| *** vnogin has joined #openstack-ansible | 00:03 | |
| *** vnogin has quit IRC | 00:08 | |
| *** odyssey4me has quit IRC | 00:23 | |
| *** odyssey4me has joined #openstack-ansible | 00:23 | |
| mnaser | gr | 01:14 |
|---|---|---|
| mnaser | anyone know of a reason why the rsync stuff in containers isnt working | 01:14 |
| mnaser | or wouldnt work | 01:14 |
| mnaser | or i guess it just takes a little while for the rsync to finish | 01:16 |
| mnaser | and there is no ansible task to wait for it:\ | 01:16 |
| *** vincent has joined #openstack-ansible | 01:28 | |
| *** vincent is now known as Guest85774 | 01:28 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible stable/queens: Install mon servers in parallel. https://review.openstack.org/559610 | 02:27 |
| *** markvoelker has quit IRC | 02:28 | |
| *** markvoelker has joined #openstack-ansible | 02:29 | |
| *** markvoelker has quit IRC | 02:34 | |
| *** chhagarw has joined #openstack-ansible | 04:07 | |
| *** udesale has joined #openstack-ansible | 04:17 | |
| *** vnogin has joined #openstack-ansible | 05:18 | |
| *** vnogin has quit IRC | 05:23 | |
| *** Guest85774 has quit IRC | 05:27 | |
| *** udesale_ has joined #openstack-ansible | 05:30 | |
| *** vincent has joined #openstack-ansible | 05:30 | |
| *** vincent is now known as Guest53306 | 05:31 | |
| *** sar has quit IRC | 05:32 | |
| *** udesale has quit IRC | 05:33 | |
| *** radeks has joined #openstack-ansible | 05:56 | |
| *** jbadiapa has joined #openstack-ansible | 06:00 | |
| *** Sha000000 has joined #openstack-ansible | 06:05 | |
| *** Sha000000 has quit IRC | 06:07 | |
| *** sar has joined #openstack-ansible | 06:19 | |
| *** eumel8 has joined #openstack-ansible | 06:25 | |
| *** armaan has joined #openstack-ansible | 06:32 | |
| *** pcaruana has joined #openstack-ansible | 06:35 | |
| *** armaan has quit IRC | 06:41 | |
| *** armaan has joined #openstack-ansible | 06:41 | |
| *** oanson has quit IRC | 06:42 | |
| *** oanson has joined #openstack-ansible | 06:44 | |
| *** udesale__ has joined #openstack-ansible | 06:46 | |
| *** udesale_ has quit IRC | 06:49 | |
| *** gokhan has quit IRC | 06:55 | |
| evrardjp | good morning | 06:57 |
| evrardjp | mnaser: which rsync? | 06:57 |
| evrardjp | if it's between repo nodes, yes it takes time | 06:58 |
| evrardjp | we should probably only enable the LB backends when they are in sync | 06:58 |
| evrardjp | but this hasn't received any love | 06:58 |
| evrardjp | people just wait usually :) | 06:58 |
| *** gokhan has joined #openstack-ansible | 06:58 | |
| evrardjp | or force the sync | 06:58 |
| niraj_singh | evrardjp: good morning | 06:59 |
| niraj_singh | i have a question that where it is mentioned that neutron agents will be on compute nodes. | 06:59 |
| niraj_singh | https://github.com/openstack/openstack-ansible/blob/master/inventory/env.d/neutron.yml | 06:59 |
| niraj_singh | i want to understand this part | 07:00 |
| evrardjp | https://docs.openstack.org/openstack-ansible/latest/reference/inventory/inventory.html | 07:01 |
| evrardjp | this will get refactored soon though | 07:01 |
| evrardjp | but have a look already | 07:01 |
| *** epalper has joined #openstack-ansible | 07:08 | |
| *** rodolof has joined #openstack-ansible | 07:08 | |
| niraj_singh | evrardjp: http://paste.openstack.org/show/718709/ have a look on this configuration please. is it correct? | 07:18 |
| *** pradiprwt has joined #openstack-ansible | 07:28 | |
| pradiprwt | Hi Everyone, I want to install multi region openstack using ansible, can anyone please suggest me how to achieve this or any references. | 07:30 |
| *** radeks has quit IRC | 07:31 | |
| *** radeks has joined #openstack-ansible | 07:31 | |
| *** electrofelix has joined #openstack-ansible | 07:41 | |
| *** DanyC has joined #openstack-ansible | 07:42 | |
| *** DanyC has quit IRC | 07:47 | |
| *** mathlin has joined #openstack-ansible | 07:47 | |
| odyssey4me | evrardjp the repo servers are taken out of the lb when they are busy syncing, that was implemented ages ago | 07:48 |
| *** radeks has quit IRC | 07:50 | |
| *** radeks has joined #openstack-ansible | 07:51 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_tempest master: [WIP] Overriding default timeouts https://review.openstack.org/557896 | 07:58 |
