| *** sdake_ has joined #openstack-ansible | 00:08 | |
| *** britthou_ has joined #openstack-ansible | 00:09 | |
| *** sdake has quit IRC | 00:11 | |
| *** britthouser has quit IRC | 00:12 | |
| *** cloudtrainme has joined #openstack-ansible | 00:15 | |
| cloudnull | evrardjp: it took all day but I finally got to the review of the patch https://review.openstack.org/218818 - looks like a good start btw. I've done my best to review the entire role / code and I like where its going. | 00:16 |
|---|---|---|
| cloudnull | I think that the effort in https://review.openstack.org/#/c/217517/ should merge with https://review.openstack.org/218818 | 00:17 |
| *** sdake has joined #openstack-ansible | 00:18 | |
| cloudnull | that said, I'll let you and Shu Shen (sshen) work that out if needed. | 00:18 |
| cloudnull | evrardjp: great work on that role and i'll try to sync up with you in the morrow | 00:19 |
| *** sdake_ has quit IRC | 00:21 | |
| *** sdake_ has joined #openstack-ansible | 00:47 | |
| *** sdake__ has joined #openstack-ansible | 00:49 | |
| *** sdake has quit IRC | 00:50 | |
| *** sdake_ has quit IRC | 00:53 | |
| *** sdake__ is now known as sdake | 00:54 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 01:00 | |
| openstackgerrit | Brian Moss proposed stackforge/os-ansible-deployment: [OSAD doc] Update link to Cloud Admin Guide https://review.openstack.org/219085 | 01:11 |
| *** BjoernT has quit IRC | 01:15 | |
| *** cloudtrainme has quit IRC | 01:55 | |
| *** tlian2 has quit IRC | 02:19 | |
| mhayden | okay, so do i make a spec first or a blueprint? | 02:38 |
| *** britthou_ has quit IRC | 02:53 | |
| *** britthouser has joined #openstack-ansible | 02:53 | |
| sigmavirus24 | mhayden: yes | 03:03 |
| sigmavirus24 | mhayden: we used to primarily use blueprints | 03:03 |
| mhayden | sigmavirus24: i see what you did there | 03:03 |
| sigmavirus24 | I think now they're sort of placeholders for tracking development via launchpad sort of | 03:03 |
| sigmavirus24 | So my workflow is write a spec, before pushing create a blueprint with the summary. Then create spec referencing the blueprint name | 03:03 |
| sigmavirus24 | Make sense/ | 03:03 |
| mhayden | gotcha | 03:08 |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment-specs: Adding spec for named veth interfaces https://review.openstack.org/219105 | 03:12 |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment-specs: Adding spec for named veth interfaces https://review.openstack.org/219105 | 03:14 |
| mhayden | i failed at rst :/ | 03:17 |
| *** Apsu is now known as JorgeSonata | 03:20 | |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment-specs: Adding spec for named veth interfaces https://review.openstack.org/219105 | 03:20 |
| mhayden | now it should pass | 03:21 |
| *** JorgeSonata is now known as Apsu | 03:22 | |
| *** shausy has joined #openstack-ansible | 03:26 | |
| *** darrenc is now known as darrenc_afk | 03:28 | |
| sigmavirus24 | mhayden: everyone fails at rst the first few times | 03:33 |
| sigmavirus24 | I've been writing rst for years now and still fail | 03:34 |
| mhayden | i missed the tox stuff in the instructions :) | 03:34 |
| mhayden | sigmavirus24: next spec will suggest a migration to cfengine | 03:34 |
| * mhayden cackles | 03:35 | |
| sigmavirus24 | dude I cfengine'd in college | 03:38 |
| *** britthou_ has joined #openstack-ansible | 03:46 | |
| *** sigmavirus24 has quit IRC | 03:48 | |
| *** britthouser has quit IRC | 03:48 | |
| *** sigmavirus24 has joined #openstack-ansible | 03:49 | |
| *** THiagoCMC has quit IRC | 03:52 | |
| *** britthou_ has quit IRC | 03:55 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 03:59 | |
| *** markvoelker has joined #openstack-ansible | 04:00 | |
| *** markvoelker_ has joined #openstack-ansible | 04:02 | |
| *** shausy has quit IRC | 04:04 | |
| *** markvoelker has quit IRC | 04:04 | |
| *** darrenc_afk is now known as darrenc | 04:54 | |
| *** cloudtrainme has joined #openstack-ansible | 04:56 | |
| *** shausy has joined #openstack-ansible | 05:21 | |
| *** cloudtrainme has quit IRC | 05:39 | |
| *** markvoelker_ has quit IRC | 05:53 | |
| *** shausy has quit IRC | 06:04 | |
| *** sdake_ has joined #openstack-ansible | 06:05 | |
| *** shausy has joined #openstack-ansible | 06:05 | |
| *** sdake has quit IRC | 06:08 | |
| *** sdake has joined #openstack-ansible | 06:11 | |
| *** sdake has quit IRC | 06:11 | |
| *** sdake has joined #openstack-ansible | 06:13 | |
| *** sdake_ has quit IRC | 06:15 | |
| *** shoutm has quit IRC | 06:17 | |
| *** Mudpuppy_ has quit IRC | 06:26 | |
| *** Mudpuppy has joined #openstack-ansible | 06:26 | |
| *** Mudpuppy has quit IRC | 06:30 | |
| *** alextricity has quit IRC | 07:22 | |
| *** shoutm has joined #openstack-ansible | 07:24 | |
| evrardjp | good morning everyone | 07:41 |
| *** gparaskevas has joined #openstack-ansible | 08:13 | |
| mattt | morning evrardjp | 08:14 |
| evrardjp | what does "NIT" mean in a review? it's like a software bug so ugly it looks like its animal counterpart with the same name? | 08:16 |
| odyssey4me | morning all | 08:17 |
| odyssey4me | evrardjp a nit is just a small minor comment which ideally should be fixed, but is generally not enought o hold back a review | 08:18 |
| evrardjp | good morning odyssey4me | 08:18 |
| evrardjp | thanks for the info | 08:18 |
| evrardjp | it's some kind of abreviation? like "not important topic"? | 08:19 |
| odyssey4me | fyi mattt hughsaunders andymccr https://review.openstack.org/218572 unblocks master by resolving the ssh conneciton failures | 08:21 |
| odyssey4me | evrardjp your comments on your changes to ssh config in ansible.cfg interest me though | 08:21 |
| odyssey4me | evrardjp nit is short for nit-pick I think | 08:21 |
| evrardjp | this is what I perceived | 08:22 |
| odyssey4me | 'Look for and criticize trivial faults' | 08:22 |
| evrardjp | I think it should be more tested | 08:22 |
| evrardjp | if it's real, it's a damn obscure bug | 08:23 |
| evrardjp | I understand with the example given the dictionary, thanks for improving my english skills ;) | 08:24 |
| odyssey4me | svg yes, 11.2.0 is being held back due to some breaking changes we found - if we can get https://review.openstack.org/218572 merged today, then I'll finalise https://review.openstack.org/217014 and we can get that merged which is a final patch for 11.2.0 | 08:39 |
| svg | thx odyssey4me | 08:40 |
| odyssey4me | we had a holiday yesterday - I did hope that we'd have more votes on the patch from the US guys, but that didn't happen :/ | 08:41 |
| *** shausy has quit IRC | 09:04 | |
