwilkers sweet thats the pvc issue then? | 14:23 |
BIOS_Hazard | Is there a place I can open a report for this all-in-one issue? | 14:23 |
BIOS_Hazard | I would be happy to submit my issue | 14:23 |
*** mgiles has joined #openstack-kolla | 14:24 | |
BIOS_Hazard | every other part of this process has been smooth so far | 14:24 |
srwilkers | sdake_, yeah, shouldve clarified. the patches are for the pvc issue. in terms of docs being ready, im hoping to have something up for review within a few days, but most likely post holiday | 14:24 |
sean-k-mooney | normally i would say open a bug here https://bugs.launchpad.net/kolla | 14:24 |
BIOS_Hazard | and knowing now that this precheck can be ignored for all-in-one, I am back to zero real issues | 14:24 |
sdake_ | srwilkers cool thanks | 14:24 |
sean-k-mooney | though with the repo split it might be https://bugs.launchpad.net/kolla-ansible now | 14:24 |
BIOS_Hazard | "the repo split" | 14:24 |
*** goldyfruit has joined #openstack-kolla | 14:25 | |
sean-k-mooney | BIOS_Hazard: the ansible deployment code is being moved into its own repo kolla-ansibel | 14:25 |
*** tonanhngo has joined #openstack-kolla | 14:25 | |
BIOS_Hazard | ah! | 14:25 |
BIOS_Hazard | thanks for clarifying | 14:25 |
sean-k-mooney | previously it was in the main kolla repo | 14:25 |
BIOS_Hazard | makes sense to separate | 14:25 |
BIOS_Hazard | I will use that one then per your recommendation | 14:26 |
*** tonanhngo has quit IRC | 14:26 | |
sean-k-mooney | the prechecks are part of the ansible playbooks so any bugs with them should go to https://bugs.launchpad.net/kolla-ansible now | 14:26 |
BIOS_Hazard | excellent | 14:26 |
BIOS_Hazard | and I will simply ignore these prechecks in my testing | 14:26 |
BIOS_Hazard | everything else set up with zero issues | 14:26 |
BIOS_Hazard | so I would expect this to work | 14:26 |
*** jtriley has joined #openstack-kolla | 14:26 | |
sean-k-mooney | the note in the docs say that the no ha proxy senario is not tested so this was obviosly not checked when the precheck was added. | 14:27 |
*** imcsk8 has quit IRC | 14:27 | |
portdirect | afternoon all o/ | 14:27 |
srwilkers | hey portdirect | 14:27 |
*** liyifeng has quit IRC | 14:28 | |
BIOS_Hazard | sean-k-mooney: would I find that in the globals file? or in the online docs? (just want to see where I missed it) | 14:28 |
sean-k-mooney | BIOS_Hazard: actully looking at master you should not be executing that test https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/prechecks/tasks/port_checks.yml#L18-L34 | 14:28 |
BIOS_Hazard | oh, I checked out stable/newton | 14:28 |
portdirect | hey srwilkers, looks like I've got a lot of k8s related logs to go through! | 14:28 |
BIOS_Hazard | should I be using master? | 14:28 |
BIOS_Hazard | hmm, my precheck playbook has this same code | 14:29 |
BIOS_Hazard | now I suspect I missed a configuration to disable haproxy :\ | 14:29 |
BIOS_Hazard | since this only runs if thats enabled (right?) | 14:29 |
*** imcsk8 has joined #openstack-kolla | 14:30 | |
sean-k-mooney | yes you sibly need to set enable_haproxy: False in the global.yml | 14:30 |
sean-k-mooney | BIOS_Hazard: https://github.com/openstack/kolla/blob/stable/newton/doc/advanced-configuration.rst#ip-address-constrained-environments | 14:31 |
BIOS_Hazard | hot damn! precheck successful | 14:31 |
sean-k-mooney | nice | 14:32 |
sean-k-mooney | just so you know the precheck can also be used multinode when you get to that point by add -i and specifying your multinode inventory | 14:32 |
BIOS_Hazard | Excellent :) | 14:32 |
BIOS_Hazard | Alright, looks like I am ready to deploy. Thanks again! | 14:33 |
sean-k-mooney | BIOS_Hazard: just so you are aware you last time i ran with enable_haproxy: "no" the deploy failed because seting up clustering for rabbitmq or mysql. | 14:34 |
sean-k-mooney | if you hit an issue with either let us know here because it may be a bug | 14:34 |
BIOS_Hazard | will do | 14:34 |
BIOS_Hazard | and otherwise, I can implement multi-node, just seemed easier to do AIO for the first test | 14:35 |
v1k0d3n | sdake_: ping | 14:35 |
*** srwilkers has quit IRC | 14:35 | |
*** srwilkers has joined #openstack-kolla | 14:35 | |
sean-k-mooney | yep all in one is simpler but as the docs say we dont regually test the enable_haproxy: "no" config so we only fix thing if people tell us its broken | 14:36 |
v1k0d3n | srwilkers mentioned you had some questions about halcyon? | 14:36 |
*** schwicht has joined #openstack-kolla | 14:36 | |
*** Bico_Fino has quit IRC | 14:36 | |
BIOS_Hazard | sean-k-mooney: I think thats reasonable. | 14:37 |
*** Bico_Fino has joined #openstack-kolla | 14:37 | |
*** srwilkers_ has joined #openstack-kolla | 14:39 | |
sean-k-mooney | v1k0d3n: halcyon? is that the kubernetes thing you wrote? | 14:41 |
sean-k-mooney | v1k0d3n: this https://github.com/v1k0d3n/halcyon-kubernetes ? | 14:42 |
openstackgerrit | zhubingbing proposed openstack/kolla-ansible: Add trove role https://review.openstack.org/400580 | 14:43 |
*** mmoster_ has quit IRC | 14:46 | |
*** YuYangWang has quit IRC | 14:49 | |
*** srwilkers_ has quit IRC | 15:04 | |
v1k0d3n | sean-k-mooney: yes | 15:06 |
*** srwilkers_ has joined #openstack-kolla | 15:06 | |
v1k0d3n | well, there's the vagrant one...which uses the link you sent as a submodule. | 15:06 |
v1k0d3n | so they can be used any way the operator really wants. | 15:06 |
openstackgerrit | Surya Prakash Singh proposed openstack/kolla-ansible: use uuidutils instead of uuid.uuid4() https://review.openstack.org/399847 | 15:07 |
v1k0d3n | i should probably add a sample inventory for admins looking to just use halcyon-kubernetes. | 15:07 |
sean-k-mooney | v1k0d3n: so the vagrent one reuses halcyon-kubernetes for all the heavy lifting | 15:07 |
v1k0d3n | (for things like bare-metal...since the vagrant one covers pretty much everything else non-bare-metal). | 15:08 |
portdirect | v1k0d3n: that would be awesome - I know a few folks who would prob pick it up if you added that. :) | 15:08 |
v1k0d3n | sean-k-mooney: yup. the vagrant one uses providers: currently libvirt, openstack and virtualbox (aws coming...it's just been a lower priority)...and then uses the submodule for doing the work. | 15:08 |
v1k0d3n | both are heavily influenced by variables, so the operator/developer as as many options as possible. | 15:09 |
sean-k-mooney | i was hopinging to extend the kolla-host playbook we have to deploy kubbernetes for kolla-kubernetess at some point but is halcyon going to be used for that instead? | 15:09 |
v1k0d3n | portdirect: the AWS portion would be helpful? | 15:09 |
*** papacz2 has quit IRC | 15:10 | |
v1k0d3n | sean-k-mooney: halcyon installs kubernetes, correct. | 15:10 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: NO_MERGE: test mitaka branch https://review.openstack.org/393877 | 15:10 |
v1k0d3n | it takes a kubeadm approach, since that is what will be supported by kuberenetes directly going forward. | 15:10 |
v1k0d3n | we're ok for now, since nobody is attempting to deploy "production" workloads of kolla-kube yet. as things firm up, we will shift support to kubeadm...(if needed) to make sure that is prod-ready in preparation for kolla-kube prod. | 15:11 |
v1k0d3n | there are so many things in flux at the moment. | 15:12 |
v1k0d3n | it would be nice to have an etherpad or some sort of document just to keep track of the state of things currently. | 15:12 |
*** senk has joined #openstack-kolla | 15:12 | |
portdirect | v1k0d3n: I'd like that, but think a few other people are looking at 'bare-metal' esq deployments. | 15:12 |
*** msimonin has joined #openstack-kolla | 15:13 | |
v1k0d3n | for instance, items about kubeadm...and where that's at. pvc's we're working with in 1.5.0beta1 because PetSets are no longer (in favor of StatefulSets...so there is a huge renaming effort going on for 1.5.0)...things like this impact the state of kolla0k8s. | 15:13 |
v1k0d3n | portdirect: bare-metal works today. just use https://github.com/v1k0d3n/halcyon-kubernetes directly. | 15:13 |
v1k0d3n | and oh! not that... | 15:14 |
v1k0d3n | sean-k-mooney: ... | 15:14 |
v1k0d3n | dont use v1k0d3n fork. use this one https://github.com/att-comdev/halcyon-vagrant-kubernetes | 15:14 |
sean-k-mooney | v1k0d3n: hi just in other channel reading back | 15:14 |
v1k0d3n | att-comdev | 15:14 |
*** senk_ has quit IRC | 15:14 | |
v1k0d3n | portdirect: if you want...you can use our bare-metal solution for MaaS in kubernetes as well if you want :) | 15:15 |
v1k0d3n | https://github.com/att-comdev/dockerfiles/tree/master/maas | 15:15 |
v1k0d3n | in deployment. | 15:15 |
*** senk_ has joined #openstack-kolla | 15:15 | |
sean-k-mooney | v1k0d3n: so currently the kolla host playbook can install the dependcies for kolla-ansible to deploy. i was thinking it might be nice to extend it to do the same for kolla-k8s but it might not be required | 15:16 |
portdirect | v1k0d3n: nice - I'll check that out for sure :) I'm started moving away from foreman (was also running that in k8s... <- dont do that) and have been playing with pixiecore quite a bit recently. | 15:17 |
*** inc0 has joined #openstack-kolla | 15:17 | |
inc0 | good morning | 15:17 |
*** senk has quit IRC | 15:18 | |
v1k0d3n | ah, portdirect i was using pixiecore as well. it's nice. danderson created that repo, and i believe he's now at coreos possibly? | 15:18 |
v1k0d3n | portdirect: were you using pixiecore for coreos installations? | 15:19 |
duonghq | morning inc0 | 15:19 |
v1k0d3n | or just for the simplicity + api? | 15:20 |
v1k0d3n | morning inc0 | 15:20 |
srwilkers_ | hey inc0 | 15:20 |
*** schwicht has quit IRC | 15:20 | |
inc0 | how it's going guys? | 15:21 |
*** diogogmt has quit IRC | 15:22 | |
portdirect | v1k0d3n: simplicity, though I've started to play with the api in the last week. I've basicly been using CentOS atomic, with an updated LTS Kernel for my stuff: https://github.com/portdirect/harbor/blob/latest/docker/mandracchio/mandracchio-repo-assets/assets/opt/mandracchio/harbor-host.json | 15:22 |
*** unicell has quit IRC | 15:25 | |
v1k0d3n | ah! ok. good stuff! | 15:25 |
v1k0d3n | this can be intalled with maas too if you're interested | 15:26 |
BIOS_Hazard | The all-in-one looks to have installed correctly. I was able to log into Horizon and everything. kolla is baller AF. thanks again for all the assistance! | 15:26 |
v1k0d3n | also...there is the bifrost/ironic stuff which will eventually make it's way into kolla-kube at some point. | 15:26 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: Bump OpenStack service version https://review.openstack.org/394271 | 15:28 |
