Thursday, 2018-06-28

« Wednesday, 2018-06-27 Index Friday, 2018-06-29 »
openstackgerritMichael Johnson proposed openstack/octavia master: DNM: Re-enable KVM on OVH hosts  https://review.openstack.org/54905400:26
openstackgerritMichael Johnson proposed openstack/octavia master: Neutron-LBaaS to Octavia migration tool  https://review.openstack.org/55442000:28
*** Guest36449 has quit IRC00:28
*** longkb has joined #openstack-lbaas00:44
*** kobis has joined #openstack-lbaas01:42
*** Johnny has joined #openstack-lbaas01:44
*** Johnny is now known as Guest2828901:44
Guest28289hi, anyone has a complete guide on how to start with octavia installation?01:45
Guest28289I have actually a production ready setup with keystone,glance,nova and neutron (not running lbaas yet) on a controller node01:46
*** kobis has quit IRC01:47
*** Guest28289 is now known as johnnytan288501:48
*** hongbin has joined #openstack-lbaas02:06
*** AlexeyAbashkin has joined #openstack-lbaas02:10
*** links has joined #openstack-lbaas02:30
*** links has quit IRC02:30
*** AlexeyAbashkin has quit IRC02:39
*** kobis has joined #openstack-lbaas03:43
lxkongjohnsom, i think this patch https://review.openstack.org/#/c/573497/ breaks heat03:46
*** kobis has quit IRC03:47
lxkongwe verify the http_method here https://github.com/openstack/octavia/blob/3d545ab1b6e133f4696fdd389472b869e86d177b/octavia/api/v2/controllers/health_monitor.py#L11703:47
lxkongbut in heat template http_method must have a value: https://github.com/openstack/heat/blob/3ec69a96f6872463cf446fc242af83bcab1abe70/heat/engine/resources/openstack/octavia/health_monitor.py#L7903:48
lxkongalso EXPECTED_CODES and URL_PATH are affected04:00
*** nmanos has joined #openstack-lbaas04:19
*** hongbin has quit IRC04:27
*** AlexeyAbashkin has joined #openstack-lbaas04:32
*** ramishra has joined #openstack-lbaas04:34
openstackgerritAllen proposed openstack/octavia-dashboard master: Show the 'Insert Headers' when listener protocol is 'HTTP' or 'HTTPS'  https://review.openstack.org/56496304:43
*** AlexeyAbashkin has quit IRC05:25
*** openstackgerrit has quit IRC05:34
*** yboaron has joined #openstack-lbaas05:49
*** openstackstatus has quit IRC05:51
*** openstackstatus has joined #openstack-lbaas05:52
*** ChanServ sets mode: +v openstackstatus05:52
*** ramishra has quit IRC06:10
*** ramishra has joined #openstack-lbaas06:12
*** openstackgerrit has joined #openstack-lbaas06:15
openstackgerritAllen proposed openstack/octavia-dashboard master: Show the 'Insert Headers' when listener protocol is 'HTTP' or 'TERMINATED_HTTPS'  https://review.openstack.org/56496306:15
*** kobis has joined #openstack-lbaas06:23
*** kobis has quit IRC06:28
*** AlexeyAbashkin has joined #openstack-lbaas06:29
*** nmanos has quit IRC06:33
*** nmanos has joined #openstack-lbaas06:33
*** ramishra has quit IRC06:35
*** ramishra has joined #openstack-lbaas06:37
*** keithmnemonic[m] has quit IRC06:40
*** AlexeyAbashkin has quit IRC06:41
*** keithmnemonic[m] has joined #openstack-lbaas06:41
*** ramishra has quit IRC06:55
*** yboaron has quit IRC06:58
*** ramishra has joined #openstack-lbaas07:00
*** peereb has joined #openstack-lbaas07:09
*** pcaruana has joined #openstack-lbaas07:10
*** tesseract has joined #openstack-lbaas07:11
*** ramishra has quit IRC07:23
*** ramishra has joined #openstack-lbaas07:24
*** ramishra has quit IRC07:29
*** ramishra has joined #openstack-lbaas07:30
bzhao__johnsom:  hi, have you done the udp test? :). For current lvs-nat, we need to open the ip_forward in lvs director.  Then we can access the VIP accross a neutron router. But I found the default amphora-haproxy didn't open it.07:34