| *** shausy has joined #openstack-ansible | 09:05 | |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: [WIP] Implementation of keepalived for haproxy https://review.openstack.org/218818 | 09:36 |
| evrardjp | odyssey4me: I've seen your review after this commit, but I'm glad I don't have to change anything... I thought of it without you... You see, progress! | 09:39 |
| svg | evrardjp: Would that patch allow for using haproxy in production? | 09:40 |
| evrardjp | that's what we'll try | 09:40 |
| evrardjp | haproxy definitely needs fine tuning in terms of timers | 09:41 |
| evrardjp | we've got our own values | 09:41 |
| evrardjp | it's really painful to setup | 09:41 |
| evrardjp | to correctly setup | 09:41 |
| evrardjp | this just improves HA of haproxy | 09:41 |
| evrardjp | by setting it in one or multiple vrrp groups, depending on what you're looking for | 09:42 |
| svg | right now we use an extrenal BigIP F5 device; might be need to not depend on that anymore | 09:47 |
| svg | *neat | 09:47 |
| *** Mudpuppy has joined #openstack-ansible | 10:00 | |
| *** Mudpuppy has quit IRC | 10:04 | |
| odyssey4me | evrardjp agreed, I think some attention should be paid to some of the backend config in haproxy | 10:07 |
| odyssey4me | svg using the F5 would still be something that we recommend, bu haproxy is becoming more usable for production in smaller deployments | 10:08 |
| *** javeriak has joined #openstack-ansible | 10:13 | |
| *** javeriak has quit IRC | 10:18 | |
| *** britthouser has joined #openstack-ansible | 11:09 | |
| *** jaypipes has joined #openstack-ansible | 11:14 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Removed default lxc profile on container create https://review.openstack.org/217014 | 11:15 |
| *** shoutm has quit IRC | 11:19 | |
| evrardjp | odyssey4me: the interesting thing is that haproxy's price doesn't scale the same way as F5 depending on your bandwidth | 11:25 |
| odyssey4me | evrardjp yes :) what I am saying though is that RPC has decided that it will not be using haproxy for load balancing, so the maintenance of the haproxy role for production use is totally up to the community that wants to use it for production | 11:28 |
| evrardjp | yes, I understood, and it seems logical | 11:28 |
| evrardjp | we are working on OSS after all | 11:28 |
| *** javeriak has joined #openstack-ansible | 11:29 | |
| odyssey4me | :) | 11:29 |
| *** pradk has joined #openstack-ansible | 11:50 | |
| *** javeriak_ has joined #openstack-ansible | 12:00 | |
| *** javeriak has quit IRC | 12:02 | |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment-specs: Adding spec for named veth interfaces https://review.openstack.org/219105 | 12:16 |
| mgariepy | good morning everyone | 12:20 |
| evrardjp | good morning mgariepy | 12:21 |
| evrardjp | is someone interested by having the nova_console secured? | 12:22 |
| evrardjp | or I am the only one? Because it's an upstream bug I think | 12:22 |
| *** javeriak_ has quit IRC | 12:39 | |
| odyssey4me | evrardjp I think we all are, but yes as I recall there's some sort of upstream bug - don't quote me on that | 12:43 |
| evrardjp | I've edited the js for nova console, and I got it working with ssl termination on my load balancers | 12:48 |
| evrardjp | no problem of protocol mismatch anymore | 12:48 |
| evrardjp | however it's not real security, the connection is secure, but the spice page still shows: "keyboard channel is insecure" | 12:49 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 12:51 | |
| *** javeriak_ has joined #openstack-ansible | 12:52 | |
| shausy | @here during one of my play, im getting the following error --> undefined variables: 'dict object' has no attribute 'vips'", 'failed': True} | 12:55 |
| shausy | any pointer for this issue | 12:55 |
| *** tlian has joined #openstack-ansible | 12:59 | |
| evrardjp | shausy: check you identation first | 13:00 |
| evrardjp | and see if you your variable has "vips" | 13:00 |
| evrardjp | the easiest way for us to help you is probably by copy pasting more content somewhere | 13:00 |
| *** scarlisle has joined #openstack-ansible | 13:04 | |
| shausy | evrardjp, i have checked, indn with yml verifier.. no issues there | 13:09 |
| mattt | shausy: i see no reference to "vips" on master, which branch are you using? | 13:11 |
| *** gparaskevas has quit IRC | 13:13 | |
| *** sdake has quit IRC | 13:17 | |
| shausy | mattt, 1.7.2 version stable we are using | 13:17 |
| odyssey4me | shausy is that your ansible version? I thikn that mattt meant what version of openstack-ansible/os-ansible-deployment are you using? | 13:19 |
| *** cloudtrainme has joined #openstack-ansible | 13:19 | |
| mattt | shausy: yep which branch of os-ansible-deployment | 13:19 |
| *** cloudtrainme has quit IRC | 13:19 | |
| openstackgerrit | Hugh Saunders proposed stackforge/os-ansible-deployment: Add variable for cirros url https://review.openstack.org/217310 | 13:20 |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Update how neutron migrations are handled https://review.openstack.org/215584 | 13:20 |
| *** sdake has joined #openstack-ansible | 13:21 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 13:24 | |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Update how neutron migrations are handled https://review.openstack.org/215584 | 13:30 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated juno to include fix for CVE-2015-3241 - 26 Aug 2015 https://review.openstack.org/217098 | 13:32 |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Update how neutron migrations are handled https://review.openstack.org/215584 | 13:34 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 13:39 | |
| palendae | odyssey4me: Does https://review.openstack.org/#/c/218572/ not have a bug associated? | 13:45 |
| odyssey4me | palendae there is - check the bottom of the commit msg | 13:46 |
| palendae | Huh, so the topic didn't update | 13:46 |
| odyssey4me | palendae ah, that's a legacy of me merging two patches from my wip testing over the weekend | 13:47 |
| odyssey4me | the topic is usually based on your branch, not on the referenced bug | 13:47 |
| *** Mudpuppy has joined #openstack-ansible | 13:49 | |
| palendae | Unless you have a bug in the first commit message | 13:51 |
| palendae | But cool, silly me for not to check the commit message :) | 13:53 |
| odyssey4me | palendae the first commit message had one ;) https://review.openstack.org/#/c/218572/1//COMMIT_MSG,cm | 13:53 |
| *** KLevenstein has joined #openstack-ansible | 13:53 | |
| cloudnull | morning | 13:54 |
| odyssey4me | o/ cl | 13:55 |
| odyssey4me | cloudnull | 13:55 |