sean-k-mooney | v1k0d3n: yep i keep leaning twords the idea of haveing a kolla-host repo that could be shared between kolla-ansibel and kolla-k8s to do all the host prep including biforst | 15:28 |
sean-k-mooney | i would also move ceph into that and maybe ovs at some point as it would allow use to do more advanced network configuration then what we do now | 15:29 |
sean-k-mooney | BIOS_Hazard: awsome | 15:30 |
portdirect | v1k0d3n: yeah, I need to look at maas: I was using foreman as it provided a nice intergration with FreeIPA; which I was using for to enroll hosts for Kerberos and Cert management/provisoning via certmonger. | 15:30 |
sean-k-mooney | BIOS_Hazard: glad to hear you got it working | 15:30 |
BIOS_Hazard | :) | 15:30 |
BIOS_Hazard | Looks like I forgot to enable ceph when I deployed | 15:30 |
BIOS_Hazard | Can I just enable it and re-run deploy? | 15:30 |
sean-k-mooney | BIOS_Hazard: you might need to run deploy then reconfigure | 15:31 |
sean-k-mooney | BIOS_Hazard: that is not really something we have tested | 15:31 |
BIOS_Hazard | fair enough | 15:31 |
BIOS_Hazard | im just gonna see what happens! I can always re-install :P | 15:31 |
sean-k-mooney | BIOS_Hazard: just so you know singel node ceph is not really a thing. you can do it but normally you should not | 15:32 |
*** papacz has joined #openstack-kolla | 15:32 | |
*** unicell has joined #openstack-kolla | 15:32 | |
BIOS_Hazard | sean-k-mooney: understood. I just have some proof-of-concept stuff I want to try and this seemed like the ideal approach. I do understand that this approach is not production-compatible | 15:33 |
sean-k-mooney | BIOS_Hazard: well its not that. ceph upstream is not intended to run on one server it assumes 3 by default | 15:33 |
BIOS_Hazard | oh, like it will fail? | 15:34 |
sean-k-mooney | BIOS_Hazard: https://github.com/openstack/kolla/blob/stable/newton/doc/ceph-guide.rst#managing-ceph has the extra stpes you need to do for all in one | 15:34 |
BIOS_Hazard | sean-k-mooney: thanks!!! | 15:34 |
*** tonanhngo has joined #openstack-kolla | 15:34 | |
sean-k-mooney | BIOS_Hazard: actully there are a few other steps but all covered in that guide | 15:34 |
inc0 | looking for someone knowing puppet | 15:35 |
*** tonanhngo has quit IRC | 15:36 | |
BIOS_Hazard | ah yes, I remember seeing this now... | 15:36 |
BIOS_Hazard | [global] | 15:36 |
BIOS_Hazard | osd pool default size = 1 | 15:36 |
BIOS_Hazard | osd pool default min size = 1 | 15:36 |
BIOS_Hazard | thanks! | 15:36 |
sean-k-mooney | inc0: sorry one of the reason i like kolla is the lack of puppet | 15:37 |
*** schwicht has joined #openstack-kolla | 15:37 | |
*** sayantani01 has joined #openstack-kolla | 15:38 | |
inc0 | sean-k-mooney, yeah...but that puts as at disadvantage in writing support for stuff in infra | 15:38 |
inc0 | stuff we could really use | 15:38 |
sean-k-mooney | inc0: infra use ansible to orcestrate puppet right | 15:39 |
inc0 | yeah | 15:39 |
inc0 | but module itself has to be puppet unfortunatly | 15:39 |
sean-k-mooney | that makes my head hurt a little but i understand why they did it | 15:39 |
sdake_ | v1k0d3n ya srwilkers_ answered em thanks | 15:39 |
sean-k-mooney | inc0: one thing i would like to add to kolla is the ablity to deploy infra services e.g. zuul/nodepool nativly so that kolla could spin up an instace of the upstream ci plus provided the ci cloud | 15:40 |
sean-k-mooney | inc0: our ci master node is currently deployed with the puppet script from infra but we use kolla for the cloud | 15:41 |
sdake_ | v1k0d3n we have something that keeps track of state in openstack called launchpad :) | 15:42 |
inc0 | I think it's cool idea, just infra people are overworked even now | 15:42 |
sean-k-mooney | inc0: well i was thinking that our team might do it when we get time | 15:42 |
inc0 | "when we get time";) | 15:42 |
zhubingbing | In 20 minutes later we have a weekly meeting | 15:42 |
zhubingbing | ;) | 15:42 |
sean-k-mooney | inc0: we have higher proiry stuff for our ci but i added it too our ci backlog. | 15:43 |
*** prameswar has joined #openstack-kolla | 15:43 | |
*** berendt has quit IRC | 15:44 | |
sean-k-mooney | inc0: assuming we are both there we should discuss this at the ptg. by then i hope we will have capasity to look at doing it. it would be a nice addtion for pike | 15:45 |
*** berendt has joined #openstack-kolla | 15:45 | |
v1k0d3n | sdake_: you keep track of the state of kubernetes and helm in launchpad? | 15:46 |
sdake_ | v1k0d3n state of implementation yup | 15:47 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: DO_NOT_MERGE: TEST MASTER BRANCH https://review.openstack.org/399897 | 15:47 |
sdake_ | each review links to launchpad | 15:47 |
sdake_ | then reviews are reviewed by core reviewers | 15:47 |
sdake_ | and hashed out | 15:47 |
sdake_ | https://launchpad.net/kolla-kubernetes/+milestone/ocata-2 | 15:47 |
v1k0d3n | i wouldn't even know where to begin keeping STATE tracking on other projects... | 15:47 |
v1k0d3n | this is very much in daily flux. | 15:48 |
sdake_ | nah not on other projects | 15:48 |
*** schwicht has quit IRC | 15:48 | |
v1k0d3n | ok | 15:48 |
sdake_ | on kolla itself | 15:48 |
v1k0d3n | i'm talking about other projects. | 15:48 |
v1k0d3n | because kolla has impacts. | 15:48 |
v1k0d3n | like petset vs statefulset | 15:48 |
*** Bico_Fino_ has joined #openstack-kolla | 15:48 | |
*** Bico_Fino has quit IRC | 15:49 | |
*** Bico_Fino_ is now known as Bico_Fino | 15:49 | |
v1k0d3n | like the current state and tracking of userland pvc claims for rdb. | 15:49 |
v1k0d3n | things like that. | 15:49 |
v1k0d3n | they impact sure, but kolla will need to work around. | 15:49 |
sdake_ | yup seems like a reasonable thing to track if someone cares to maintain it :) | 15:50 |
*** Serlex has quit IRC | 15:52 | |
sean-k-mooney | maybe you could use tags to track it? | 15:53 |
Bico_Fino | sean-k-mooney: I upgrade the ironic-ui version to 2.1.1 | 15:53 |
Bico_Fino | sean-k-mooney: Got this error -> You have offline compression enabled but key "8745f6f37b1fe160569fae654663c71a" is missing from offline manifest. You may need to run "python manage.py compress". | 15:53 |
Bico_Fino | looks easy to fix. ;) | 15:53 |
sean-k-mooney | Bico_Fino: did you run the upgrade playbook to change the version or redeploy with that version? | 15:54 |
v1k0d3n | so just as an example sdake_: https://github.com/kubernetes/kubernetes/issues/27430 | 15:54 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla: DO_NOT_MERGE: TEST NEWTON BRANCH https://review.openstack.org/386981 | 15:54 |
v1k0d3n | so there is a hard break from 1.4 to 1.5 | 15:54 |
*** tonanhngo has joined #openstack-kolla | 15:54 | |
Bico_Fino | sean-k-mooney: nope. Where is this upgrade playbook? | 15:54 |
v1k0d3n | the resource type from petset will no longer be available, and there is no plan to keep running workloads either. | 15:55 |
v1k0d3n | it's a hard break. things like this need tracked, otherwise a project like kolla-k8s will be in a bug fix mode often. | 15:55 |
*** tonanhngo has quit IRC | 15:56 | |
sean-k-mooney | Bico_Fino: you can use kolla-ansible updgade to upgrade between releases but that was never tested for ironic as well ironic was broken untill this releases | 15:56 |
v1k0d3n | keeping an eye ahead in one place would help ease fixes later. | 15:56 |
Bico_Fino | sean-k-mooney: but that will work for plugins? Example: ironic-ui | 15:56 |
sean-k-mooney | Bico_Fino: well for the plugins you can just run python manage.py compress in the template override i guess | 15:57 |
sean-k-mooney | upgrade works by upgrading the docker images so if the plugin is in that it might work | 15:57 |
Bico_Fino | sean-k-mooney: I just entered the horizon container and installed the plugin | 15:58 |
sean-k-mooney | in any case we actully dont use the horizon plugin for ironic in our enviroment so not sure | 15:58 |
inc0 | guess what's time is it!? its meeeeeting time! | 15:58 |
sean-k-mooney | Bico_Fino: ah there is a way to do that at build time | 15:58 |
Bico_Fino | sean-k-mooney: the ironic-ui is working now, but the button Enroll Node don’t. | 15:59 |
Bico_Fino | Need do more tests. | 15:59 |
sean-k-mooney | #openstack-meeting3? | 15:59 |
duonghq | do we still use meeting-4? | 15:59 |
sean-k-mooney | duonghq: i guess we should check the wiki | 15:59 |
duonghq | wiki tell that we use no-4 | 16:00 |
sean-k-mooney | cool | 16:00 |
duonghq | inc0 sdake_ , ping | 16:00 |
openstackgerrit | satya proposed openstack/kolla-ansible: Added separate interface for rabbitmq as rabbitmq requires primary interface to communicate https://review.openstack.org/401329 | 16:00 |
sdake_ | sup | 16:00 |
duonghq | do we use meeting4 as usual? | 16:00 |
*** senk_ has quit IRC | 16:01 | |
sdake_ | yes duonghq | 16:01 |
*** jascott1 has joined #openstack-kolla | 16:01 | |
sean-k-mooney | Bico_Fino: https://github.com/openstack/kolla/blob/stable/newton/doc/image-building.rst#plugin-functionality discribes how to add plugins at build time | 16:01 |
duonghq | hmm Heat guys still use that | 16:02 |
Bico_Fino | thanks sean-k-mooney | 16:02 |
*** Pavo has quit IRC | 16:02 | |
Satya_ | hi | 16:03 |
Satya_ | please review https://review.openstack.org/#/c/401329/ | 16:03 |
Satya_ | jeffrey there? | 16:03 |
Satya_ | hi sdake | 16:03 |
*** tonanhngo has joined #openstack-kolla | 16:03 | |
*** schwicht has joined #openstack-kolla | 16:04 | |
*** strigazi is now known as strigazi_AFK | 16:04 | |
*** Pavo has joined #openstack-kolla | 16:07 | |
*** magicboiz has joined #openstack-kolla | 16:09 | |
*** lrensing has joined #openstack-kolla | 16:09 | |
Bico_Fino | sean-k-mooney: fixed the Enroll issue, collectstatic is our friend. ;) | 16:10 |
*** cnf has left #openstack-kolla | 16:10 | |
kristia__ | sdake_: Hi do you think gpu passthrough will work on openstack-ansible? | 16:11 |
*** DaveTurner has joined #openstack-kolla | 16:14 | |
v1k0d3n | so sdake_working on the pvc today. | 16:16 |
v1k0d3n | most likely tonight, honestly. | 16:16 |
sdake_ | v1k0d3n sweet - its key for 1.0.0 of kolla to be a reality | 16:17 |
v1k0d3n | talking with kubeadm folks on how i could add custom controller to it. | 16:17 |
sdake_ | rather kolla-kubernetes | 16:17 |
v1k0d3n | yup | 16:17 |
*** eaguilar has joined #openstack-kolla | 16:21 | |
*** mdnadeem has joined #openstack-kolla | 16:22 | |
*** fguillot has quit IRC | 16:24 | |
*** jgriffith_away is now known as jgriffith | 16:25 | |
*** unicell has quit IRC | 16:25 | |
*** fguillot has joined #openstack-kolla | 16:26 | |
*** dkehn__ has joined #openstack-kolla | 16:28 | |
*** Satya_ has quit IRC | 16:31 | |
*** Satya_ has joined #openstack-kolla | 16:31 | |
openstackgerrit | satya proposed openstack/kolla-ansible: Added separate interface for rabbitmq as rabbitmq requires primary interface to communicat https://review.openstack.org/401341 | 16:33 |