| evrardjp | odyssey4me: oh I thought we didn't do it :) | 08:04 |
| evrardjp | cool | 08:04 |
| evrardjp | I might have found the tempest timeout issue but I'm not 100% sure because I don't know tempest | 08:05 |
| *** shardy has joined #openstack-ansible | 08:05 | |
| *** shardy has quit IRC | 08:05 | |
| evrardjp | so let's see how my last patch goes | 08:05 |
| evrardjp | it's just a docs issue | 08:05 |
| *** shardy has joined #openstack-ansible | 08:06 | |
| evrardjp | if this is working as expected, I will change the default timeouts, to be more reliable | 08:06 |
| evrardjp | in the last two cases I see, the instructions for the teardown were launched, and it's during the teardown process that it timed out, not even during the build | 08:06 |
| *** gameon has joined #openstack-ansible | 08:17 | |
| evrardjp | hwoarang is not here today -- he is usually early, I wonder if he is alright | 08:17 |
| evrardjp | odyssey4me: can I get this to merge? https://review.openstack.org/#/c/552081/ it's been pending for a while | 08:24 |
| evrardjp | andymccr: ^ | 08:24 |
| pradiprwt | evrardjp : Is there deployment guide for multi region openstack ? | 08:27 |
| evrardjp | nope there is none. But there is a guide for swift multi region I think | 08:29 |
| evrardjp | It solely depends on what you want to do | 08:29 |
| pradiprwt | If I want to achieve this what are things need to do manually ? | 08:31 |
| evrardjp | I don't know what you want to achieve | 08:32 |
| evrardjp | :) | 08:32 |
| evrardjp | "multi region" is so vast | 08:32 |
| evrardjp | do you want one keystone, one galera cluster, multiple compute cells? | 08:32 |
| evrardjp | or is it multiple federated keystones? | 08:33 |
| evrardjp | or is that multiple rabbit/galera clusters? | 08:33 |
| evrardjp | this isn't clear to me | 08:33 |
| pradiprwt | Yeah similar | 08:33 |
| evrardjp | well | 08:33 |
| evrardjp | it must be clear to you in order for us to help you | 08:33 |
| evrardjp | we cannot presume your design, your networking, etc. | 08:33 |
| evrardjp | it solely depends on your needs, your team skills, etc. | 08:34 |
| pradiprwt | 2 different openstack environment managing by single horizon | 08:34 |
| *** DanyC has joined #openstack-ansible | 08:36 | |
| *** shardy has quit IRC | 08:36 | |
| evrardjp | pradiprwt: I cannot answer that question. | 08:37 |
| evrardjp | this isn't "simply that" | 08:37 |
| *** shardy has joined #openstack-ansible | 08:37 | |
| evrardjp | or I don't understand the question | 08:37 |
| evrardjp | yes it looks possible with openstack-ansible | 08:37 |
| pradiprwt | evrardjp : Thank you, I am working on this .. I will let you with more details if I need any help | 08:39 |
| *** DanyC has quit IRC | 08:40 | |
| evrardjp | pradiprwt: you need to know networking constraints mostly | 08:46 |
| evrardjp | and what you really mean by "two different" openstack | 08:46 |
| evrardjp | if you really want two different openstacks, just deploy it twice :) | 08:46 |
| evrardjp | cloudnull: you know there was no necessity to break builds like mhayden :p You're a worthy successor indeed -- this time its systemd-networkd that breaks suse | 08:50 |
| pradiprwt | evrardjp: yes, I will deploy two environment but I want to access it from single horizon by changing region | 08:50 |
| odyssey4me | pradiprwt how you decide to deploy is up to you - but each region should be implemented using its own inventory | 08:51 |
| pradiprwt | yes | 08:51 |
| *** vnogin has joined #openstack-ansible | 08:52 | |
| odyssey4me | there are many, many ways to implement multi-region - but the clear guideline from us is to ensure that you keep each region in its own inventory | 08:52 |
| odyssey4me | *ansible inventory | 08:52 |
| pradiprwt | Any suggestion for integration after deployment | 08:52 |
| evrardjp | yeah, deploy it twice, remove horizon from the second inventory | 08:53 |
| pradiprwt | ok | 08:53 |
| evrardjp | then ... I don't know | 08:53 |
| pradiprwt | :) | 08:53 |
| evrardjp | probably worth asking other ppl like logan- or simply checking upstream documentation | 08:53 |
| odyssey4me | pradiprwt I don't understand your question. | 08:54 |