| cloudnull | how was your bank holiday ? :) | 13:58 |
| odyssey4me | cloudnull much sleeping and movie watching was done, oh and good food | 13:59 |
| cloudnull | sounds like a perfect day! | 13:59 |
| palendae | Huh | 13:59 |
| cloudnull | evrardjp: did you ever get the dynamic inventory bits sorted ? | 14:00 |
| * cloudnull still reading scroll back | 14:00 | |
| evrardjp | cloudnull: didn't got the chance to do so | 14:01 |
| evrardjp | it's not that it doesn't work | 14:01 |
| evrardjp | it's that it overrides some of my work | 14:01 |
| cloudnull | ah. where is it overriding ? | 14:01 |
| evrardjp | do you mind if we take this one as example? https://review.openstack.org/#/c/218818/ | 14:02 |
| cloudnull | sur e | 14:03 |
| evrardjp | If I rename the conf.d/keepalived.yml.example into keepalived.yml | 14:03 |
| evrardjp | I do any operation | 14:03 |
| evrardjp | I will see that in my openstack_inventory.json: | 14:03 |
| *** yaya has joined #openstack-ansible | 14:03 | |
| cloudnull | yes it merges on ansible playbook execution. | 14:04 |
| *** cloudtrainme has joined #openstack-ansible | 14:04 | |
| evrardjp | physical_host_group: "keepalived_hosts" | 14:04 |
| evrardjp | it must be ok, but if I use the same hosts for another purpose (let's say haproxy)? | 14:05 |
| evrardjp | if it overwrites the physical_host_group, won't it be problematic? | 14:05 |
| *** KLevenstein has quit IRC | 14:05 | |
| evrardjp | atm the inventory seems fine for me | 14:05 |
| evrardjp | but I'd like to make sure it will always be the same for everyone | 14:06 |
| *** spotz_zzz is now known as spotz | 14:06 | |
| evrardjp | and good morning cloudnull ;) | 14:06 |
| *** KLevenstein has joined #openstack-ansible | 14:07 | |
| *** phalmos has joined #openstack-ansible | 14:07 | |
| evrardjp | odyssey4me: you said you had a nice holiday with good food? Is that even possible in the UK? (I meant having good food?) >< | 14:07 |
| evrardjp | I guess lots of restaurants with a different origin than UK :p | 14:08 |
| *** javeriak has joined #openstack-ansible | 14:10 | |
| cloudnull | evrardjp: yes that could be problematic as the physical_host_group is used in a few places. that said its a legacy variable and only used within inventory. | 14:11 |
| cloudnull | http://cdn.pasteraw.com/9j4gfvej1i3nyat7ez3dvbb1tyd8mru | 14:11 |
| palendae | mhayden: Updated https://review.openstack.org/#/c/219105/ with comments | 14:11 |
| *** jmckind has joined #openstack-ansible | 14:11 | |
| cloudnull | ^ thats where the physical_host_group variable is defined. | 14:11 |
| *** javeriak_ has quit IRC | 14:11 | |
| evrardjp | it's maybe because I used a wrong way to define a new type of service | 14:12 |
| evrardjp | probably* | 14:12 |
| mhayden | palendae: thanks -- looking them over | 14:12 |
| cloudnull | so it should not cause deployment problems and in the liberty/mitaka timeframe, if palendae gets some time to work on upgrade implementations, we should be able to clean up some of the todos and make those kinds of things go away | 14:12 |
| evrardjp | that's good news cloudnull | 14:13 |
| evrardjp | just to be sure: | 14:13 |
| evrardjp | on this: https://review.openstack.org/#/c/218818/3/etc/openstack_deploy/env.d/keepalived.yml | 14:13 |
| palendae | So cloudnull and evrardjp, I'm working on specs for upgrade stuff (like liberty's upgrade path and an upgrade gate job) and docs for the env, conf, and inventory, but it's been an 'as we have time' thing | 14:13 |
| evrardjp | I think if ppl want to bring their own kind of components, it could be useful to understand how env.d really works | 14:14 |
| palendae | Yep | 14:15 |
| palendae | I agree | 14:15 |
| evrardjp | while it seemed obvious at first, when reading the env.d files, creating some env.d has overwrite impact | 14:15 |
| evrardjp | (if hosts are reused for example) | 14:15 |
| evrardjp | also here: | 14:16 |
| evrardjp | https://review.openstack.org/#/c/218818/3/etc/openstack_deploy/env.d/keepalived.yml | 14:16 |
| evrardjp | in container_skel: | 14:16 |
| evrardjp | the first line, here keepalived_container, should be "standardized" | 14:16 |
| evrardjp | I had the feeling it was weird to have some kind of group of containers called keepalived_containers and another group called keepalived_container | 14:17 |
| openstackgerrit | Matthew Kassawara proposed stackforge/os-ansible-deployment: Disable scatter-gather offload on host bridges https://review.openstack.org/219292 | 14:21 |
| openstackgerrit | Matthew Kassawara proposed stackforge/os-ansible-deployment: Disable scatter-gather offload on host bridges https://review.openstack.org/219292 | 14:25 |
| *** alejandrito has joined #openstack-ansible | 14:26 | |
| *** javeriak has quit IRC | 14:28 | |
| cloudnull | sorry was in side convos here. evrardjp host wise it should be ok to reuse hosts in various files as long as the groups are different in the confi files. | 14:37 |
| evrardjp | that makes sense | 14:37 |
| cloudnull | maybe we need to revist the precedence system in how those conf files are used to generate inv json. | 14:38 |
| odyssey4me | cloudnull so fyi I had some conversations with -infra and did a little research on a few topics | 14:38 |
| evrardjp | so there is nothing to change for the env.d/conf.d in my commit | 14:38 |
| cloudnull | so that we can have it be more deterministic instead of doing an inventory hash merge. | 14:38 |
| evrardjp | I was just surprised that merging was doing overwrite | 14:38 |
| cloudnull | evrardjp: i dont think so, and the physical_host_group variable can be ignored. | 14:39 |
| cloudnull | odyssey4me: on what ? | 14:39 |
| evrardjp | but if it's on non-important stuff, then it's ok | 14:39 |
| odyssey4me | with regards to https://review.openstack.org/218611 I was investigating whether there was a difference made to the success of builds if we don't use the rackspace mirror specifically | 14:39 |
| odyssey4me | it does seem to improve things when builds are running in hpcloud | 14:39 |
| odyssey4me | there's some speculation in -infra about whether hpcloud intercepts apt requests to archive.ubuntu and directs them to local caches | 14:40 |
| cloudnull | yea in ci there were 6 successes 5 of which were in hp on https://review.openstack.org/#/c/218611 | 14:40 |
| odyssey4me | but nothing concrete - I'd like to run some tests in your hpcloud account if that's ok? | 14:40 |
| cloudnull | sure. | 14:40 |
| cloudnull | ill make you a tenant | 14:41 |