*** srwilkers_ has left #openstack-kolla | 16:33 | |
*** eaguilar has quit IRC | 16:36 | |
*** kristia__ is now known as kristian__ | 16:36 | |
*** senk has joined #openstack-kolla | 16:37 | |
*** zhubingbing__ has joined #openstack-kolla | 16:37 | |
openstackgerrit | satya proposed openstack/kolla-ansible: Added separate interface for rabbitmq as rabbitmq requires primary interface to communicate https://review.openstack.org/401341 | 16:39 |
*** zhubingbing has quit IRC | 16:39 | |
*** mdnadeem has quit IRC | 16:42 | |
*** schwicht has quit IRC | 16:48 | |
*** rmart04 has quit IRC | 16:50 | |
Satya_ | Hi jeffrey4l | 16:56 |
Jeffrey4l | Satya_, sup | 16:56 |
*** Pavo has quit IRC | 16:56 | |
Satya_ | Please review https://review.openstack.org/#/c/401341/ | 16:57 |
*** rhallisey_ has joined #openstack-kolla | 16:57 | |
Jeffrey4l | ok | 16:57 |
Satya_ | Thanks | 16:57 |
Satya_ | i messed with the precious commit so abandoned that... | 16:58 |
*** Pavo has joined #openstack-kolla | 16:59 | |
v1k0d3n | and we shift back to here :) | 16:59 |
Pavo | morning | 16:59 |
duonghq | morning Pavo | 17:00 |
*** sp__ has quit IRC | 17:00 | |
portdirect | ok I'm quite up for documenting getting a kolla-k8s dev env set up (also morning Pavo) | 17:00 |
zhubingbing__ | hi | 17:00 |
zhubingbing__ | i have 2 problemes | 17:00 |
zhubingbing__ | about elk | 17:01 |
zhubingbing__ | 1.https://blueprints.launchpad.net/kolla/+spec/heka-deprecation | 17:01 |
Pavo | I agree portdirect | 17:01 |
zhubingbing__ | 2.https://blueprints.launchpad.net/kolla/+spec/bp-elasticsearch-upgrade | 17:01 |
Pavo | and morning duonghq and portdirect | 17:01 |
*** srwilkers_ has joined #openstack-kolla | 17:01 | |
Pavo | so........ who wants to walk me through a kolla-k8s deployment? | 17:04 |
*** athomas has quit IRC | 17:04 | |
Pavo | as in from start to finish | 17:04 |
pbourke | zhubingbing__: what are the problems | 17:05 |
zhubingbing__ | 1.https://blueprints.launchpad.net/kolla/+spec/heka-deprecation | 17:05 |
pbourke | that is a spec, not a problem :) | 17:05 |
zhubingbing__ | We have not considered this matter | 17:06 |
pbourke | ok I get you | 17:06 |
portdirect | Pavo: I dont want to yet; as we dont yet have to docs together for the current state of the project - soon as we do I'd love you to be the guinea pig; your input and insight would be really valuble | 17:06 |
Pavo | I am game | 17:06 |
pbourke | zhubingbing__: someone will need to research both, draw up a conclusion in a spec | 17:06 |
zhubingbing__ | We have not decided which one, so I can go to the test | 17:06 |
zhubingbing__ | ok | 17:06 |
Pavo | and also game for anything else | 17:06 |
pbourke | zhubingbing__: sounds good | 17:06 |
*** egonzalez90 has quit IRC | 17:07 | |
Pavo | also got my TLS working from outside portdirect | 17:08 |
sdake_ | oenstack workflow http://docs.openstack.org/infra/manual/developers.html | 17:08 |
Pavo | finally | 17:08 |
sdake_ | rather OpenStack workflow | 17:08 |
Pavo | check it out https://ddi.hopto.org | 17:08 |
sdake_ | this unfortunately doesn't cover blueprints much | 17:08 |
sdake_ | mgiles ^^ | 17:08 |
Pavo | can someone test the non https redirect for me at ddi.hopto.org | 17:08 |
mgiles | sdake thanks | 17:08 |
mgiles | rather... sdake_ thanks. :) | 17:09 |
*** sdake_ is now known as sdake | 17:09 | |
sdake | np mgiles | 17:09 |
openstackgerrit | Merged openstack/kolla: Remove adding repo in magnum-base Dockerfile https://review.openstack.org/401032 | 17:10 |
openstackgerrit | Merged openstack/kolla: Bump OpenStack service version https://review.openstack.org/394271 | 17:10 |
portdirect | whoop! routing? | 17:10 |
duonghq | sdake, who is take care helm packaging/neutron? | 17:10 |
zhubingbing__ | pbourke thank you | 17:11 |
sdake | duonghq sign up for work item in the blueprints | 17:11 |
duonghq | I'm try to grab something from work items | 17:11 |
duonghq | but neutron is in progress without assignee | 17:11 |
*** senk has quit IRC | 17:11 | |
duonghq | I'm confusing | 17:11 |
sdake | duonghq yup - workflow hard :) | 17:12 |
sdake | duonghq moment | 17:12 |
sdake | https://launchpad.net/kolla-kubernetes/+milestone/ocata-2 | 17:12 |
duonghq | as I see, it's kfox1111? | 17:12 |
*** rhallisey_ has quit IRC | 17:12 | |
Jeffrey4l | Satya_, done nit. | 17:12 |
sdake | click the helm link | 17:12 |
*** rhallisey_ has joined #openstack-kolla | 17:12 | |
sdake | it has work items in it | 17:12 |
duonghq | sdake, I'm already on this bp, I mean the work items in this | 17:13 |
duonghq | for example: the neutron is inprogress w/o assignee | 17:13 |
sdake | duonghq scroll down | 17:13 |
sdake | duonghq to WORK ITEMS | 17:13 |
sdake | or search for unassigned | 17:13 |
duonghq | I'm try to find out with item do not have anybody take care yet | 17:14 |
sdake | those are the ones that are unassigned | 17:14 |
sdake | duonghq right neutron is kfox1111 | 17:14 |
duonghq | (unassigned)neutron: INPROGRESS <- somebody should update this | 17:14 |
duonghq | ya | 17:14 |
portdirect | duonghq: it would be great if someone took on mariadb, and keystone | 17:15 |
sdake | duonghq duonghq can you update it since your there | 17:15 |
duonghq | so I'll take the mariadb or nova, need some review | 17:16 |
duonghq | I'll update tomorrow | 17:16 |
duonghq | in the bp | 17:16 |
sdake | duonghq sounds good | 17:16 |
portdirect | ok - I can get going on keystone | 17:17 |
*** msimonin has quit IRC | 17:17 | |
portdirect | then we will at least have a base for everything else above it :) | 17:17 |
duonghq | portdirect, I can update the bp for you | 17:18 |
duonghq | portdirect, can I mark it as inprogress? | 17:18 |
portdirect | duonghq: great - thanks (what TZ are you in?) | 17:18 |
duonghq | UTC+7, | 17:19 |
portdirect | duonghq: it will be in 8 hours :) | 17:19 |
duonghq | portdirect, np | 17:19 |
rhallisey_ | sdake, did you get kube setup> | 17:19 |
sdake | rhallisey_ nope and have to jet atm | 17:19 |
sdake | family emergency | 17:19 |
sdake | see folks later | 17:19 |
rhallisey_ | kk | 17:19 |
duonghq | cya sdake | 17:19 |
sdake | enjoy holidays if you got em | 17:20 |
portdirect | bye sdake | 17:20 |
*** portdirect is now known as portdirect_away_ | 17:21 | |
openstackgerrit | Merged openstack/kolla: Install mkfs.vfat tool in ironic-conductor image https://review.openstack.org/400709 | 17:22 |
*** portdirect_away_ is now known as portdirect_back_ | 17:23 | |
duonghq | I need some sleep, see you later guys | 17:26 |
*** duonghq has quit IRC | 17:27 | |
*** hogepodge has joined #openstack-kolla | 17:28 | |
openstackgerrit | zhubingbing proposed openstack/kolla-ansible: Add trove role https://review.openstack.org/400580 | 17:28 |
*** diogogmt has joined #openstack-kolla | 17:29 | |
*** matrohon has quit IRC | 17:30 | |
*** hogepodge has quit IRC | 17:33 | |
openstackgerrit | zhubingbing proposed openstack/kolla-ansible: Add trove role https://review.openstack.org/400580 | 17:34 |
*** zhubingbing__ has quit IRC | 17:37 | |
pbourke | vpnaas is fubared on the redhat base | 17:38 |
pbourke | im very surprised it works even for ubuntu | 17:39 |
*** Bico_Fino_ has joined #openstack-kolla | 17:39 | |
*** Bico_Fino has quit IRC | 17:40 | |
*** Bico_Fino_ is now known as Bico_Fino | 17:40 | |
*** msimonin has joined #openstack-kolla | 17:41 | |
*** shardy has quit IRC | 17:43 | |
*** mkoderer has quit IRC | 17:43 | |
*** hogepodge has joined #openstack-kolla | 17:43 | |
pbourke | Jeffrey4l: ping | 17:44 |
Jeffrey4l | pbourke, pong | 17:44 |
pbourke | Jeffrey4l: im just looking at your 'move glance precheck' change | 17:45 |
pbourke | Jeffrey4l: whats the purpose of the new module | 17:45 |
openstackgerrit | Merged openstack/kolla: Update docs around TrivialFix https://review.openstack.org/397712 | 17:45 |
Jeffrey4l | pbourke, yep. get all container status by using one module. | 17:45 |
openstackgerrit | satya proposed openstack/kolla-ansible: Added separate interface for rabbitmq as rabbitmq requires primary interface to communicate https://review.openstack.org/401341 | 17:45 |
Jeffrey4l | just like ansible's docker_image_facts | 17:45 |
Satya_ | hi jeffrey | 17:46 |
*** matrohon has joined #openstack-kolla | 17:46 | |
Satya_ | accommodated the comment | 17:46 |
Jeffrey4l | Satya_, hi. ( please use jeffrey4l rather than jeffrey, then i can get a notification ;) ) | 17:47 |
Satya_ | sue :) jeffrey4l | 17:47 |
Jeffrey4l | much better ;) | 17:47 |
Satya_ | Thanks | 17:47 |
pbourke | Jeffrey4l: why not just kolla_docker | 17:48 |
Jeffrey4l | kolla_docker just support one container and normally it failed when container is not exist. | 17:49 |
Jeffrey4l | pbourke, | 17:49 |
pbourke | Jeffrey4l: it would just be nice if we could get away without another module | 17:50 |
pbourke | Jeffrey4l: but if you think not then I get it's ok | 17:50 |
pbourke | *guess | 17:50 |
*** lrensing has quit IRC | 17:51 | |
Jeffrey4l | i think it should be OK. extend kolla_docker will make it more complicated and kolla_container_facts make it simple enough. ;) | 17:52 |
v1k0d3n | portdirect_back_: | 17:52 |
Jeffrey4l | i should be OK to write our own module, especially when it do not exist in official modules. | 17:52 |
v1k0d3n | hey man, so for the changes for pvc ( inc0 you may be interested in this as well)...i'm just replacing the following: | 17:53 |
*** eaguilar has joined #openstack-kolla | 17:53 | |
Jeffrey4l | pbourke, ^^ | 17:53 |
v1k0d3n | in /etc/kubernetes/manifests/kube-controller-manager.json just make that line point to our image: "image": "quay.io/attcomdev/kube-controller-manager:v1.5.0-beta.1" | 17:53 |
v1k0d3n | and boom. pvc works for rdb | 17:53 |
*** sdake has quit IRC | 17:54 | |
*** kristian__ has quit IRC | 17:55 | |
*** dave-mccowan has joined #openstack-kolla | 17:56 | |
*** dave-mccowan has quit IRC | 17:57 | |
*** alteriks has quit IRC | 17:58 | |
*** ppalacios has quit IRC | 18:00 | |
Pavo | anyone know how to fix init-runonce when using TLS I am getting these errors | 18:00 |
Pavo | http://paste.openstack.org/show/590242/ | 18:00 |
*** Pavo has quit IRC | 18:02 | |
*** Pavo has joined #openstack-kolla | 18:03 | |
*** senk has joined #openstack-kolla | 18:05 | |
*** Bico_Fino has quit IRC | 18:07 | |
sbezverk | v1k0d3n: if you are going (I hope you do) push this fix upstream, please let me know Kubernetes issue ID.. | 18:08 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: NO_MERGE: Testing https://review.openstack.org/401370 | 18:10 |