*** kobis has joined #openstack-lbaas07:36
*** ramishra has quit IRC07:39
*** ispp has joined #openstack-lbaas07:54
*** numans has joined #openstack-lbaas07:55
*** numans_ has quit IRC07:56
*** yboaron has joined #openstack-lbaas07:59
*** rpittau has quit IRC08:03
*** rpittau has joined #openstack-lbaas08:03
*** JudeC_ has quit IRC08:05
*** threestrands has joined #openstack-lbaas08:06
*** threestrands has quit IRC08:13
*** ramishra has joined #openstack-lbaas08:25
*** apuimedo has joined #openstack-lbaas08:29
*** numans_ has joined #openstack-lbaas08:30
*** numans has quit IRC08:30
isppJohnson, I was able to resolve the issue, in my case I didn't have any data I would care about, but in case I would.. this would be really bad, I am investigating why this happened,08:40
*** yboaron_ has joined #openstack-lbaas09:01
*** yboaron has quit IRC09:04
*** yboaron_ has quit IRC09:05
*** yboaron_ has joined #openstack-lbaas09:05
*** salmankhan has joined #openstack-lbaas09:17
*** rcernin has quit IRC09:56
*** johnnytan2885 has quit IRC10:00
*** yboaron_ has quit IRC10:02
*** numans_ has quit IRC10:06
*** yboaron_ has joined #openstack-lbaas10:09
*** kobis has quit IRC10:23
*** ispp has quit IRC10:25
*** kobis has joined #openstack-lbaas10:40
*** numans has joined #openstack-lbaas10:41
*** peereb has quit IRC10:43
*** ramishra has quit IRC10:44
*** salmankhan has quit IRC10:45
*** ThomasWhite has joined #openstack-lbaas10:48
*** ispp has joined #openstack-lbaas10:49
ThomasWhiteHi all, in order for the source IP to be shown to clients/apps/instances received load balanced connections, do any changes need to be made to the neutron conf? We are running Newton (soon to be upgraded a few times) with Neutron LBaaSv210:49
openstackgerritCarlos Goncalves proposed openstack/octavia master: Make scenario gates use devstack-tempest parent  https://review.openstack.org/54103911:00
ThomasWhiteOk, just realised I was having one of those moments with that question. So HTTP does passthrough with the header, but can it be done with TCP? I know HAProxy can do it with send-proxy parameter, but can I achieve this in Openstack on lbaasv2?11:23
*** numans_ has joined #openstack-lbaas11:27
openstackgerritZhaoBo proposed openstack/octavia master: UDP jinja template  https://review.openstack.org/52542011:28
openstackgerritZhaoBo proposed openstack/octavia master: UDP for [2]  https://review.openstack.org/52965111:28
openstackgerritZhaoBo proposed openstack/octavia master: UDP for [3][5][6]  https://review.openstack.org/53939111:28
*** sapd_ has joined #openstack-lbaas11:33
*** sapd has quit IRC11:33
*** numans_ has quit IRC11:36
*** ramishra has joined #openstack-lbaas11:50
*** ramishra has quit IRC11:51
*** ramishra has joined #openstack-lbaas11:51
*** pcaruana has quit IRC11:56
*** amuller has joined #openstack-lbaas11:56
*** fnaval has quit IRC12:22
waytHello folks, I've discovered you've a blueprint about l3 active active implementation in Octavia, and also some merge request that was never merged. Is the subject still up to date ?12:27
*** jlaffaye_ is now known as jlaffaye12:28
*** longkb has quit IRC12:30
*** fnaval has joined #openstack-lbaas12:34
johnsomThomasWhite: Only native Octavia Pike or newer supports the “proxy” protocol for TCP.  You can run a newer version of Octavia on a newton cloud, so grab the Queens version of Octavia and you will have what you need.  The deprecated neutron-lbaas does not support it.12:41