| odyssey4me | I saw a post that hpcloud has a specific mirror, but -infra doesn't appear to be using it - I'm thinking that perhaps we should use the generic approach of using the hosts repositories for AIO's - but then for gate jobs we can select a repo based on known upstream repository URL's that give the best success. | 14:41 |
| palendae | odyssey4me: Did the image-based approach fail | 14:42 |
| palendae | ? | 14:42 |
| odyssey4me | ie an AIO build should use whatever's configured on the initial host build - then the gate checks should use known repositories which work best for rax/hpcloud | 14:42 |
| odyssey4me | palendae image-based approach? | 14:42 |
| palendae | odyssey4me: I thought you linked to something that showed we could deploy our own custom OS images on infra | 14:43 |
| *** alextricity_h has joined #openstack-ansible | 14:43 | |
| odyssey4me | I think that the above approach for the AIO vs Gate configs of the repositories is the best way to be dev friendly, but also configure what we think is best for the gate. It seperates the concerns better. | 14:43 |
| palendae | True enough | 14:43 |
| odyssey4me | palendae ah, that was the next thing I wanted to mention to cloudnull | 14:44 |
| palendae | The image based approach would save time, but also not exercise certain things | 14:44 |
| palendae | But if those things are broken in infra, not much point in exercising them either | 14:44 |
| odyssey4me | I didn't actually do an image build test of any sort (yet) - but it does seem that there are projects that are building images and publishing them. | 14:44 |
| *** javeriak has joined #openstack-ansible | 14:44 | |
| palendae | Or teaming with infra to fix it | 14:44 |
| odyssey4me | This would mean that we could publish LXC images however we choose - perhaps every tag or whatever. | 14:46 |
| odyssey4me | This could be very helpful for upgrade testing - to make the check run a heck of a lot quicker. | 14:46 |
| odyssey4me | It would also be useful for role specific gate checks where we may wish to use the images for the infrastructure, then only do the full deploys for the openstack bits... or something to that effect. | 14:47 |
| odyssey4me | Anyway, more specifically it opens up the options. | 14:47 |
| *** shoutm has joined #openstack-ansible | 14:49 | |
| *** phalmos has quit IRC | 14:49 | |
| *** sdake_ has joined #openstack-ansible | 14:53 | |
| *** k_stev has joined #openstack-ansible | 14:54 | |
| *** phalmos has joined #openstack-ansible | 14:55 | |
| *** Mudpuppy has quit IRC | 14:57 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Remove hardcoded config drive enforcement https://review.openstack.org/218480 | 14:58 |
| *** sdake has quit IRC | 14:58 | |
| *** Mudpuppy has joined #openstack-ansible | 14:58 | |
| *** KLevenstein has quit IRC | 14:58 | |
| *** yaya has quit IRC | 14:58 | |
| odyssey4me | here's another funny | 14:59 |
| *** KLevenstein has joined #openstack-ansible | 14:59 | |
| odyssey4me | the issue in hpcloud-b4 seems very specifically tied to the containers being unable to communicate with dns or the outside world in any way | 15:00 |
| odyssey4me | only the containers though, the host is perfectly fine | 15:00 |
| palendae | odyssey4me, cloudnull, mattt, andymccr, hughsaunders, sigmavirus24, other cores - I want to submit a spec for upgrade testing, however, I don't think it fits just in kilo or just in liberty. Anyone have opinions where to put it? Top of specs repo? | 15:05 |
| sigmavirus24 | palendae: yes | 15:05 |
| *** galstrom_zzz is now known as galstrom | 15:06 | |
| sigmavirus24 | put it in liberty because it's new feature development | 15:06 |
| sigmavirus24 | just because it's in one doesn't mean it stops applying to future releases | 15:06 |
| odyssey4me | palendae it needs a finite target | 15:06 |
| *** sdake_ is now known as sdake | 15:06 | |
| sigmavirus24 | "This will be implemented for liberty" | 15:06 |
| palendae | Ok, my plan was to implement in kilo, too | 15:06 |
| palendae | And I think kilo and liberty implementations may vary | 15:06 |
| palendae | So perhaps we have different specs? | 15:06 |
| odyssey4me | agreed with sigmavirus24 - the upgrade blueprint must specifically target kilo->liberty | 15:06 |
| palendae | Alright, so are people saying to leave kilo as is? | 15:07 |
| sigmavirus24 | palendae: that might work | 15:07 |
| palendae | I'd really like to fix that script in kilo... | 15:07 |
| palendae | But Liberty is greenfield, too | 15:07 |
| odyssey4me | palendae if at some point it's decided that kilo's method needs an overhaul, then that should be blueprinted then | 15:07 |
| sigmavirus24 | palendae: that works too | 15:07 |
| sigmavirus24 | kilo specs are still open | 15:07 |
| sigmavirus24 | as far as I remember | 15:07 |
| sigmavirus24 | odyssey4me: I think most of us agree it needs an overhall sadly | 15:08 |
| palendae | odyssey4me: Fair. I was going to start there because I know at least part of what needs to be done | 15:08 |
| odyssey4me | for now, liberty is soon - let's get it right in a greener field, then learn lessons and apply them to the kilo upgrade method if we can afford the time | 15:08 |
| *** javeriak_ has joined #openstack-ansible | 15:08 | |
| odyssey4me | sigmavirus24 sure we all agree, but do we have the time and resource allocation to do it | 15:08 |
| sigmavirus24 | odyssey4me: sure | 15:09 |
| sigmavirus24 | Deciding to do the work and deciding the work needs to be done are two different things | 15:09 |
| odyssey4me | palendae I suppose I'm saying that we should focus on one and do it well - not on both | 15:09 |
| palendae | Ok, so I'll work on that. I'd really like to get cores looking at the liberty upgrade path spec too, and stop merging changes for kilo run-upgrades.sh into master | 15:09 |
| sigmavirus24 | As in "we know this is something broken and should be fixed" versus "we can fix that now" | 15:09 |
| palendae | odyssey4me: Yeah, and that makes sense | 15:09 |
| sigmavirus24 | Specs are an indication of the former, not the later | 15:09 |
| sigmavirus24 | *latter | 15:09 |
| *** javeriak has quit IRC | 15:09 | |
| odyssey4me | palendae I haven't seen a run-upgrade patch into master for ages | 15:10 |
| palendae | There were enough to invalidate my WIP patch that guts it | 15:10 |
| * palendae shrugs | 15:10 | |
| *** shoutm has quit IRC | 15:10 | |
| palendae | Mostly just want to be clear, but I can't be too stern about it with out the spec accepted | 15:10 |
| palendae | Gonna go respond to comments in it right now | 15:11 |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Update how neutron migrations are handled https://review.openstack.org/215584 | 15:11 |
| palendae | odyssey4me, sigmavirus24 Do you remember the title for the epoch discussion on the openstack ML? | 15:15 |