openstackgerrit | Jeffrey Zhang proposed openstack/kolla-ansible: NO_MERGE: Testing https://review.openstack.org/401370 | 18:11 |
*** senk has quit IRC | 18:13 | |
*** pbourke has quit IRC | 18:14 | |
*** pbourke has joined #openstack-kolla | 18:14 | |
*** senk has joined #openstack-kolla | 18:14 | |
*** senk has quit IRC | 18:17 | |
*** eaguilar has quit IRC | 18:17 | |
v1k0d3n | yeah, really strange why the kube-controller just doesn't include the rbd fixes... | 18:19 |
v1k0d3n | sbezverk: ^^ | 18:19 |
*** portdirect_back_ is now known as portdirect_ | 18:20 | |
v1k0d3n | you saw my notes on getting it working though? | 18:20 |
portdirect_ | v1k0d3n: yo! | 18:20 |
v1k0d3n | hey portdirect_ | 18:20 |
v1k0d3n | you see my message? | 18:20 |
portdirect_ | no? | 18:20 |
v1k0d3n | in /etc/kubernetes/manifests/kube-controller-manager.json just make that line point to our image: "image": "quay.io/attcomdev/kube-controller-manager:v1.5.0-beta.1" | 18:21 |
v1k0d3n | then rdb patches are in place | 18:21 |
v1k0d3n | for pvc | 18:21 |
portdirect_ | oh - sweet | 18:21 |
v1k0d3n | going to update update halcyon | 18:21 |
*** eaguilar has joined #openstack-kolla | 18:21 | |
v1k0d3n | sbezverk: then you can use halcyon for the ceph rdb stuff you've been working on as well | 18:22 |
v1k0d3n | :) | 18:22 |
*** unicell has joined #openstack-kolla | 18:24 | |
*** unicell has quit IRC | 18:25 | |
*** unicell has joined #openstack-kolla | 18:27 | |
inc0 | ok, so gates and registry | 18:27 |
inc0 | Jeffrey4l, moved good point, if we implement registry in infra | 18:28 |
inc0 | depends-on cross project wont' work | 18:28 |
inc0 | as images in registry will always be out of merged changes | 18:28 |
inc0 | so you can't gate ansible code on unmerged change to images | 18:28 |
Jeffrey4l | yep. we need support two case: when there is no depends-on in commit message, use the tarball registry directly. otherwise, build the image by using the depends-on kolla ref | 18:29 |
inc0 | that's going to be hard to implement:/ | 18:30 |
*** msimonin has quit IRC | 18:31 | |
sean-k-mooney | any reson not to tag images with the changeid | 18:32 |
sean-k-mooney | then for depends on you could pull the tagged images | 18:32 |
*** ppalacios has joined #openstack-kolla | 18:33 | |
sean-k-mooney | Jeffrey4l: inc0 ^^ | 18:33 |
inc0 | sean-k-mooney, thing is, we want to pull images from unmerged change | 18:33 |
inc0 | and we won't keep all unmerged change images in registry | 18:33 |
*** ppalacios has left #openstack-kolla | 18:33 | |
sean-k-mooney | hum ok if you dont keep all unmerge changes then yes that wont work | 18:34 |
inc0 | keeping unmerged changes in registry would destroy it;) | 18:34 |
inc0 | every patchset -> new build and upload | 18:34 |
inc0 | and it wouldn't take long to have 1000s of garbage images | 18:35 |
Jeffrey4l | i do not think tarball site can has enough disk for tagging images with changeid. since the tarball file will not be removed or override. | 18:35 |
sean-k-mooney | yes basically with a periodic task to delete untagged images | 18:35 |
inc0 | so no, we will only keep up-to-date master images | 18:35 |
sean-k-mooney | ok then as Jeffrey4l said i guess you will need two workflows | 18:35 |
sean-k-mooney | one for normal commits and second for dependson | 18:36 |
inc0 | or we just agree to having red gates until kolla change merges | 18:36 |
inc0 | and do recheck after it merges | 18:36 |
inc0 | not ideal, but I for one think it will be managable | 18:36 |
inc0 | you need to wait for kolla merge before ansible code merge anyway | 18:36 |
inc0 | and gates aren't meant for debugging, you should have it tested locally | 18:36 |
Jeffrey4l | do u mean do not use depends-on in any way? | 18:37 |
inc0 | no, you do use depends-on | 18:37 |
inc0 | from ansible to kolla | 18:37 |
sean-k-mooney | well yes and no. unit tests basic deployment yes you should do locallly gates are for testing full intergration | 18:37 |
inc0 | just not for gates - to ensure that ansible change won't merge before kolla change | 18:37 |
Jeffrey4l | hmm, let me think.. | 18:37 |
inc0 | kolla change merges - > you do recheck on ansible patchset and gates should green-up | 18:38 |
Jeffrey4l | i strongly against this gates aren't meant for debugging, you should have it tested locally | 18:38 |
inc0 | what he said^ | 18:38 |
inc0 | infra has limited resources as they are | 18:38 |
inc0 | let's not use them to avoid testing code locally | 18:38 |
sean-k-mooney | yes but oftten things break in the gate that work locally. | 18:39 |
sean-k-mooney | well not often. i test every locally before i push but not everyone does | 18:39 |
inc0 | yeah, but that will become apparent after kolla change emrges | 18:39 |
Jeffrey4l | sean-k-mooney, yep. | 18:39 |
inc0 | only thing you lose with this solution is that you won't see this coming *before* change to kolla merges | 18:40 |
inc0 | Jeffrey4l, how about addint experimental gate to ansible with build and deploy | 18:40 |
inc0 | so you'll be able to call it on demand | 18:40 |
Jeffrey4l | inc0, nice solution ;) | 18:40 |
Jeffrey4l | like it. | 18:41 |
inc0 | it will limit our options regarding scenerios tested | 18:41 |
inc0 | (as seting up registry fees up lots of resources and time and we can use it to increase number of scenerios) | 18:41 |
inc0 | but it's something | 18:41 |
Jeffrey4l | acually depends-on should be edge case, there should few patch need this. | 18:41 |
inc0 | exactly | 18:41 |
*** jgriffith is now known as jgriffith_away | 18:42 | |
inc0 | and we should not use gates for debugging anyway | 18:42 |
sean-k-mooney | well any patch chain that adds a new service will | 18:42 |
Jeffrey4l | OK. let's make things more clear. | 18:42 |
sean-k-mooney | but yes they should be the minority of changes | 18:42 |
inc0 | sean-k-mooney, yeah, but I hope we won't add that many new services tbh;) | 18:42 |
inc0 | I think we have lots of them already | 18:43 |
Jeffrey4l | 1. in kolla, how the deploy gate image come from? should be build locally, right? | 18:43 |
inc0 | kolla will have image building | 18:43 |
Jeffrey4l | sean-k-mooney, when adding new service, there should be no gate/job to test it. so depends-on is useless. | 18:43 |
inc0 | not sure how we'll solve deploy in kolla, but I guess we can use zuul cloner for both deploy gates for ansible and k8s | 18:43 |
inc0 | yeah, Jeffrey4l is right, unless you create gate scenerio for your service in same patch, it's not going to do anything anyway | 18:44 |
sean-k-mooney | inc0: well kolla could use the master of kolla-ansible or you could say that kolla will not deploy at all jsut build but then you have an issue with docker changes breakin ansible | 18:44 |
inc0 | Jeffrey4l, but I think kolla changes should never have depend-on ansible change | 18:44 |
inc0 | which means change to image is good if master kolla-ansible works | 18:45 |
Jeffrey4l | inc0, deploy in kolla should use building image, and use zuul cloner to pull kolla-ansible or kolla-k8s. | 18:45 |
inc0 | if one of changes ongoing in kolla-ansible doesn't well, tough luck | 18:45 |
Jeffrey4l | hrm | 18:46 |
sean-k-mooney | if you dont have a depens-on form kolla-ansible to kolla you could have ansible chages merge that expect contianers that have not merged so depens on is needed | 18:46 |
inc0 | kolla is common core | 18:46 |
inc0 | it's baseline | 18:46 |
*** sayantani01 has quit IRC | 18:46 | |
*** portdirect_ is now known as portdirect_away | 18:46 | |
inc0 | changes to kolla takes precedence on changes to kolla-ansible | 18:46 |
*** sayantani01 has joined #openstack-kolla | 18:47 | |
Jeffrey4l | inc0, technical yes. but depends-on make it easier. and when u using zuul-cloner, you can use depends-on | 18:47 |
inc0 | yeah, but again, deadlock | 18:47 |
inc0 | changes can't have depends-on each other | 18:47 |
inc0 | as zuul won't allow it to merge before other merges | 18:48 |
inc0 | and that creates deadlock | 18:48 |
Jeffrey4l | we can use depends-on in kolla. use tarball image in kolla-ansilbe/kolla-k8s | 18:48 |
sean-k-mooney | well if you never allow depens-on from kolla to kolla-* then no deadlock | 18:48 |
inc0 | so dependency should always go ansible > kolla | 18:48 |
inc0 | thats my point | 18:48 |
*** kristian__ has joined #openstack-kolla | 18:48 | |
Jeffrey4l | inc0, ansible-> kolla only use by experimental. | 18:48 |
Jeffrey4l | right? | 18:48 |
inc0 | so deploy gates in Kolla should *always* pull merged master from ansible | 18:48 |
inc0 | no, hold on | 18:49 |
inc0 | depends-on is more than just pulling code in gates | 18:49 |
*** cnf has joined #openstack-kolla | 18:49 | |
inc0 | it's main role is that Zuul gates will -2 merge of changeX if it depends on unmerged changeY | 18:49 |
*** dave-mccowan has joined #openstack-kolla | 18:49 | |
Jeffrey4l | yep, you are right. kolla -> kolla-ansible is bad. | 18:49 |
*** rhallisey_ has quit IRC | 18:49 | |
inc0 | so you can put 2*+2+A on ansible change but it won't land until depends-on change lands | 18:49 |
sean-k-mooney | yep that is its main role | 18:50 |
inc0 | so kolla will always pull master ansible and k8s | 18:50 |
Jeffrey4l | inc0, get your point. so rule: never use depends-on in kolla project. | 18:50 |
inc0 | and kolla-ansible will use tarballed registry + experimental gate with build | 18:50 |
inc0 | yeah | 18:50 |
*** newmember has quit IRC | 18:50 | |
inc0 | how does that sound? | 18:51 |
sean-k-mooney | depens on in kolla could be uses aslong as you dont depend on kolla-ansibel or kolla-k8s | 18:51 |
Jeffrey4l | the deploy jobs in kolla will never be voting, right? inc0 | 18:51 |
Jeffrey4l | sean-k-mooney, yep. | 18:51 |
sean-k-mooney | e.g. depens-on to glance would be fine for say a pending config change | 18:51 |
inc0 | Jeffrey4l, I don't expect them to, no | 18:51 |
inc0 | but if never? dunno | 18:52 |
Jeffrey4l | sean-k-mooney, but kolla depends-on nothing actually. | 18:52 |
inc0 | not initially | 18:52 |
*** eaguilar has quit IRC | 18:52 | |
*** kristian__ has quit IRC | 18:53 | |
Jeffrey4l | inc0, how about in kolla-ansible use build image when depends-on exist in commit message? | 18:53 |
sean-k-mooney | yes but as the config gen code will live in kolla if i am extending kolla for an unmerged config change i can and should use depends-on to track that | 18:53 |
sean-k-mooney | anyway that beside the point sorry | 18:53 |
inc0 | Jeffrey4l, then it can't be voting | 18:53 |
*** sayantani01 has quit IRC | 18:53 | |
inc0 | I don't want to change voting gate logic based on commit message.. | 18:54 |
inc0 | so, I'd rather have separate gates | 18:54 |