johnsomwayt: yes, it is still being worked on. However the developers involved shifted to another internal project for a bit. It will likely be the Stein release before it is12:43
johnsomdone12:43
*** pcaruana has joined #openstack-lbaas12:44
waytjohnsom: alright, thanks for the info12:46
*** rcernin has joined #openstack-lbaas12:52
*** rcernin has quit IRC13:08
*** numans_ has joined #openstack-lbaas13:27
*** numans_ has quit IRC13:32
*** kobis has quit IRC13:46
*** yboaron has joined #openstack-lbaas14:00
*** yboaron_ has quit IRC14:03
*** kobis has joined #openstack-lbaas14:09
*** kobis has quit IRC14:11
*** kobis has joined #openstack-lbaas14:13
*** ispp has quit IRC14:14
*** yboaron has quit IRC14:24
*** kobis has quit IRC14:30
*** Swami has joined #openstack-lbaas14:35
ThomasWhitejohnsom: Many thanks, sounds like a plan. Will let you know how it goes when we figure it out.14:36
*** Swami_ has joined #openstack-lbaas14:37
*** fnaval has quit IRC14:38
*** Swami has quit IRC14:39
*** kobis has joined #openstack-lbaas14:42
*** fnaval has joined #openstack-lbaas14:42
*** numans_ has joined #openstack-lbaas14:45
johnsomSome good news for Thursday, it looks like KVM may be working at OVH again.  We had a number of good runs last night and in working with them we can't reproduce it anymore.14:49
johnsomI'm going to do more re-checks just to make sure before we re-enable KVM for OVH.  Still not sure on limestone, we haven't hit one of their hosts with our test yet.14:50
*** Swami_ has quit IRC14:51
*** johnnytan2885 has joined #openstack-lbaas14:54
openstackgerritMichael Johnson proposed openstack/octavia master: Fix a requirements mis-match  https://review.openstack.org/57883715:00
*** kobis has quit IRC15:03
*** Swami has joined #openstack-lbaas15:03
*** yamamoto has quit IRC15:12
*** yamamoto has joined #openstack-lbaas15:12
*** lxkong has quit IRC15:39
*** ispp has joined #openstack-lbaas15:40
*** numans has quit IRC15:52
*** ispp has quit IRC16:20
*** raopajay has quit IRC16:26
*** ramishra has quit IRC16:34
*** kobis has joined #openstack-lbaas16:34
*** tesseract has quit IRC16:38
*** kobis has quit IRC16:43
*** colin- has joined #openstack-lbaas16:45
*** Swami has quit IRC16:53
*** ramishra has joined #openstack-lbaas16:59
*** kobis has joined #openstack-lbaas17:15
openstackgerritMichael Johnson proposed openstack/octavia master: Neutron-LBaaS to Octavia migration tool  https://review.openstack.org/55442017:24
*** SumitNaiksatam has joined #openstack-lbaas17:24
*** phuoc_ has joined #openstack-lbaas17:26
*** ramishra has quit IRC17:26
openstackgerritMichael Johnson proposed openstack/octavia master: Neutron-LBaaS to Octavia migration tool  https://review.openstack.org/55442017:28
*** kbyrne has quit IRC17:29
openstackgerritMichael Johnson proposed openstack/neutron-lbaas master: Add a job to test the nlbaas2octavia migration.  https://review.openstack.org/57894217:29
*** phuoc has quit IRC17:29
*** kbyrne has joined #openstack-lbaas17:29
*** kobis has quit IRC17:34
*** SumitNaiksatam_ has joined #openstack-lbaas18:12
*** SumitNaiksatam has quit IRC18:13
*** SumitNaiksatam_ is now known as SumitNaiksatam18:13
*** pcaruana has quit IRC18:14
*** SumitNaiksatam has quit IRC18:20
rm_workjohnsom: do you know what happened to ispp? >_>18:23
johnsomUmm, so. I think they were on yesterday.18:24
rm_worki meant like18:25
johnsomThey had something odd in a container install that had a DB revision I couldn't find giving an error. I guess a rebuild fixed it. Maybe someone had loaded a patch that had a version in it.18:25