| odyssey4me | Once you have multiple regions setup, horizon is informed about the other regions from the service catalog. | 08:54 |
| odyssey4me | So you just add the other region endpoints to the catalog. | 08:54 |
| odyssey4me | This is a once-off, so you can do it by hand or implement automation if you want. | 08:55 |
| odyssey4me | But it's a trivial action. | 08:55 |
| *** vnogin has quit IRC | 08:56 | |
| mhayden | evrardjp: glad to see cloudnull carrying on my traditions | 09:13 |
| *** mbuil has joined #openstack-ansible | 09:20 | |
| *** gameon has quit IRC | 09:23 | |
| *** armaan has quit IRC | 09:24 | |
| *** mma has joined #openstack-ansible | 09:29 | |
| *** armaan has joined #openstack-ansible | 09:30 | |
| mma | Hey, I notice that the token expiration time is set to 43200 by default in OSA. Looking at the security guide for openstack it says the default is 1 hour and it is recommended to lower it. (https://docs.openstack.org/security-guide/identity/tokens.html). What is the reason for the high expiration value in osa? | 09:31 |
| *** DanyC has joined #openstack-ansible | 09:32 | |
| odyssey4me | mma because snapshots take a while, and especially when using a swift back-end for the storage of images your token lifetime needs to be as long as your largest operation executed using a token | 09:34 |
| Tahvok | odyssey4me: why is this default good? https://review.openstack.org/#/c/549287/ | 09:34 |
| odyssey4me | mma so if your users are snapshotting a 3TB instance, how long does it take to complete | 09:34 |
| odyssey4me | to my knowledge there is not yet a way for the operation to renew its token if it expires while still busy doing the operation | 09:34 |
| Tahvok | odyssey4me: it's the same as this 2 reviews: https://review.openstack.org/#/c/549285/ https://review.openstack.org/#/c/549286/ | 09:35 |
| *** armaan has quit IRC | 09:35 | |
| odyssey4me | Tahvok :/ a role should be able to stand alone - I guess it seems that's not a common consensus - please reply to the comment with that reference and I'll remove my -1 | 09:36 |
| mma | odyssey4me: Got it thanks. Makes sense. Too bad there is no renew token support yet since from a regulatory perspective we are obliged to state reason for each individual case where we do not meet what is declared in the security guide. But that is our headache, at least now we have the reason confirmed. :) | 09:36 |
| mma | odyssey4me: Thanks a lot | 09:36 |
| odyssey4me | mma sure thing | 09:36 |
| odyssey4me | mma obviously you're able to change the default and set your own value - but hopefully now you understand the risk | 09:36 |
| Tahvok | odyssey4me: but localhost default, is pretty bad... Because you do want the user to specify it.. | 09:37 |
| evrardjp | Tahvok: we do | 09:37 |
| evrardjp | The role should be standalone | 09:37 |
| evrardjp | and group vars to do the wiring if need be | 09:37 |
| evrardjp | the amount of group vars should be reduced because there are not many integration points | 09:37 |
| evrardjp | generally it's all the service catalog | 09:37 |
| Tahvok | so perhaps I need to update the other 2 merged reviews and create the localhost default? | 09:37 |
| evrardjp | I am not sure what's your case | 09:38 |
| niraj_singh | evrardjp: any other way so that i can install monitor serivces on all compute host | 09:38 |
| evrardjp | maybe you don't even need localhost. | 09:38 |
| Tahvok | evrardjp: check this: https://review.openstack.org/#/c/549287/ | 09:38 |
| evrardjp | niraj_singh: are you working on the role right now ? | 09:38 |
| evrardjp | niraj_singh: if it's for the role, give a static inventory | 09:38 |
| niraj_singh | yes | 09:38 |
| evrardjp | the env.d is for the integration into the openstack-ansible | 09:38 |
| odyssey4me | Tahvok it doesn't have to be a localhost default, but if we're relying on a variable to be set we should either make sure we check for it or provide a default - right now there'd be an obscure error | 09:38 |
| evrardjp | that can be done later | 09:38 |
| evrardjp | I am busy I can't deal with this right now niraj_singh | 09:39 |
| odyssey4me | evrardjp I guess in this case we can let it slide to progress the reduction of group_vars | 09:39 |