| sigmavirus24 | palendae: no but it was started by harlowja | 15:16 |
| sigmavirus24 | That was back in June maybe? | 15:16 |
| sigmavirus24 | Perhaps earlier? | 15:16 |
| sigmavirus24 | If you search by month and look for harlowja and me in a conversation with epoch somewhere in the title you'll find it | 15:16 |
| palendae | http://lists.openstack.org/pipermail/openstack-dev/2015-July/069085.html ? | 15:17 |
| odyssey4me | palendae [openstack-dev] [openstack-ansible] [os-ansible-deployment] Kilo -> Liberty Upgrade Problems | 15:18 |
| sigmavirus24 | palendae: that's it | 15:19 |
| odyssey4me | oh, you're meaning the other discussion | 15:19 |
| odyssey4me | the one that spurred this one | 15:19 |
| odyssey4me | that was fragmented among many spats | 15:19 |
| palendae | I'll link both | 15:19 |
| palendae | Why not | 15:19 |
| sigmavirus24 | odyssey4me: I think you're confused. It was one thread but dotted over a couple of weeks | 15:20 |
| sigmavirus24 | I didn't want to get too angry in one sitting | 15:20 |
| sigmavirus24 | That isn't quite healthy | 15:20 |
| odyssey4me | lol | 15:20 |
| sigmavirus24 | There isn't enough alcohol in Wisconsin to calm me down when people say "downtream are the only people who should care about that" | 15:21 |
| sigmavirus24 | Because that's not how that works | 15:21 |
| sigmavirus24 | You're using versions semantically already, you should be as semantic as possible to be as clear to all users what those versions mean | 15:21 |
| * sigmavirus24 steps away from keyboard for a short period of time | 15:21 | |
| *** galstrom is now known as galstrom_zzz | 15:23 | |
| bgmccollum | vPOTAT.0 | 15:23 |
| *** galstrom_zzz is now known as galstrom | 15:23 | |
| *** k_stev has quit IRC | 15:26 | |
| sigmavirus24 | Hey potatoes are actually useful bgmccollum | 15:26 |
| sigmavirus24 | You can run a cloud on potato power | 15:26 |
| sigmavirus24 | Not an openstack cloud because those just don't run | 15:26 |
| sigmavirus24 | But any other cloud, certainly | 15:26 |
| palendae | odyssey4me, cloudnull: Replied to comments on https://review.openstack.org/#/c/207713/, but working on a new draft too | 15:29 |
| *** k_stev has joined #openstack-ansible | 15:31 | |
| *** jmckind has quit IRC | 15:31 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: [WIP] Diagnostic test https://review.openstack.org/219313 | 15:31 |
| *** javeriak has joined #openstack-ansible | 15:32 | |
| *** galstrom is now known as galstrom_zzz | 15:33 | |
| *** cloudtra_ has joined #openstack-ansible | 15:33 | |
| *** javeriak_ has quit IRC | 15:33 | |
| *** cloudtrainme has quit IRC | 15:33 | |
| *** jmckind has joined #openstack-ansible | 15:34 | |
| *** alop has joined #openstack-ansible | 15:34 | |
| *** galstrom_zzz is now known as galstrom | 15:44 | |
| *** cloudtra_ has quit IRC | 15:44 | |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Update how neutron migrations are handled https://review.openstack.org/215584 | 15:51 |
| *** shausy has quit IRC | 15:56 | |
| *** k_stev has quit IRC | 16:01 | |
| *** k_stev has joined #openstack-ansible | 16:01 | |
| *** k_stev has quit IRC | 16:02 | |
| *** k_stev has joined #openstack-ansible | 16:02 | |
| *** k_stev has quit IRC | 16:03 | |
| *** k_stev has joined #openstack-ansible | 16:03 | |
| *** jmckind has quit IRC | 16:03 | |
| *** sdake_ has joined #openstack-ansible | 16:05 | |
| palendae | Bug tirage? | 16:06 |
| palendae | triage, too | 16:06 |
| *** k_stev1 has joined #openstack-ansible | 16:06 | |
| *** k_stev has quit IRC | 16:08 | |
| *** k_stev2 has joined #openstack-ansible | 16:08 | |
| odyssey4me | uh cloudnull ^ ? | 16:08 |
| *** k_stev2 has quit IRC | 16:08 | |
| *** sdake has quit IRC | 16:09 | |
| *** k_stev has joined #openstack-ansible | 16:09 | |
| cloudnull | ah . yea. sorry . | 16:10 |
| * cloudnull in email hell | 16:10 | |
| cloudnull | first up https://bugs.launchpad.net/openstack-ansible/+bug/1488639 | 16:10 |
| openstack | Launchpad bug 1488639 in openstack-ansible "Profile Ansible tasks to find the slowest ones" [Undecided,New] | 16:10 |
| *** phalmos has quit IRC | 16:11 | |
| *** k_stev1 has quit IRC | 16:11 | |
| cloudnull | related PR https://review.openstack.org/#/c/216849/ | 16:11 |
| odyssey4me | mhayden if you had assigned it to yourself then the master review would have been auto-added :) | 16:11 |
| odyssey4me | to me it's a wishlist item, and should only be in gate tests - not run for production environments | 16:12 |
| cloudnull | I like the use case and change that said i can understand the desire to keep unneeded bits out of the repo | 16:12 |
| cloudnull | if some of the other cores can weigh in on it it would be great | 16:12 |
| odyssey4me | it can also make some of our scripts simpler, so I'm game for the change | 16:12 |
| cloudnull | next https://bugs.launchpad.net/openstack-ansible/+bug/1488650 | 16:12 |
| openstack | Launchpad bug 1488650 in openstack-ansible "Should be able to specify account names in user_variables" [Undecided,New] | 16:12 |
| cloudnull | this is something that we have in kilo and beyond | 16:13 |
| cloudnull | the question is do we need / want it for juno | 16:13 |
| odyssey4me | I am not a fan of backporting anything to Juno, personally. | 16:14 |
| cloudnull | i say no. being that juno goes sec patch only in about two months. | 16:14 |
| odyssey4me | any features I mean | 16:14 |
| odyssey4me | real bugs, sure - but not features | 16:14 |
| *** phalmos has joined #openstack-ansible | 16:14 | |
| cloudnull | im of the oppinion that if thats a feature needed, kilo is where its at | 16:14 |
| odyssey4me | yup, kilo's been out for ages - it's time that those that use juno upgrade to it | 16:15 |
| palendae | odyssey4me: Keep in mind Kilo upgrades are not simple | 16:15 |
| palendae | So I see why people would want it in Juno | 16:15 |
| palendae | That said, we can't keep taking features back | 16:15 |
| odyssey4me | palendae if they're on juno, then they are already running in the way juno was designed | 16:15 |
| palendae | Sure | 16:15 |
| odyssey4me | it's pointless to change that way at this late stage | 16:15 |
| palendae | Agreed | 16:16 |
| cloudnull | Agreed. | 16:16 |
| *** jaypipes has quit IRC | 16:16 | |
| palendae | Just saying that waving it off as "upgrade to Kilo" is complicated | 16:16 |
| odyssey4me | if people are having trouble with upgrades, the issues with the upgrades or the changes made in kilo need to be adressed - rather than trying to make juno work like kilo. | 16:16 |
| palendae | Agreed | 16:17 |