inc0 | for build | 18:54 |
Jeffrey4l | i do not it change anything. | 18:54 |
Jeffrey4l | the ansible changes won't be merged until its depends docker change is merged. | 18:55 |
sean-k-mooney | why not only execute the deploy in the gate pipeline and just to do the image build in the check pipeline | 18:55 |
*** sayantani01 has joined #openstack-kolla | 18:56 | |
inc0 | sean-k-mooney, most of the time we won't need to build images at all | 18:56 |
Jeffrey4l | sean-k-mooney, for kolla project? isn't there a rule: gate pipeline should be exist in check pipeline? | 18:56 |
sean-k-mooney | so before you merge a docker change it will pull master or stable/X ansible and test it can deploy but normaly it will only check it can build | 18:56 |
inc0 | why waste time? | 18:56 |
*** sayantani01 has quit IRC | 18:56 | |
*** sayantani01 has joined #openstack-kolla | 18:56 | |
sean-k-mooney | for kolla. it should save time as you dont always have run the deploy job | 18:57 |
Jeffrey4l | inc0, if we do not support depends-on, we break the zuul voting/merge progress, imho | 18:57 |
*** DaveTurner has quit IRC | 18:57 | |
sean-k-mooney | changes to kolla would only trigger the build job and would only trigger deploy job after +2+w. just a taught | 18:58 |
Jeffrey4l | in in default. | 18:58 |
inc0 | no, depends-on will work just fine | 18:58 |
Jeffrey4l | sean-k-mooney, i think it will slow down the merge speed. | 18:59 |
inc0 | sean-k-mooney, no, we need to know if it deploys before +2 | 18:59 |
*** dave-mcc_ has joined #openstack-kolla | 18:59 | |
sean-k-mooney | inc0: even if its non voteing? | 18:59 |
inc0 | sean-k-mooney, what's value of it if it's not voting and only runs after merge? | 18:59 |
Jeffrey4l | sean-k-mooney, we want all jobs voting, inc0 right? | 18:59 |
inc0 | "oh...we merged broken code" | 19:00 |
inc0 | Jeffrey4l, I don't think we'll have all jobs voting | 19:00 |
inc0 | but I want to have more of them | 19:00 |
sean-k-mooney | my point was it should be voteing if we only run it on merge | 19:00 |
Jeffrey4l | or at least all job green. | 19:00 |
inc0 | like...build binary won't be voting anytime soon | 19:00 |
inc0 | all jobs green - agree | 19:00 |
sean-k-mooney | i taught you were saying the deploy would be non voting for kolla | 19:00 |
inc0 | deploy will be non voting for kolla | 19:01 |
inc0 | but will be information for cores | 19:01 |
Jeffrey4l | sean-k-mooney, after two day, i got +w, but i found deploy gate red, because i made a mistake ( like wrong shell script syntax error ). then i start from beginning. | 19:02 |
*** dave-mccowan has quit IRC | 19:02 | |
*** prameswar has quit IRC | 19:02 | |
Jeffrey4l | that will be terrible. | 19:02 |
*** msimonin has joined #openstack-kolla | 19:03 | |
sean-k-mooney | thats valid. | 19:03 |
sean-k-mooney | is the intent for the deploy jobs to be voteing eventually? | 19:04 |
Jeffrey4l | sean-k-mooney, in kolla project, i do not think so. | 19:04 |
Jeffrey4l | if the deploy gate failed, it may cause dead lock . | 19:04 |
Jeffrey4l | kolla changes red, and kolla-ansible changed red, too. | 19:05 |
Jeffrey4l | so kolla only enable build image voting. and kolla-ansible/kolla-k8s only enable deploy job voting. | 19:05 |
sean-k-mooney | if +2 is depentend on deploy gate passing in kolla then it is effectivly voteing but i think we are rat holing | 19:05 |
Jeffrey4l | inc0, again please re-consider the support depends-on in kolla-ansible or kolla-k8s. i still it should be OK. even though the gate broken, we can push a fix in kolla, then the gate will be green. | 19:07 |
inc0 | Jeffrey4l, depends on will work on kolla-ansible -> kolla | 19:08 |
Jeffrey4l | without experimental? | 19:08 |
inc0 | I'd keep build gates experimental | 19:09 |
inc0 | because most of changes wont ever need that | 19:09 |
sean-k-mooney | experimental for kolla-ansible that is right inc0? | 19:09 |
Jeffrey4l | wait. add build gate in kolla-ansible? | 19:09 |
inc0 | and few that needs that will have voting gates red until kolla change merges | 19:09 |
inc0 | ok, again, problem is: we have change (changeX) in kolla-ansible that requres kolla change (changeY) | 19:10 |
inc0 | so state 1. changeX and changeY not merged | 19:10 |
inc0 | voting gates in changeX will be red because we have image missing or something | 19:11 |
openstackgerrit | Serguei Bezverkhi proposed openstack/kolla-kubernetes: WIP Hierarchical approach for helm charts https://review.openstack.org/401394 | 19:11 |
inc0 | but we can call check experimental and we'll call version of jobs that will also build images and pull changeY | 19:11 |
inc0 | so we'll have voting gates red and experimental gates green (or red, but then you know you need to fix stuff) | 19:12 |
inc0 | then, changeY merges in kolla | 19:12 |
inc0 | and all the gates in changeX becomes green, free to merge | 19:12 |
*** msimonin1 has joined #openstack-kolla | 19:12 | |
sean-k-mooney | yes that workflow should work fine | 19:13 |
sean-k-mooney | it will also work for k8s | 19:13 |
inc0 | yeah | 19:13 |
Jeffrey4l | inc0, all the issue is : we need run `recheck` after change Y is merged. | 19:13 |
inc0 | yeah | 19:13 |
inc0 | not a big deal imho | 19:13 |
inc0 | especially that our gates will have much more scenerios than experimental | 19:14 |
sean-k-mooney | well wont zull to a merge check after the depens-on change has merged | 19:14 |
inc0 | because we'll have more time in gates for valuable tests | 19:14 |
sean-k-mooney | could you trigger the recheck automatically? | 19:14 |
*** kristian__ has joined #openstack-kolla | 19:14 | |
inc0 | not sure, maybe | 19:14 |
*** jgriffith_away is now known as jgriffith | 19:14 | |
inc0 | either way, really not a big deal | 19:14 |
Jeffrey4l | and when using depends-on, may be all the job except experimental is a wast :( | 19:14 |
*** msimonin has quit IRC | 19:14 | |
Jeffrey4l | just fyi. | 19:15 |
sean-k-mooney | its only a wast if the depens-on is to kolla | 19:15 |
Jeffrey4l | i am Ok with the current implement this :) | 19:15 |
Jeffrey4l | kolla-ansible should only support kolla. ( it do not use others project's git url ) | 19:16 |
*** kristian__ has quit IRC | 19:16 | |
inc0 | Jeffrey4l, waste, yeah, not a bit waste tho | 19:16 |
*** kristian__ has joined #openstack-kolla | 19:16 | |
Jeffrey4l | anyway, let's implement this ;) | 19:16 |
Jeffrey4l | so conclusion: | 19:17 |
inc0 | cool, I'll keep talking to infra about getting registry up one way or another | 19:17 |
Jeffrey4l | in kolla project: 1. voting build source gate ( done ) | 19:17 |
Jeffrey4l | 2. push tarball after merged. | 19:17 |
inc0 | or 6 different tarballs | 19:17 |
Jeffrey4l | 3. in deploy gate, using build | 19:17 |
Jeffrey4l | inc0, yep. | 19:18 |
Jeffrey4l | that's all, anything else? | 19:18 |
inc0 | well, if 3 means adding deploy gates to kolla - non voting | 19:18 |
inc0 | than that's all | 19:18 |
Jeffrey4l | 4. deploy jobs will be non-voting | 19:19 |
inc0 | yeah | 19:19 |
Jeffrey4l | in kolla-ansible project: 1. pull image from tarball site | 19:19 |
Jeffrey4l | 2. add experimental jobs to use build image | 19:19 |
Jeffrey4l | 3. enable deploy+source gate | 19:20 |
Jeffrey4l | anything else? | 19:20 |
sean-k-mooney | Jeffrey4l: why pull from tarball site? | 19:20 |
Jeffrey4l | sean-k-mooney, that why we want set docker registry in infra | 19:20 |
sean-k-mooney | yes but you would pulll the images from the regestry | 19:21 |
sean-k-mooney | not tarballs of kolla | 19:21 |
inc0 | sean-k-mooney, if we run registry in infra | 19:21 |
inc0 | tarball == tarball of built images | 19:21 |
Jeffrey4l | sean-k-mooney, ^^ yes | 19:21 |
harlowja | qq, if i wanted to also get the test dependencies into images, whats the best way | 19:21 |
inc0 | sean-k-mooney, either way, this is only couple of solutions to "how to have registry in infra" | 19:22 |
harlowja | seems like the usual install only installs the non-test dependencies | 19:22 |
*** fragatina has quit IRC | 19:22 | |
inc0 | harlowja, what do you mean by test dependencies? | 19:22 |
harlowja | https://gist.github.com/harlowja/5fabeed44f957859a099ca33ecc79297 | 19:22 |
Jeffrey4l | pip install -r test-requirement.txt? | 19:22 |
harlowja | ya | 19:22 |
harlowja | Jeffrey4l i'm doing that | 19:22 |
sean-k-mooney | a tarbasll of all built images will take much more space the storing in a regestry as you loose the space saving of the deltas | 19:22 |
Jeffrey4l | sean-k-mooney, tarball means /var/lib/registry tarball file. | 19:23 |
harlowja | but Jeffrey4l since pip has a shitty dep resolver, u can't really do that by itself and expect it to be correct | 19:23 |
Jeffrey4l | it shouldn't that big. | 19:23 |
inc0 | it will be as big as registry itself...;) | 19:23 |
harlowja | https://github.com/openstack/kolla/blob/master/docker/glance/glance-base/Dockerfile.j2#L36 (seems like it just installs the project requirements, not say the test ones) | 19:23 |
Jeffrey4l | 2 - 3 GB may be? | 19:23 |
inc0 | maybe less even | 19:24 |
sean-k-mooney | ok so you are just taring up the regestry. you are not running a regestry and pulling the images indivgually over http | 19:24 |
inc0 | per distro | 19:24 |
inc0 | sean-k-mooney, correct | 19:24 |
Jeffrey4l | sean-k-mooney, yep. | 19:24 |
Jeffrey4l | harlowja, why u need that? | 19:24 |
sean-k-mooney | ok that was the disconnect | 19:24 |
Jeffrey4l | sean-k-mooney, sorry? | 19:25 |
sean-k-mooney | i was assuming you would just push to a regsitry as normal and pull as normal | 19:25 |
harlowja | Jeffrey4l to be able to test that the apps inside the images unittests still work post-build | 19:25 |
inc0 | sean-k-mooney, it was idea about "how to enable registry in infra without actually installing it there" | 19:25 |
*** lrensing has joined #openstack-kolla | 19:25 | |
Jeffrey4l | harlowja, you can add all you need in openstack-base image. because all the pip requirements exist in upper-constrain.txt file. | 19:26 |
harlowja | i don't want to have to install the whole upper-constraints file, lol | 19:26 |
Jeffrey4l | harlowja, run unittest in doker by using kolla? interesting. | 19:26 |
*** sayantani01 has quit IRC | 19:26 | |
harlowja | ya, https://gist.github.com/harlowja/971645fd17013c9fcd13f37aed3c0d34 (what our jenkins is doing) | 19:27 |
Jeffrey4l | harlowja, it will be hard to install in each service. because you may need install some develop packages. | 19:27 |
sean-k-mooney | inc0: i see. rsync might work better. basically build in registry into the base image nightly via node pool, rsync latest regstry as part of kolla-ansible job | 19:27 |
harlowja | Jeffrey4l i don't mind hard | 19:27 |