| odyssey4me | Tahvok just reply to my suggestion with reference to the other reviews which have merged and I'll change my vote | 09:39 |
| evrardjp | I don't know I haven't checked that particular case. | 09:39 |
| niraj_singh | evrardjp: np thank you | 09:39 |
| evrardjp | think about role input and outputs, and it should be clear what we need | 09:39 |
| Tahvok | odyssey4me: I can create 2 more reviews for the other 2 that were merged, and do the default thingy | 09:40 |
| evrardjp | all the rest should be removed | 09:40 |
| odyssey4me | I think that we'll very likely change up how some of this stuff is done next cycle when we start to look at roles more like components. | 09:40 |
| evrardjp | not sure what that meant | 09:40 |
| evrardjp | but it would be good indeed to be in sync | 09:40 |
| Tahvok | So I'll add the default, and create 2 more reviews | 09:41 |
| evrardjp | odyssey4me: could I have a review of this please? https://review.openstack.org/#/c/559059/ | 09:43 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible-os_ironic master: Move ironic_glance_api_servers variable into role https://review.openstack.org/549287 | 09:43 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible master: [Docs] Restructure inventory documentation https://review.openstack.org/552081 | 09:44 |
| evrardjp | I want to get the docs done. I am tired of it now. | 09:45 |
| evrardjp | eumel8: are you there? | 09:48 |
| evrardjp | eumel8: Should I build a machine with https://review.openstack.org/#/c/559290/ ? | 09:48 |
| evrardjp | d34dh0r53: if you're looking for places to help there are many: bugs, the gates are failing for all the upgrade jobs, the release process like we discussed. | 09:49 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_heat master: Update paste, policy and rootwrap configurations 2018-03-31 https://review.openstack.org/558077 | 09:53 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible-os_nova master: Add default to nova_glance_api_servers variable https://review.openstack.org/559682 | 09:54 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible-os_cinder master: Add default to cinder_glance_api_servers variable https://review.openstack.org/559684 | 09:57 |
| Tahvok | Done. Please review | 09:57 |
| *** admin0 has joined #openstack-ansible | 09:57 | |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible-os_designate master: Get the adminuri variable from adminurl variable https://review.openstack.org/548353 | 10:07 |
| niraj_singh | getting error while running repo-builed playbook "fatal: [aio1_repo_container-9d49f9a9]: FAILED! => {"attempts": 5, "changed": false, "cmd": "/usr/local/bin/pip2 install -U --isolated --constraint http://172.29.236.100:8181/os-releases/master/ubuntu-16.04-x86_64/requirements_absolute_requirements.txt ", "msg": "\n:stderr: ERROR: You must give at least one requirement to install (see \"pip help install\")\n"}" | 10:12 |
| odyssey4me | niraj_singh that's due to a recent pip update, and that's been solved in later releases - update your OSA checkout and bootstrap ansible again to pull in the most recent role updates | 10:13 |
| admin0 | afternoon | 10:14 |
| niraj_singh | ok | 10:14 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ceph_client master: Adapt the list of ceph packages for SUSE https://review.openstack.org/559688 | 10:15 |
| evrardjp | niraj_singh: the role test should not care about that. | 10:16 |
| evrardjp | you should start step by step. | 10:16 |
| evrardjp | I suggested to first write a spec | 10:17 |
| evrardjp | then get started on the role | 10:17 |
| evrardjp | then we help you integrate it | 10:17 |
| evrardjp | please read this again: https://docs.openstack.org/openstack-ansible/latest/contributor/contribute.html | 10:17 |
| niraj_singh | already submitted the spec https://review.openstack.org/#/c/556756/ | 10:18 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible master: Remove non used variables https://review.openstack.org/549271 | 10:19 |
| evrardjp | niraj_singh: thanks for the update :) | 10:20 |
| evrardjp | niraj_singh: I have posted a comment right now | 10:20 |
| niraj_singh | ok | 10:20 |
| evrardjp | but my link above should help you | 10:21 |
| evrardjp | it explains how to work on a role | 10:21 |
| evrardjp | and it includes a section about testing | 10:21 |