| palendae | So I think that bug needs to have a link to the change in Kilo with that explanation | 16:17 |
| cloudnull | ok . | 16:18 |
| *** jmckind has joined #openstack-ansible | 16:18 | |
| cloudnull | updated | 16:21 |
| cloudnull | next https://bugs.launchpad.net/openstack-ansible/+bug/1488670 | 16:21 |
| openstack | Launchpad bug 1488670 in openstack-ansible "configure_diskspace will attempt to use small partitions" [Undecided,New] | 16:21 |
| *** sdake_ is now known as sdake | 16:21 | |
| odyssey4me | there's a review in for that | 16:21 |
| palendae | Thanks | 16:22 |
| cloudnull | related https://review.openstack.org/#/c/216880/ | 16:22 |
| odyssey4me | yup, that's it | 16:22 |
| odyssey4me | some discussion there - I agree that we need to do better, but I'm thinking that the current method is a little too exclusionary | 16:23 |
| odyssey4me | we've suggested a smaller limit or something | 16:23 |
| odyssey4me | anyway, in terms of bug this is assigned to neill and is low priority | 16:23 |
| *** javeriak_ has joined #openstack-ansible | 16:24 | |
| cloudnull | bug updated with context, assigned ot neil, and marked triaged . | 16:24 |
| cloudnull | next https://bugs.launchpad.net/openstack-ansible/+bug/1489169 | 16:24 |
| openstack | Launchpad bug 1489169 in openstack-ansible "Speed up deployment by upgrading cached container image" [Undecided,New] | 16:24 |
| *** javeriak has quit IRC | 16:25 | |
| odyssey4me | I like the idea | 16:26 |
| odyssey4me | wish list | 16:26 |
| odyssey4me | assign to mhayden | 16:26 |
| cloudnull | kk | 16:26 |
| odyssey4me | but a cautionary note - putting it into the play would probably make it update the cache every time? | 16:26 |
| odyssey4me | how does that affect already running containers | 16:27 |
| odyssey4me | and how does that affect consistency when adding new containers later on | 16:27 |
| *** jaypipes has joined #openstack-ansible | 16:27 | |
| cloudnull | +1 i like the approach and I think we can iterate on further improving that process for more speed gains. | 16:28 |
| cloudnull | we can drop a when clause to only do it for new container creates. | 16:28 |
| cloudnull | or if a variable is specifically specified | 16:29 |
| evrardjp | that would be even better | 16:29 |
| palendae | Yeah | 16:29 |
| cloudnull | +1 odyssey4me can you drop a not e in that bug to that effect? | 16:29 |
| evrardjp | by default, update (old process), and skip if variable is set | 16:29 |
| odyssey4me | cloudnull sure | 16:30 |
| palendae | I know the apt module has a conditional you can add to only update things if it's been >n seconds since last update, but this looks like it's just using a command | 16:30 |
| evrardjp | this is for the whole container thing, this is even better | 16:31 |
| evrardjp | why not think also of one update and then a distribution of the updated container template to all physical hosts? | 16:31 |
| *** cloudtrainme has joined #openstack-ansible | 16:32 | |
| cloudnull | yea theres an upstream bug in that module which makes it so the update doesn't "always" happen as one would assume. we have a fix upstream to address that but its currently in review https://github.com/ansible/ansible-modules-core/pull/1517 | 16:32 |
| cloudnull | palendae: ^ re apt module. | 16:32 |
| palendae | Ah, ok | 16:32 |
| palendae | That was really a tangent | 16:32 |
| cloudnull | evrardjp: thats also a good idea . | 16:32 |
| cloudnull | odyssey4me: and andymccr have been pondering how to do tha t | 16:32 |
| cloudnull | maybe you three can come to gether to work out a solution ? | 16:33 |
| evrardjp | why not :) | 16:33 |
| cloudnull | i think that would be ideal in the end to be able to distribute an updated image internally using something like the repo container infrastrucutre however there are some logistical issues with that need to be address. | 16:34 |
| odyssey4me | Added this note: It may be useful to download the cache to the deployment host (and if the file is there, don't download it again), then update it and distribute it to the rest of the hosts from there. That may cater for all needs. | 16:34 |
| cloudnull | and with a meeting of the minds antonym odyssey4me evrardjp we can make it all go :) | 16:34 |
| palendae | I think you mean andymccr there :) | 16:35 |
| cloudnull | s/antonym/ andymccr/ | 16:35 |
| palendae | Channel's getting bigger, our tab completion strings are getting longer | 16:35 |
| cloudnull | indeed | 16:35 |
| cloudnull | :) | 16:35 |
| cloudnull | next https://bugs.launchpad.net/openstack-ansible/+bug/1489251 | 16:35 |
| openstack | Launchpad bug 1489251 in openstack-ansible "Wrong pip package versions after upgrade" [Undecided,New] | 16:36 |
| palendae | Looks like this is Juno -> Juno minor versions | 16:36 |
| palendae | But may be relevant to any upgrades | 16:36 |
| *** Mudpuppy has quit IRC | 16:37 | |
| cloudnull | this is an issue with us needing to downgrade a package due to upstream pin changes. | 16:37 |
| prometheanfire | do we not support downgrades? | 16:37 |
| cloudnull | sadly pip does not have a good resolver to make that a reality without some manual intervention . | 16:37 |
| cloudnull | sigmavirus24: cc ^ | 16:37 |
| palendae | yeah, pip doesn't work very well there.. | 16:38 |
| odyssey4me | this only affects the deployment host though | 16:38 |
| cloudnull | ++ | 16:38 |
| palendae | odyssey4me: With these specific packages, but it may very well be a general problem, no? | 16:38 |
| cloudnull | on install of a given service pip will install all of the packages that are required. which will do a downgrade in general. | 16:39 |
| odyssey4me | well, it will affect how the repo server builds out the final package list | 16:39 |
| odyssey4me | other than that, requirements.txt is only used by bootstrap-ansible as I recall | 16:40 |
| palendae | Yeah, true - though openstack services do tend to shift dependencies back when problems are found | 16:40 |
| cloudnull | if im not mistaken pyyaml would be replaced if nova was reinstalled in this case. | 16:40 |
| palendae | And we don't clean it up | 16:40 |
| *** KLevenstein_ has joined #openstack-ansible | 16:42 | |
| prometheanfire | easier to do a full regen every run? | 16:42 |
| *** KLevenstein has quit IRC | 16:42 | |
| *** KLevenstein_ is now known as KLevenstein | 16:42 | |
| prometheanfire | recompile the deps? | 16:42 |
| odyssey4me | prometheanfire that is not possible for juno | 16:42 |
| prometheanfire | ah | 16:42 |
| odyssey4me | juno's repo is built upstream and is done per tag | 16:42 |
| *** Mudpuppy has joined #openstack-ansible | 16:43 | |