harlowja | lol | 19:27 |
inc0 | sean-k-mooney, so tarballs.o.o already have infra for it | 19:28 |
sean-k-mooney | inc0: on merge push delta with rsync to infra hosting | 19:28 |
inc0 | we just say "after merge I want this file uploaded" | 19:28 |
Jeffrey4l | re "after merge I want this file uploaded" we need this. otherwise , it may cause kolla-ansible failed. | 19:28 |
Jeffrey4l | for one day. | 19:28 |
sean-k-mooney | inc0: if it will prune all the old versions for you automatically then ya that will work too | 19:28 |
inc0 | it will literally override full registry | 19:29 |
harlowja | Jeffrey4l http://i.imgur.com/Miu2rSp.png what our jenkins is currently doing | 19:29 |
sean-k-mooney | Jeffrey4l: i was suggesting "after merge rsync delta" | 19:29 |
Jeffrey4l | named the file like: registry_centos_source_master.tar.gz | 19:29 |
sean-k-mooney | inc0: yes the only down side is every kolla-ansibel build will also have to download the full regestry | 19:30 |
Jeffrey4l | sean-k-mooney, does rsync possible right now? if so, it will be nice. | 19:30 |
sean-k-mooney | inc0: and every merged kolla change will have to upload it. | 19:30 |
sean-k-mooney | Jeffrey4l: im not sure. how are the docs uploaded | 19:30 |
inc0 | sean-k-mooney, yup, well, not ideal | 19:30 |
Jeffrey4l | iirc, it use scp. | 19:30 |
sean-k-mooney | scp would work for the registry too | 19:31 |
sean-k-mooney | rsycn can run over it and reduce the amount of bandwidth used | 19:31 |
Jeffrey4l | but it is not delta transport. | 19:31 |
Jeffrey4l | yep | 19:32 |
sean-k-mooney | basically with tarballs its a 2-3G upload everytimes a kolla change and a 2-3GB download for every time a kolla-ansible job runs | 19:32 |
*** schwicht has joined #openstack-kolla | 19:32 | |
*** msimonin1 has quit IRC | 19:33 | |
sean-k-mooney | with rsyc its a couple of megs hopfully since previous nights build | 19:33 |
inc0 | sean-k-mooney, but for rsync we need different hosting | 19:33 |
inc0 | not tarballs.o.o | 19:33 |
Jeffrey4l | we can specify the tarball's image. like we won't need bifrost right now. then we can not push it into tarball site. | 19:33 |
sean-k-mooney | yes perhaps. | 19:34 |
sean-k-mooney | well rsync can be an optimsiate at a later date | 19:34 |
inc0 | nah, I'd keep all of them | 19:34 |
*** eaguilar has joined #openstack-kolla | 19:34 | |
inc0 | simply because when someone actually will need bifrost | 19:35 |
inc0 | I don't want to change infra | 19:35 |
Jeffrey4l | fyi, during scp, it use this plugin https://ant.apache.org/manual/Tasks/scp.html | 19:35 |
Jeffrey4l | inc0, sean-k-mooney i added all conclusion in here https://etherpad.openstack.org/p/kolla-repo-split | 19:39 |
inc0 | thanks Jeffrey4l | 19:44 |
inc0 | I'll keep digging registry | 19:44 |
sean-k-mooney | inc0: what is the current issue with the registry by the way? | 19:45 |
Jeffrey4l | cool | 19:45 |
sean-k-mooney | you posted a review before? | 19:45 |
inc0 | sean-k-mooney, decision on direction we want to take with infea | 19:45 |
sean-k-mooney | ah ok | 19:45 |
inc0 | either they host normal registry - then we need to create puppet for installing it | 19:45 |
inc0 | or tarball hack | 19:45 |
inc0 | then we need to implement it ourselves | 19:45 |
sean-k-mooney | i see. | 19:46 |
sean-k-mooney | i would personally install docker and just spin up a regesty in docker. | 19:47 |
sean-k-mooney | aparently there is already a docker puppet module https://puppet.com/blog/simplify-managing-docker-puppet | 19:48 |
sean-k-mooney | https://github.com/garethr/garethr-docker | 19:49 |
jrich523 | hey guys, i realize this is mildy off topic, but im trying to hit the API for login and im going to the /v3/auth/token with a post based on a payload defined in the docs, and its telling me i need to auth first... i though thats what i was doing? | 19:49 |
*** dave-mcc_ has quit IRC | 19:49 | |
inc0 | jrich523, I'd say check if password is correct:) | 19:53 |
jrich523 | hmm well, it is, i copy pasted it :) | 19:53 |
jrich523 | seems like an odd error for that tho... but i guess based on the error msgs i've seen, thats normal for openstack lol | 19:54 |
jrich523 | "The request you have made requires authentication." | 19:54 |
inc0 | jrich523, that's how error "wrong password" looks like too | 19:54 |
jrich523 | hmm ok | 19:54 |
inc0 | clients will display it with any auth problem | 19:55 |
jrich523 | perhaps its the domain/scope, tried to leave it as default, i'll try being specific | 19:55 |
inc0 | check OS_AUTH_USERNAME and stuff | 19:55 |
inc0 | envs | 19:55 |
jrich523 | lol awesome | 19:55 |
jrich523 | yup im pulling that from the RC file | 19:55 |
inc0 | yeah it's not great error handling | 19:55 |
jrich523 | making calls from powershell, read in the RC file to grab what i need to auth | 19:55 |
inc0 | maybe powershell adds some strange windows'y whitespaces?;) | 19:56 |
*** fragatina has joined #openstack-kolla | 19:56 | |
jrich523 | lol perhaps, but probably not.. i've used powershell against a LOT of REST api's | 19:56 |
inc0 | I never used it, so I'm jsut shooting blind | 19:57 |
jrich523 | arent we all :) | 19:57 |
inc0 | hear hear | 19:58 |
jrich523 | boom! | 19:58 |
jrich523 | using the domain rather than default did it | 19:58 |
jrich523 | hmm, weird since the default domain is the domain im using, either way, thanks! | 19:59 |
inc0 | np, glad it works | 19:59 |
sean-k-mooney | inc0: im not a puppet guy but naively http://paste.openstack.org/show/590252/ should be all that is needed no? | 20:00 |
inc0 | well we want to add auth too | 20:00 |
inc0 | but yeah, it's not like hard either way | 20:00 |
inc0 | running registry outside docker is super easy too | 20:01 |
inc0 | apt-get install docker-registry -> docker-registry path/to/config/file | 20:01 |
inc0 | or service docker regsitry start | 20:01 |
sean-k-mooney | oh ok. when ever i have had to use puppt i just made it exec bash so im a bad person but it worth it to not learn ruby | 20:02 |
*** tonanhngo has quit IRC | 20:05 | |
inc0 | lol | 20:06 |
jrich523 | lol... it responded, but i dont see the token object in the response as the docs say | 20:07 |
jrich523 | nvm i think i know what they did | 20:07 |
*** sayantani01 has joined #openstack-kolla | 20:09 | |
*** msimonin has joined #openstack-kolla | 20:12 | |
*** inc0 has quit IRC | 20:16 | |
*** sayantani01 has quit IRC | 20:16 | |
*** sayantani01 has joined #openstack-kolla | 20:17 | |
*** eaguilar has quit IRC | 20:19 | |
*** msimonin has left #openstack-kolla | 20:25 | |
*** fragatina has quit IRC | 20:30 | |
openstackgerrit | Paul Belanger proposed openstack/kolla: DO NOT MERGE https://review.openstack.org/401412 | 20:38 |
*** pabelanger has joined #openstack-kolla | 20:38 | |
openstackgerrit | satya proposed openstack/kolla-ansible: Added separate interface for rabbitmq Enable rabbitmq to listen on different interface when primary hostname not based on api_interface. https://review.openstack.org/401341 | 20:39 |
*** v1k0d3n has quit IRC | 20:46 | |
*** newmember has joined #openstack-kolla | 20:47 | |
openstackgerrit | Paul Belanger proposed openstack/kolla: DO NOT MERGE https://review.openstack.org/401412 | 20:47 |
*** Satya_ has quit IRC | 20:49 | |
*** sayantani01 has quit IRC | 20:51 | |
*** flaper87 has quit IRC | 20:52 | |
*** sayantani01 has joined #openstack-kolla | 20:53 | |
*** sdake has joined #openstack-kolla | 20:54 | |
sdake | hey peeps | 20:54 |
sean-k-mooney | hey :) it pretty quite at the momen | 20:56 |
sean-k-mooney | moment | 20:56 |
sdake | which moment | 20:56 |
sdake | or quiet :) | 20:57 |
sean-k-mooney | my spelling should never be trusted | 20:57 |
pabelanger | for the gate-kolla-dsvm-build-ubuntu-binary-ubuntu-xenial-nv job, where would the images be stored on the filesystem it produces? /var/lib/docker? | 20:59 |
sean-k-mooney | yes | 21:00 |
sean-k-mooney | basically /var/lib/docker/image/btrfs | 21:01 |
*** fragatina has joined #openstack-kolla | 21:01 | |
sean-k-mooney | the last part is storage driver dependent | 21:02 |
*** lrensing has quit IRC | 21:04 | |
*** schwicht has quit IRC | 21:04 | |
pabelanger | thanks | 21:06 |
-openstackstatus- NOTICE: Due to a problem with the cinder volume backing the log server, jobs are failing with POST_FAILURE. Please avoid issuing 'recheck' commands until the issue is resolved. | 21:08 | |
*** ChanServ changes topic to "Due to a problem with the cinder volume backing the log server, jobs are failing with POST_FAILURE. Please avoid issuing 'recheck' commands until the issue is resolved." | 21:08 | |
openstackgerrit | Alexis Rivera proposed openstack/kolla-ansible: renaming-kolla-build-command https://review.openstack.org/401005 | 21:12 |
*** inc0 has joined #openstack-kolla | 21:13 | |
*** unicell has quit IRC | 21:14 | |
*** jtriley has quit IRC | 21:15 | |
*** unicell1 has joined #openstack-kolla | 21:21 | |
pabelanger | so, if I wanted to export an image, looks like: docker save lokolla/ubuntu-binary-base | gzip > lokolla_ubuntu-binary-base.tar.gz would be an example | 21:22 |
inc0 | pabelanger, we don't want to do that | 21:23 |
inc0 | reason being | 21:23 |
pabelanger | right, I am just testing docker commands | 21:23 |
inc0 | we have ubuntu binary and ubuntu source | 21:23 |
inc0 | ok | 21:23 |
pabelanger | looks like it sqashs things | 21:23 |
inc0 | it does | 21:23 |
inc0 | but on the other hand | 21:23 |
inc0 | if you look at docker registry storage volume | 21:24 |
inc0 | it understands layers | 21:24 |
*** fguillot has quit IRC | 21:24 | |
inc0 | so it will be lot smaller than sum of images | 21:24 |
inc0 | currently in our build gates we always download ubuntu base image | 21:24 |
pabelanger | /var/lib/docker is currently 7gb | 21:24 |
pabelanger | and looks to be a loopback mount | 21:25 |
inc0 | pabelanger, depends on storage driver | 21:25 |
inc0 | docker info <- this command should tell you which storage driver you use | 21:25 |
pabelanger | btrfs | 21:25 |
inc0 | ok, that's cool | 21:25 |
inc0 | ubuntu xenial I presume | 21:26 |
pabelanger | this is a gate-kolla-dsvm-build-ubuntu-binary-ubuntu-xenial-nv vm btw | 21:26 |
pabelanger | I held a node to poke at it | 21:26 |
*** schwicht has joined #openstack-kolla | 21:26 | |
inc0 | ok | 21:26 |
inc0 | so if we use full registry | 21:27 |
pabelanger | so, step 1, could be unmount /var/lib/docker, compress and store on tarballs.o.o. To see how much time that takes | 21:27 |
inc0 | I wouldn't do that | 21:27 |
pabelanger | if it is a really long time, then I say we move back to docker-register in infra | 21:27 |
inc0 | I'd rather run local registry | 21:27 |
inc0 | push images there | 21:27 |