rm_workwhat thing happened to their octavia install18:25
rm_workah18:25
rm_workoh right, k18:25
rm_workwas wondering if they were talking about a failover storm or something18:25
johnsomNope it was this: https://pastebin.com/XwPcZG2G18:27
*** ajun has joined #openstack-lbaas18:46
*** ajun is now known as abaindur18:48
openstackgerritSwaminathan Vasudevan proposed openstack/neutron-lbaas master: neutron-lbaas haproxy agent prevent vif unplug when failover occurs  https://review.openstack.org/57896619:12
*** kobis has joined #openstack-lbaas19:14
*** abaindur has quit IRC19:23
*** kobis has quit IRC19:35
*** abaindur has joined #openstack-lbaas19:35
*** abaindur has quit IRC20:02
*** abaindur has joined #openstack-lbaas20:03
*** amuller has quit IRC20:09
*** abaindur has quit IRC20:21
*** abaindur has joined #openstack-lbaas20:21
openstackgerritMichael Johnson proposed openstack/neutron-lbaas master: Add a job to test the nlbaas2octavia migration.  https://review.openstack.org/57894220:32
rm_workdifferent patchset?20:37
rm_worki assumed it'd just be folded in20:38
johnsomI need it to run from the neutron-lbaas repo to have the neutron-lbaas DB and LB to migrate...20:59
rm_workoh is the script in octavia?21:00
johnsomI guess I could do it from octavia... now that I think of it21:00
rm_worki figured it'd be in neutron-lbaas21:00
johnsomYeah, I had the code in octavia land21:00
rm_workseems like it'd be kinda cruft in octavia21:00
rm_workbut if it was in n-lbaas it'd go away when n-lbaas did21:00
* rm_work shrugs21:00
johnsomYeah, that is kind of true....21:00
johnsomYeah, I will move it over. I think originally I was thinking I would need Octavia objects, but the path I took had no dependencies.21:01
openstackgerritGerman Eichberger proposed openstack/neutron-lbaas master: Gate API test for the lbaasv2-proxy plugin  https://review.openstack.org/53935021:03
johnsomI will move it over after this test run, I'm working on figuring out how to parse the config files from ansible21:03
*** ivve has quit IRC21:08
*** numans_ has quit IRC21:35
*** numans has joined #openstack-lbaas21:36
*** nmanos has quit IRC21:40
*** rcernin has joined #openstack-lbaas21:42
*** fnaval has quit IRC21:44
abaindurjohnsom: I was able to build the octavia image, but only on an Ubuntu 14 vm21:47
johnsomOk, yeah, I think the folks in the DIB channel found the bug in diskimage-builder last night.21:48
abaindursee my msgs earlier in #openstack-dib. When I tried on another centos machine (running 7.5), it gets an error about debootstrap not found. But in the errors from yesterday on my centos 7.2 machine, I dont see it ever even loading the 08-debootstrap element21:48
abaindurany idea why on cntos 7.5 it might be trying this, when i do not see it on my centos 7.2 run?21:48
abaindurthat element is being loaded from the diskimage_builder module, not the octavia/elements21:48
johnsomHmm, no. Frankly I don't use centos much.21:49
abaindur2018-06-28 06:31:23.399 | Copying hooks /usr/lib/python2.7/site-packages/diskimage_builder/elements/debootstrap/root.d/08-debootstrap21:49
johnsomYeah, it's a DIB issue and not an octavia issue21:49
abainduri never see that in my centos 7.2 failure from yday ^^21:49
abaindurah ok, i mainly use centos21:49
johnsomWe don't override or replace any of their elements21:49
abainduri am wondering if ubuntu can only build an ubuntu diskimage?21:49
abaindurI havent tried building a centos amphora on my centos yet21:50
johnsomYeah, there are other here in the channel that are centos/rhel folks. I'm just not one of them.21:50