| odyssey4me | the bug in question doesn't really say that there's any serious problem other than inconsistency | 16:43 |
| palendae | So we should ask if there is any other serious problem | 16:44 |
| odyssey4me | yeah - if this doesn't affect any running services, then what's the problem? | 16:45 |
| *** Bjoern_ has joined #openstack-ansible | 16:45 | |
| palendae | I asked | 16:45 |
| palendae | In a comment | 16:45 |
| odyssey4me | otherwise it sounds like a low priority issue | 16:45 |
| palendae | Can be for now, until we get a response | 16:46 |
| cloudnull | ok marked incomplete | 16:46 |
| palendae | ^ | 16:46 |
| prometheanfire | next? | 16:46 |
| cloudnull | next https://bugs.launchpad.net/openstack-ansible/+bug/1489451 | 16:47 |
| openstack | Launchpad bug 1489451 in openstack-ansible "Not able to connect memcached_container while running setup-infrastructure.yml " [Undecided,New] | 16:47 |
| cloudnull | i think this is an issue with the key distribution bits that is being replaced by the process developed by andymccr. | 16:49 |
| odyssey4me | oh? I thought that python was already installed in the image? | 16:49 |
| cloudnull | it is. | 16:50 |
| odyssey4me | and why is this not happening in our gate checks? | 16:50 |
| cloudnull | it is on contianer create. | 16:51 |
| cloudnull | https://github.com/stackforge/os-ansible-deployment/blob/kilo/playbooks/roles/lxc_container_create/tasks/container_create.yml#L256-L266 | 16:51 |
| odyssey4me | he suggested using just 'python' as the package name instead of 'python2.7' | 16:52 |
| odyssey4me | maybe this relates to an old mirror or something? | 16:52 |
| cloudnull | maybe... im pondering and dont really know how all that happened. | 16:53 |
| javeriak_ | hey guys, somewhat off-topic, but I'm trying to upload a glance image with cloudfiles storage as the backend, the image is 1G, and forever stuck on saving, im guessing some glance size parameters need to be changed? | 16:53 |
| evrardjp | upload from url or directly? connection speed? what's forever? | 16:54 |
| evrardjp | did you try with the progress bar using the cli? | 16:54 |
| palendae | afk a few, sleeping laptop while I move | 16:55 |
| *** cloudtrainme has quit IRC | 16:55 | |
| cloudnull | javeriak_: what evrardjp said would be of help. | 16:55 |
| javeriak_ | evrardjp upload through cli, and nope didnt try with progress bar so all i see is 'Saving' | 16:56 |
| cloudnull | in that way you'd have a bit more upload stats | 16:56 |
| cloudnull | we're done with the general bug triage . | 16:56 |
| cloudnull | thanks everyone | 16:56 |
| Bjoern_ | javariak_: glance_container-1fec626a:~# grep 5120 /etc/glance/glance-api.conf | 16:56 |
| Bjoern_ | swift_store_large_object_size = 5120 | 16:56 |
| evrardjp | javeriak_: using glance cli use --progress | 16:57 |
| Bjoern_ | look inside the glance-registry.log for errors and turn on debugging so you would see how it uses the swift client | 16:57 |
| evrardjp | I'm off for today | 16:57 |
| evrardjp | see you tomorrow, hope you'll find your solution javeriak_ | 16:57 |
| javeriak_ | Bjoern_ yes thats 5120 and no errors i checked all logs, evrardjp let me give that a try thanks | 16:58 |
| *** Bjoern_ is now known as BjoernT | 16:58 | |
| *** sdake_ has joined #openstack-ansible | 17:06 | |
| *** sdake has quit IRC | 17:06 | |
| *** sdake has joined #openstack-ansible | 17:07 | |
| BjoernT | javeriak_: look inside the image_location table from the glance database and verify that the swift url is currect. Did you set swift_store_endpoint_type to internalURL or publicURL ? | 17:10 |
| *** sdake_ has quit IRC | 17:10 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: [WIP] Temporarily pin netaddr to not use 0.7.16 https://review.openstack.org/219364 | 17:11 |
| *** k_stev1 has joined #openstack-ansible | 17:11 | |
| *** k_stev has quit IRC | 17:12 | |
| javeriak_ | BjoernT, the swift backend settings are fine, ive had success uploading cirros images, the custom image im trying with atm is just too big i suppose, and the connectivity also seems to be crap, the progress bar is moving snail pace, so atleast now i wont get impatient and terminate it | 17:14 |
| javeriak_ | BjoernT: to answer your question, internalURL wouldn't work so I set it to publicURL | 17:15 |
| BjoernT | lol, yeah I always run tcpump from a glance container to see if stuff is still trickling into to cloudfiles which is rather slow. Yes I was hoping you had set it to publicUrl | 17:15 |
| *** k_stev1 has quit IRC | 17:16 | |
| *** abitha has joined #openstack-ansible | 17:37 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 17:54 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: [WIP] Temporarily pin netaddr to not use 0.7.16 https://review.openstack.org/219385 | 18:04 |
| *** javeriak_ has quit IRC | 18:06 | |
| *** cloudtrainme has joined #openstack-ansible | 18:12 | |
| *** javeriak has joined #openstack-ansible | 18:19 | |
| *** javeriak_ has joined #openstack-ansible | 18:22 | |
| *** javeriak has quit IRC | 18:24 | |
| *** k_stev has joined #openstack-ansible | 18:27 | |
| *** jwagner_away is now known as jwagner | 18:30 | |
| *** jwagner is now known as jwagner_away | 18:41 | |
| *** cloudtrainme has quit IRC | 18:43 | |
| mhayden | odyssey4me: whee! | 18:43 |
| *** jwagner_away is now known as jwagner | 18:49 | |
| odyssey4me | mhayden haha | 18:56 |
| * odyssey4me shakes a fist at random library changes at inconvenient times | 18:56 | |
| mhayden | odyssey4me: i feel all open sourcey | 18:57 |
| *** javeriak has joined #openstack-ansible | 18:57 | |
| *** javeriak_ has quit IRC | 18:57 | |
| *** k_stev has quit IRC | 18:59 | |
| *** woodard has joined #openstack-ansible | 19:00 | |
| *** BjoernT has quit IRC | 19:02 | |
| *** gparaskevas has joined #openstack-ansible | 19:03 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 19:04 | |
| *** javeriak_ has joined #openstack-ansible | 19:05 | |
| *** javeriak has quit IRC | 19:06 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: [WIP] Temporarily pin netaddr to use <0.7.16 https://review.openstack.org/219364 | 19:07 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: [WIP] Temporarily pin netaddr to use <0.7.16 https://review.openstack.org/219385 | 19:08 |
| *** sdake_ has joined #openstack-ansible | 19:12 | |
| *** cloudtrainme has joined #openstack-ansible | 19:12 | |
| *** sdake has quit IRC | 19:13 | |
| *** sdake has joined #openstack-ansible | 19:13 | |
| *** cloudtrainme has quit IRC | 19:14 | |
| *** sdake_ has quit IRC | 19:18 | |
| *** cloudtrainme has joined #openstack-ansible | 19:18 | |
| *** scarlisle has quit IRC | 19:34 | |