inc0 | and tarball it's storage | 21:27 |
pabelanger | okay | 21:27 |
inc0 | it will be less error-prone | 21:28 |
pabelanger | so, after you build, import into local registry then export | 21:28 |
pabelanger | then upload that data | 21:28 |
pabelanger | ? | 21:28 |
inc0 | yeah | 21:29 |
pabelanger | okay | 21:30 |
inc0 | I'll try to build ubuntu source, upload it to fresh registry | 21:30 |
inc0 | and will check how much tarball will take | 21:30 |
*** schwicht has quit IRC | 21:31 | |
inc0 | pabelanger, we don't have swift available anywhere right? | 21:32 |
pabelanger | inc0: if you dump them into the images folder, I can setup the publisher macro today | 21:33 |
pabelanger | inc0: no | 21:33 |
inc0 | I'll put them in some better directory - I can work that one out | 21:33 |
pabelanger | inc0: so, $WORKSPACE/images or name of choice | 21:34 |
inc0 | good thing about that all - we already run docker in gates, so running registry is literally single command | 21:34 |
pabelanger | but needs to be under $WORKSPACE for publisher to work | 21:34 |
inc0 | $WORKSPACE/images it is | 21:34 |
pabelanger | I'll setup macro now | 21:35 |
*** Jeffrey4l has quit IRC | 21:35 | |
inc0 | so if that's how we do it | 21:35 |
inc0 | we'll have 6 different tars there | 21:35 |
inc0 | binary+source * ubuntu+oraclelinux+centos | 21:35 |
*** Jeffrey4l has joined #openstack-kolla | 21:35 | |
inc0 | I'll make sure that tar.gz will be named correctly | 21:36 |
inc0 | each gate will upload different flavor | 21:36 |
inc0 | and each deploy gate will download correct flavor | 21:36 |
inc0 | to minimize amount of data | 21:36 |
*** f13o has joined #openstack-kolla | 21:36 | |
pabelanger | Yes, the publisher will match *.tar.gz, storing them in tarballs.o.o/(openstack-)kolla/images | 21:37 |
pabelanger | can't check what format you use today, since our drive is offline | 21:37 |
inc0 | cool | 21:37 |
*** schwicht has joined #openstack-kolla | 21:37 | |
inc0 | tar.gz works | 21:37 |
*** mgiles has quit IRC | 21:39 | |
*** schwicht has quit IRC | 21:39 | |
*** schwicht has joined #openstack-kolla | 21:40 | |
inc0 | pabelanger, ok can't build source because tarballs are down;) | 21:41 |
*** MarcisK has joined #openstack-kolla | 21:47 | |
*** srwilkers_ has quit IRC | 21:51 | |
MarcisK | Hi, guys! Couple of months ago I asked here if there are any plans on implementing SELinux for nova-compute. Last time there were no plans on doing that. Are there any updates regarding SELinux? | 21:51 |
*** jtriley has joined #openstack-kolla | 21:52 | |
*** schwicht has quit IRC | 21:55 | |
inc0 | MarcisK, it's not about lack of plans | 21:56 |
inc0 | it's about lack of people to do it:) | 21:56 |
inc0 | we want it, just nobody had written this yet | 21:56 |
sdake | kolla runs on selinux does it not? | 21:56 |
inc0 | well, it requires to not be enforcing | 21:57 |
sdake | the gate enables selinux | 21:57 |
inc0 | afair | 21:57 |
sdake | pabelanger could verify - my understanding is centos enables selinux by default | 21:57 |
sdake | and our gate scripts do not disable it | 21:57 |
MarcisK | The biggest issue is with nova-compute. Normally libvirt would put every instance into it's own "selinux domain", isolating it from another instances and protecting the host from virtualization security bugs | 21:58 |
inc0 | sdake, either you're wrong or this doc is https://github.com/openstack/kolla/blob/d75847772d0e880329eef5867caf4b87035d2ae0/doc/security.rst#selinux | 21:58 |
MarcisK | But as far as I've seen - in kolla all libvirt guests are running into a same selinux domain | 21:58 |
*** jtriley has quit IRC | 21:59 | |
sdake | inc0 doc is wrong pretty sure | 21:59 |
sdake | let me grep repo for selinux moment | 21:59 |
inc0 | I'm not sure, I'm ubuntu guy | 21:59 |
sean-k-mooney | im not sure selinux in its default config on centos will work. | 22:00 |
sean-k-mooney | i know that at least on ubuntu you sometimes have apparmor issues with libvirt | 22:01 |
sdake | ya kolla-ansible does not disable selinux | 22:01 |
inc0 | sdake, ofc it doesn't | 22:02 |
sdake | those docs re selinux are old | 22:02 |
inc0 | but it may require | 22:02 |
sdake | i mena the gate scripts there of | 22:02 |
inc0 | it to be not enforcing | 22:02 |
inc0 | kk | 22:02 |
sdake | and i'm certain infra enabled selinux on all images | 22:02 |
sean-k-mooney | sdake: selinux may be disabled in the image building | 22:02 |
sean-k-mooney | sdake: after all devstack requires selinux to be off | 22:02 |
*** Pavo has quit IRC | 22:02 | |
sdake | that is because devstack touches a whole lotta files | 22:03 |
*** Pavo has joined #openstack-kolla | 22:03 | |
sdake | and doesnt do it through proper channels | 22:03 |
sdake | so MarcisK the short answer i think from our discussion is kolla works on selinux | 22:03 |
sean-k-mooney | does kolla have custom images or does it use the default infra images | 22:03 |
MarcisK | RDO worked just fine with SELinux. Both controller and compute. | 22:03 |
MarcisK | I'm now deploying kolla on a lab machine. Will provide you with additional details soon. | 22:04 |
sdake | sean-k-mooney we use default infra images, there is no way I know of to make our own "custom images" | 22:04 |
inc0 | sdake, btw read through log here | 22:04 |
sdake | as far as libvirt putting every instance into its own security domain | 22:04 |
sdake | no clue on that | 22:04 |
inc0 | around 3hrs ago | 22:05 |
inc0 | me, Jeffrey4l, sean-k-mooney and pabelanger devised a way to enable registry-like behaviour | 22:05 |
inc0 | in infra | 22:05 |
sdake | nice | 22:05 |
sdake | i'll take your word for it :) | 22:05 |
sdake | not super interested in working on gating personally | 22:06 |
*** tonanhngo has joined #openstack-kolla | 22:06 | |
*** tonanhngo has quit IRC | 22:08 | |
*** BIOS_Hazard has quit IRC | 22:10 | |
sean-k-mooney | does kolla support osp currently? | 22:10 |
*** v1k0d3n has joined #openstack-kolla | 22:10 | |
sean-k-mooney | as in RHEL osp as a build target | 22:11 |
MarcisK | Btw, guys, while my kolla setup is getting ready. Are there any known large production deployments of kolla? | 22:13 |
*** tonanhngo has joined #openstack-kolla | 22:13 | |
britthouser | I believe its untested sean-k-mooney. The package names should be the same as CentOS. And the RHEL container is *supposed* to inherit the entitlements from the host. But we’ve had issues with that in the past. | 22:13 |
britthouser | So it seems like it would be straightforward, but I’m betting there are few devils in the details. | 22:14 |
*** tonanhngo has quit IRC | 22:14 | |
inc0 | MarcisK, how much is large?:) | 22:14 |
sean-k-mooney | britthouser: the reason i ask is for ovs-dpdk. its part of osp but not in rdo on centos | 22:14 |
MarcisK | inc0: good question :) for example, at least 30 compute nodes? | 22:15 |
inc0 | hah, I was thinking 200+, yeah, we have several ~100 | 22:15 |
inc0 | that I know of | 22:15 |
inc0 | but I expect that there is much more | 22:15 |
inc0 | Pavo, how's your "small" PoC going?;) | 22:16 |
inc0 | if you think of images | 22:16 |
inc0 | not ansible part | 22:16 |
inc0 | then there is one rather big I know | 22:16 |
inc0 | 14000 servers | 22:16 |
inc0 | or something like that | 22:16 |
*** harbor has joined #openstack-kolla | 22:17 | |
sean-k-mooney | inc0: can you remeber if vlan tenant networks were tested in osic or in general? | 22:17 |
MarcisK | inc0: thanks :) | 22:18 |
*** slagle has quit IRC | 22:18 | |
inc0 | sean-k-mooney, I think we didnt test vlan | 22:18 |
inc0 | in general yeah | 22:18 |
inc0 | but you need to play around nutron ini | 22:19 |
sean-k-mooney | we dont actully document how to enable vlan tenant networks | 22:19 |
sean-k-mooney | ya i am doing that now | 22:19 |
inc0 | no I dont think so | 22:19 |
inc0 | we should | 22:19 |
*** harbor is now known as portdirect__ | 22:19 | |
sean-k-mooney | im also going to enable the conntrack security group dirver that way but should i make that an option in ansible? | 22:20 |
pabelanger | sdake: yes, centos-7 has selinux enforced | 22:20 |
pabelanger | http://nodepool.openstack.org/dib.centos-7.log is the log of our builds | 22:20 |
inc0 | sean-k-mooney, I don't see why not | 22:20 |
sean-k-mooney | inc0: well is more how much shold be done via config override and just document and how much should be done as ansible/template change and variable in global.yml | 22:21 |
inc0 | pabelanger, registry for ubuntu binary takes 2.7Gig | 22:23 |
inc0 | after tar | 22:23 |
inc0 | source should be little less | 22:23 |
sean-k-mooney | how fast are the network connection on the vms in the gate? | 22:24 |
sean-k-mooney | 1G? | 22:24 |
pabelanger | first step is to get it onto tarballs.o.o | 22:24 |
pabelanger | then we can import it into afs | 22:24 |
inc0 | pabelanger, tarballs are local to regions or hosted somewhere? | 22:24 |
pabelanger | each region then has a cache | 22:25 |
portdirect__ | sbezverk: ping | 22:25 |
inc0 | cool, upload won't be that common | 22:25 |
pabelanger | tarballs.o.o is in rackspace | 22:25 |
pabelanger | but each cloud region has a mirror | 22:25 |
inc0 | ok, so most of the time we'll have nice 10Gig | 22:25 |
pabelanger | moving them to the mirror will address bandwidth issues | 22:25 |
inc0 | pabelanger, any way for gate to know if it's merge-gate or just normal check? | 22:26 |
inc0 | we wouldn't tar-gz these things every time | 22:26 |
inc0 | it takes couple minutes to tar.gz whole repo | 22:27 |
*** f13o has quit IRC | 22:27 | |
pabelanger | I am building a post pipeline job to do that | 22:27 |
pabelanger | so, after code is merged into git, we build and upload them | 22:28 |
inc0 | cool | 22:28 |
pabelanger | should be uploaded shortly | 22:28 |
inc0 | let me know where you want code to be | 22:28 |
pabelanger | will uses experimental first, so we can iterate on testing | 22:28 |
openstackgerrit | BorneMace proposed openstack/kolla-ansible: Ignore dom0 qemu processes during destroy https://review.openstack.org/399203 | 22:28 |
pabelanger | then, once good. move it into post | 22:28 |
inc0 | cool | 22:28 |
inc0 | thanks a lot | 22:28 |
pabelanger | np, once we get this working and your gates voting, I can circle back to a per region docker-registry | 22:29 |
pabelanger | just don't have time right now to focus on it | 22:29 |
*** kristian__ has quit IRC | 22:29 | |
sean-k-mooney | just out of interest. | 22:30 |
sean-k-mooney | since we will basically be uploading all container when the merge to infra | 22:31 |
inc0 | cool, thanks | 22:31 |
sean-k-mooney | can we also auto push to docker hub? | 22:31 |
inc0 | sean-k-mooney, well, that will be harder | 22:31 |