johnsomNah, it *should* be possible to build ubuntu on a centos host21:50
abaindurWhat does the disk size in the image matter, given that we have a amp flavor, btw?21:50
abaindurif I just create it with a 2 GB disk, but my flavor has a larger disk21:51
abaindurDoes the image itself really matter? Wondering if we can just create 1 image, or the need to host multiple images w/ different -s sizes21:51
johnsomIt's the size of the filesystem inside the image after decompression.  So you could have a flavor with a larger size, but the filesystem would still be 2GB or whatever the image is built with21:52
abaindurWhat is a good recommendation for size? in a production use case?21:53
abainduris a 4GB or 8GB disk big enough?21:53
johnsomIt depends on your connections rate and if you disable connection logging in the configuration (which may have not merged yet).  The 2GB is not big enough for production for sure.21:54
johnsomIf you have a low number of connections smaller is fine, but otherwise a larger disk is good. I think one provider is running with 10GB, but I don't know how much of that is getting used.21:55
johnsomThin provisioned it doesn't matter too much, you are really just setting a cap21:56
abaindurah ok thanks21:57
*** KeithMnemonic has quit IRC22:04
*** lxkong has joined #openstack-lbaas22:04
*** j-hinman has joined #openstack-lbaas22:12
*** j-hinman has quit IRC22:19
rm_workyeah i need to up the size i use in my images, still running with 3G22:29
rm_workgonna run out of space at some point due to logging :/22:29
abaindurWhere is logging configured in amp? Is there a way to disable it? https://docs.openstack.org/octavia/pike/configuration/configref.html#controller_worker.amp_flavor_id22:37
abaindurIs there a way to setup logrotate in that case, or do the logs grow unbounded by default?22:38
johnsomLogrotate is on by default22:38
johnsomI think it is daily though22:38
johnsomThis is the log disable patch: https://review.openstack.org/#/c/566741/22:39
rm_workthere is a patch up to allow it to be disabled22:39
johnsomStill needs some reviews22:39
rm_work^^ yes that22:39
rm_workjohnsom: will logrotate actually keep them from building up forever?22:39
rm_workit has like a max-kept?22:39
johnsomYes, it does22:39
rm_workcool ok22:39
rm_workless worried now :P22:39
johnsomBut if your log rate is high enough you will fill the disk before the daily rotate/compress22:40
rm_workyes was more worried about unbounded22:40
johnsomSo it looks like the latest patches may have fixed the nested virt issue.  Still testing, but we may be able to re-enable that22:41
rm_workwhat did they patch?22:41
johnsomThere are three KVM fixes in the latest kernel.  Nothing that points directly to the issue, but we are struggling to reproduce it now on OVH and limeston22:41
johnsomlimestone22:42
abaindurjohnsom: What nested virt issues? we are using nested virtualization for testing purposes :)22:42
johnsomYeah, so are we.  We had two hosting providers that qemu would crash when using KVM.  It started over a year ago.22:43
johnsomWe had to exclude them from using kvm22:43
johnsomWhich was painful as it can take 50 minutes off our test runs22:43
*** fnaval has joined #openstack-lbaas22:45
*** threestrands has joined #openstack-lbaas23:05
*** threestrands has quit IRC23:05
*** threestrands has joined #openstack-lbaas23:05
*** abaindur has quit IRC23:08
*** abaindur has joined #openstack-lbaas23:22
*** threestrands has quit IRC23:38
« Wednesday, 2018-06-27 Index Friday, 2018-06-29 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!