| *** mcarden has quit IRC | 19:34 | |
| *** tobasco_ has quit IRC | 19:34 | |
| *** mrodden has quit IRC | 19:34 | |
| *** raginbajin has quit IRC | 19:34 | |
| *** harlowja has quit IRC | 19:34 | |
| *** Sam-I-Am has quit IRC | 19:34 | |
| *** jmckind has quit IRC | 19:34 | |
| *** b3rnard0 has quit IRC | 19:34 | |
| *** bogeyon18 has quit IRC | 19:34 | |
| *** cloudnull has quit IRC | 19:34 | |
| *** jwagner has quit IRC | 19:34 | |
| *** b3rnard0 has joined #openstack-ansible | 19:34 | |
| *** bogeyon18 has joined #openstack-ansible | 19:34 | |
| *** cloudnull has joined #openstack-ansible | 19:34 | |
| *** jwagner has joined #openstack-ansible | 19:34 | |
| *** k_stev has joined #openstack-ansible | 19:34 | |
| *** scarlisle has joined #openstack-ansible | 19:34 | |
| *** mcarden has joined #openstack-ansible | 19:34 | |
| *** tobasco_ has joined #openstack-ansible | 19:34 | |
| *** mrodden has joined #openstack-ansible | 19:34 | |
| *** raginbajin has joined #openstack-ansible | 19:34 | |
| *** Sam-I-Am has joined #openstack-ansible | 19:34 | |
| *** harlowja has joined #openstack-ansible | 19:35 | |
| *** harlowja has quit IRC | 19:35 | |
| *** harlowja has joined #openstack-ansible | 19:35 | |
| *** abitha has quit IRC | 19:37 | |
| *** jmckind has joined #openstack-ansible | 19:39 | |
| *** sdake_ has joined #openstack-ansible | 19:45 | |
| *** harlowja has quit IRC | 19:47 | |
| *** k_stev has quit IRC | 19:47 | |
| *** scarlisle has quit IRC | 19:47 | |
| *** mcarden has quit IRC | 19:47 | |
| *** tobasco_ has quit IRC | 19:47 | |
| *** mrodden has quit IRC | 19:47 | |
| *** raginbajin has quit IRC | 19:47 | |
| *** Sam-I-Am has quit IRC | 19:47 | |
| *** sdake has quit IRC | 19:49 | |
| *** sdake has joined #openstack-ansible | 19:49 | |
| *** harlowja has joined #openstack-ansible | 19:52 | |
| *** k_stev has joined #openstack-ansible | 19:52 | |
| *** scarlisle has joined #openstack-ansible | 19:52 | |
| *** mcarden has joined #openstack-ansible | 19:52 | |
| *** tobasco_ has joined #openstack-ansible | 19:52 | |
| *** mrodden has joined #openstack-ansible | 19:52 | |
| *** raginbajin has joined #openstack-ansible | 19:52 | |
| *** Sam-I-Am has joined #openstack-ansible | 19:52 | |
| *** sdake_ has quit IRC | 19:53 | |
| *** javeriak has joined #openstack-ansible | 20:01 | |
| *** javeriak_ has quit IRC | 20:02 | |
| *** cloudtrainme has quit IRC | 20:12 | |
| *** javeriak_ has joined #openstack-ansible | 20:25 | |
| *** javeriak has quit IRC | 20:26 | |
| *** cloudtrainme has joined #openstack-ansible | 20:28 | |
| *** jwagner is now known as jwagner_away | 20:34 | |
| *** alextricity_h has quit IRC | 20:34 | |
| *** andymccr has quit IRC | 20:39 | |
| coolj | cloudnull: https://bugs.launchpad.net/openstack-ansible/+bug/1489225 | 20:41 |
| openstack | Launchpad bug 1489225 in openstack-ansible "Allow configuration of config_drive_format in nova.conf" [Undecided,New] | 20:41 |
| *** afranc202 has joined #openstack-ansible | 20:46 | |
| mattt | coolj: why do you need to use config drive (honest question) | 20:48 |
| *** KLevenstein has quit IRC | 20:49 | |
| *** KLevenstein has joined #openstack-ansible | 20:49 | |
| *** k_stev has quit IRC | 20:50 | |
| *** sdake has quit IRC | 20:53 | |
| *** k_stev has joined #openstack-ansible | 20:59 | |
| *** k_stev has quit IRC | 21:09 | |
| coolj | mattt: we don't *need* it per se, but we have had issues with api saturation from customers using puppet on a bunch of instances to poll metadata | 21:10 |
| mattt | coolj: ah, i see! | 21:10 |
| mattt | coolj: wonder if that was due to a misconfigured neutron server (not enough threads/workers?) | 21:11 |
| *** pradk has quit IRC | 21:11 | |
| mattt | (and not server per se, but neutron metadata server) | 21:11 |
| *** pradk has joined #openstack-ansible | 21:12 | |
| coolj | mattt: i can find the tickets to see exactly what we tuned but as i recall we tried upping api_workers any a bunch of other stuff but the bottleneck was that neutron-metadata-proxy was a single thread/process. i might be misremembering the exact details | 21:14 |
| *** gparaskevas has quit IRC | 21:15 | |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment-specs: Adding spec for named veth interfaces https://review.openstack.org/219105 | 21:16 |
| mattt | coolj: we can look into it, there may be no other solution but to enable config drive, but it may be worth looking into in more detail | 21:18 |
| coolj | mattt: what is the downside of enabling config-drive if we can also use vfat for the drive? afaik live-migration was the only reason to disable and that works as long as vfat format is used instead of iso9660 | 21:20 |
| *** javeriak_ has quit IRC | 21:21 | |
| mattt | coolj: tbh i'm not really aware of any | 21:21 |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment: Used named veth pairs that match container https://review.openstack.org/219457 | 21:21 |
| coolj | cool, just wanted to be aware if there are further issues with having it enabled | 21:23 |
| mhayden | cloudnull: bumped the named-veths spec and tossed in a WIP commit | 21:24 |
| mattt | coolj: oh, does vfat work on non-linux images ? | 21:28 |
| *** tlian2 has joined #openstack-ansible | 21:31 | |
| *** tlian has quit IRC | 21:34 | |
| coolj | mattt: good question! i can test! since it's fat32 i suspect 99% of our instances will have no problem with it (mostly linux with a sprinkling of windows), but i can testopenindiana and freebsd. any others you can think of to test? | 21:36 |
| *** alop has quit IRC | 21:37 | |
| *** Mudpuppy has quit IRC | 21:39 | |
| *** jmckind has quit IRC | 21:52 | |
| *** phalmos has quit IRC | 21:52 | |
| mattt | coolj: may be worht testing it on windows also :P | 21:55 |
| *** Bjoern_ has joined #openstack-ansible | 22:08 | |
| *** galstrom is now known as galstrom_zzz | 22:11 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 22:18 | |
| *** scarlisle has quit IRC | 22:26 | |
| *** jaypipes has quit IRC | 22:40 | |
| *** alejandrito_ has joined #openstack-ansible | 22:43 | |
| *** alejandrito_ has quit IRC | 22:45 | |
| *** alejandrito has quit IRC | 22:45 | |
| *** spotz is now known as spotz_zzz | 22:49 | |
| *** arbrandes has quit IRC | 22:50 | |
| *** andymccr has joined #openstack-ansible | 22:57 | |
| *** KLevenstein has quit IRC | 23:00 | |
| *** cloudtrainme has quit IRC | 23:03 | |
| *** arbrandes has joined #openstack-ansible | 23:03 | |
| *** Bjoern_ has quit IRC | 23:47 | |
| *** shoutm has joined #openstack-ansible | 23:48 | |
| *** woodard has quit IRC | 23:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!