inc0 | it takes good amount of time for upload | 22:31 |
inc0 | and network traffic | 22:31 |
sdake | pabelanger thanks for the confirmation | 22:31 |
sean-k-mooney | ok i guess we could push that to them. | 22:32 |
sean-k-mooney | we could set up a git webhook that would have them rebuild the continers on each commit | 22:32 |
sean-k-mooney | its not needed but it would be neat | 22:32 |
inc0 | or do nightly somewhere | 22:32 |
pabelanger | it would be easier to do, yes. The next step would be to download said tarballs in a secure vm (which has username / password) for docker hub, then do the upload from it. | 22:33 |
*** matrohon has quit IRC | 22:33 | |
pabelanger | that's the process we do for publishing to pypi for python packages | 22:33 |
sean-k-mooney | well you could upload the built images but dockerhub support building the continers for you as far a i know | 22:34 |
sean-k-mooney | we would just have to have a webhook to trigger the build the issue would be creeating the docker file templates automatically | 22:35 |
sean-k-mooney | either would work though | 22:35 |
inc0 | someday;) | 22:35 |
inc0 | building with dockerhub won't work because afair it doesn't support COPY | 22:36 |
sean-k-mooney | ah ok we basically would have to run kolla-build --template-only also before sending it to them | 22:37 |
*** pomac has quit IRC | 22:38 | |
sean-k-mooney | honestly i think a nightly build would be more the sufficent form most usecases | 22:39 |
inc0 | sean-k-mooney, even with --template-only, still COPY won't work, so source build is a no-no | 22:40 |
inc0 | but pushing nightly is doable | 22:40 |
sean-k-mooney | it would be nice for the stable branches once upgrade to same branch is supported. | 22:41 |
inc0 | it is supported | 22:42 |
sean-k-mooney | not that you upgrade nightly but if you are not building your own images it would be nice to be able to pull a regurally pattached/build image | 22:42 |
*** huikang has joined #openstack-kolla | 22:42 | |
inc0 | minor upgrades (3.0.0->3.0.1) are supported ofc | 22:42 |
sean-k-mooney | i can upgrade form old stable/newton to latest stable/newton? | 22:42 |
inc0 | ofc | 22:42 |
sdake | inc0 the consensus view is kolla-ansible runs on selinux - since it runs that way in the gate | 22:43 |
inc0 | sdake, cool | 22:43 |
inc0 | then docs needs update | 22:43 |
sdake | whether or not it meets MarcisK 's requirements is another question | 22:43 |
sdake | ok i'll take that particlar part on - which plenty of conditionals :) | 22:44 |
sdake | which/with | 22:44 |
sdake | (that part = fixing the docs) | 22:44 |
sdake | scoped to selinux only | 22:45 |
sdake | clear? | 22:45 |
inc0 | sure, whatever you can do:) | 22:50 |
inc0 | I'm going off guys, happy turkey day | 22:50 |
inc0 | I was meant to say happy turkey, but I bet they're not exactly happiest bunch around now | 22:50 |
inc0 | bye! | 22:51 |
*** inc0 has quit IRC | 22:51 | |
sean-k-mooney | is today thanksgiving in the us | 22:51 |
sean-k-mooney | i taught that was next week? | 22:51 |
sean-k-mooney | not from us so dont understand this holiday other then like christmas but not celebrating meal between native americans and european settelers | 22:52 |
*** ChanServ changes topic to "Looking for Ansible deployment? git checkout http://github.com/openstack/kolla-ansible; New to Kolla? Please read the documentation here: http://docs.openstack.org/developer/kolla/; Kolla IRC meetngs on Wednesdays @ 16:00 UTC - see agenda @ https://goo.gl/OXB0DL - IRC channel is *LOGGED* @ http://goo.gl/3mzZ7b (old logs from #kolla http://goo.gl/VKpPzA):" | 22:52 | |
-openstackstatus- NOTICE: The affected filesystems on the log server are repaired. Please leave 'recheck' comments on any changes which failed with POST_FAILURE. | 22:52 | |
*** sayantani01 has quit IRC | 22:53 | |
*** MarcisK has quit IRC | 22:59 | |
pron | hmm is it posible that running kolla inside openstack fails cause of networking issues? | 23:02 |
sean-k-mooney | pron: it depend on what you are trying to do | 23:06 |
pron | i created 10-15 instances | 23:06 |
pron | and tryng ot play with kolla | 23:07 |
sean-k-mooney | kolla on openstack over a vlan tenatn networking with vxlan for kolla should work fine | 23:07 |
pron | but it stuck atm with not being able to to konnect to network alias | 23:07 |
sean-k-mooney | network alias? | 23:07 |
pron | emm | 23:07 |
pron | that another ip that keepalived puts on interface | 23:07 |
pron | ;P | 23:07 |
pron | and then haproxy uses ir | 23:07 |
pron | it* | 23:08 |
sean-k-mooney | ah you have to add that other ip to the allowed adress pairs on the neutron port | 23:08 |
pron | ah m can i do that with gui? | 23:09 |
sean-k-mooney | the admin gui definetly not sure about the tenent verion | 23:09 |
pron | okay then you propably can point me to the cli line :P | 23:10 |
* pron smiles and looks sean-k-mooney in the eye | 23:10 | |
sean-k-mooney | let me check ... look for cli docs | 23:11 |
*** bmace has quit IRC | 23:12 | |
pron | i am not yet familiar with them | 23:12 |
*** bmace has joined #openstack-kolla | 23:12 | |
sean-k-mooney | neutron port-update --allowed-address-pair ip_address=<IP_ADDR> <port uuid> | 23:13 |
pron | that rises more questions that it solves but thx :P | 23:13 |
pron | i somehow tought its atm one universal cli for everything | 23:14 |
sean-k-mooney | yes there is the unified openstack client | 23:14 |
sean-k-mooney | aparently port updates are not supproted yet | 23:14 |
pron | ah i seee | 23:14 |
sean-k-mooney | hench i gave the neutorn client example | 23:15 |
pron | i see | 23:15 |
* pron scratches scratchable parts | 23:15 | |
*** pomac has joined #openstack-kolla | 23:15 | |
pron | thx =) | 23:15 |
sean-k-mooney | so in our case the neutron port you use for you haproxy interface should be updated with the vip address | 23:16 |
*** liyifeng has joined #openstack-kolla | 23:16 | |
pron | aye | 23:16 |
pron | i suspected that! | 23:17 |
sean-k-mooney | i know this at least work for all in one deployment of kolla on openstack. have not tried multinode yet | 23:18 |
*** tonanhngo has joined #openstack-kolla | 23:18 | |
*** srwilker has joined #openstack-kolla | 23:20 | |
*** tonanhngo has quit IRC | 23:20 | |
pron | for all n one i didnt bother i just used floating ip | 23:25 |
pron | ;D | 23:25 |
pron | sean-k-mooney: looks like it did the trick | 23:29 |
*** dkehn__ has quit IRC | 23:29 | |
sean-k-mooney | cool are you deploying multiple controlers? | 23:30 |
pron | yes | 23:30 |
pron | i am not sure what to use for hardware provisioning | 23:30 |
pron | and insalling | 23:31 |
pron | checked out cobller but it looks liek a huge mess | 23:31 |
sean-k-mooney | well there some support for bifrost that i added last cycle but the code is pretty new. others have used maas | 23:31 |
pron | i do use some silly stuff for debian deployment alrdy | 23:32 |
pron | i have no clue what is bitforst tho ;x | 23:32 |
sean-k-mooney | pron: https://github.com/openstack/kolla/blob/master/doc/bifrost.rst its standalone ironic | 23:33 |
sean-k-mooney | i use it at home but we have not tested it at scale. | 23:34 |
*** msimonin1 has joined #openstack-kolla | 23:35 | |
*** msimonin1 has quit IRC | 23:35 | |
pron | so you end with soem registry of phisical hardware on the host you run it? or the hardware ends in your openstack ? | 23:36 |
pron | ( i am pretty new there :P ) | 23:36 |
*** sdake has quit IRC | 23:37 | |
sean-k-mooney | you provide a yaml inventor of the node you want to deploy and os on and then run kolla-ansible deploy-servers wich enroles them in the standalone instance of ironic and ironic cleans them down and install the os | 23:38 |
pron | ahaa | 23:39 |
pron | i will look into it thx | 23:39 |
*** tonanhngo has joined #openstack-kolla | 23:40 | |
*** sdake has joined #openstack-kolla | 23:40 | |
*** sdake has quit IRC | 23:40 | |
*** sdake has joined #openstack-kolla | 23:40 | |
*** tonanhngo has quit IRC | 23:40 | |
*** huikang has quit IRC | 23:40 | |
*** huikang has joined #openstack-kolla | 23:41 | |
pron | sean-k-mooney: you actualy helped a lot today and saved so much time thx | 23:42 |
sean-k-mooney | no worries | 23:42 |
sean-k-mooney | glad to help | 23:42 |
pron | another thign i am wondering how do i tell kolla i want use tagged vlans for things | 23:43 |
*** lamt has quit IRC | 23:43 | |
sean-k-mooney | for example your storage network or api network? | 23:43 |
sean-k-mooney | or for tenant netowrks? | 23:44 |
pron | yeh | 23:44 |
pron | no no storage/mgmt/etc | 23:44 |
sean-k-mooney | well for api,storage etc. current that has to be done before you deploy kolla. then you tell kolla to use the vlan subport | 23:44 |
pron | i see | 23:45 |
sean-k-mooney | so instead of storage_interface: "enp8s0f0" you would do storage_interface: "enp8s0f0.1" or whatever | 23:45 |
pron | i point kolla to vlan interface | 23:45 |
sean-k-mooney | yep | 23:45 |
pron | sean-k-mooney: that bitfrost thing can do that for me? | 23:45 |
*** huikang has quit IRC | 23:46 | |
sean-k-mooney | this is somthing i would like to add to the kolla-ansible bootstrap-servers command in the future but its not there now. | 23:46 |
sean-k-mooney | pron: maybe. not directly but you could bake it into the image. | 23:46 |
pron | wich means i have to do my own ansible work for that atm | 23:46 |
sean-k-mooney | ya basically | 23:46 |
sean-k-mooney | ubuntu maas and cobler i think can do it but im not that familar with them | 23:47 |
pron | hmm image? | 23:47 |
pron | it doesnt use some kind of preseed/kickstart? | 23:47 |
sean-k-mooney | the os image you deploy to the server. | 23:47 |
pron | deploy is abstraction :P can be million ways :P | 23:48 |
sean-k-mooney | well bifrost use ironic so you can use cloud-init to configure the server which is installed via pxe boot and ipmi by default | 23:48 |
sean-k-mooney | ironic has a bunch of backend drivers | 23:49 |
pron | guess it sounds like a homework for me =) unless someone wants to give me free lecture | 23:49 |
pron | so i have to read ironic things and then move to bitfrost | 23:51 |
sean-k-mooney | that is one option yes but as you are probably aware there are many ways to do baremetal deployment so the final choice will depend on what you are looking for | 23:52 |
pron | i am looking for something more elegant then http://hands.com/d-i/ | 23:56 |
pron | ;=) | 23:56 |
*** newmember has quit IRC | 23:57 | |
pron | but this is fallback for me | 23:57 |
*** tonanhngo has joined #openstack-kolla | 23:58 | |
*** goldyfruit has quit IRC | 23:59 | |
*** tonanhngo has quit IRC | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!