| evrardjp | which should help you | 10:21 |
| evrardjp | then we can think of integration | 10:21 |
| evrardjp | added another comment niraj_singh :) | 10:22 |
| niraj_singh | yes | 10:22 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible master: Fix parameter name for removing compute host https://review.openstack.org/559573 | 10:23 |
| evrardjp | thanks Tahvok ! | 10:27 |
| evrardjp | good catch | 10:27 |
| Tahvok | Yeah, needed to remove a compute host yesterday, and got it :) | 10:27 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible master: Fix parameter name for removing compute host https://review.openstack.org/559573 | 10:31 |
| Tahvok | evrardjp: please review again. There was another place to change it | 10:31 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-tests master: Add Ceph experimental job for roles https://review.openstack.org/559693 | 10:32 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ceph_client master: Add ceph integrated repo testing https://review.openstack.org/559695 | 10:34 |
| evrardjp | Tahvok: oh I missed it in my grep I guess | 10:34 |
| *** nicolasbock has joined #openstack-ansible | 10:34 | |
| evrardjp | odyssey4me: https://review.openstack.org/#/q/topic:experimental_test_should_trigger_ceph_job+(status:open+OR+status:merged) | 10:35 |
| *** savvas has joined #openstack-ansible | 10:35 | |
| evrardjp | should be useful before merging https://review.openstack.org/#/c/559688/ | 10:35 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible master: Make the on metal job voting https://review.openstack.org/559404 | 10:37 |
| evrardjp | odyssey4me: for https://review.openstack.org/#/c/559688/, we said that release notes was for deployer impacting changes -- should that impact them, as its a vars/main.yml change? | 10:41 |
| evrardjp | I can do it if you want | 10:42 |
| odyssey4me | evrardjp it is deployer impacting due to the documentation that was adjusted, so it's not just a vars/main change | 10:43 |
| odyssey4me | it's also a user config change if the user is using ceph for gnocchi | 10:43 |
| evrardjp | I am not sure this should be in the docs tbh | 10:44 |
| evrardjp | I just grepped over everything | 10:44 |
| odyssey4me | evrardjp that linters timeout on the integrated gate is bothering me - I've seen it timeout often when doing the inventory tests... maybe we should increase the timeout to 15 mins? | 10:44 |
| odyssey4me | evrardjp but it is in the docs, and that specifically tells people how to add the gnocchi config, and that method's changing... so yes, it needs to be communicated | 10:45 |
| evrardjp | odyssey4me: yes it's bothering me too now. It's not like the first time I see that. But I thought it was an infra issue | 10:46 |
| evrardjp | Let me check how long our tests take | 10:46 |
| evrardjp | odyssey4me: fair point for the gnocchi case | 10:47 |
| *** exodusftw has quit IRC | 10:48 | |
| evrardjp | odyssey4me: to what I see it's not much that fail during the linting | 10:50 |
| evrardjp | I see git clone failures, or just the whole process thing taking time | 10:51 |
| evrardjp | I will extend to 15 minutes. | 10:51 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-tests master: Extend linters time out https://review.openstack.org/559701 | 10:53 |
| evrardjp | odyssey4me: ^ | 10:53 |
| *** exodusftw has joined #openstack-ansible | 10:54 | |
| odyssey4me | evrardjp I find myself thinking that even 30 mins for linters is fine given that all other tests take longer, typically. | 10:55 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ceph_client master: Adapt the list of ceph packages for SUSE https://review.openstack.org/559688 | 10:56 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ceph_client master: Adapt the list of ceph packages for SUSE https://review.openstack.org/559688 | 10:57 |
| evrardjp | odyssey4me: I don't think we should go that far. I'd rather know what's going on with more data | 10:57 |
| evrardjp | but 15 minutes vs 30 minutes is a moot conversation -- I can go for 30 | 10:58 |
| evrardjp | if you feel the need, just update the patch | 11:00 |
| evrardjp | it's fine for me | 11:00 |
| *** savvas has quit IRC | 11:02 | |
| *** savvas has joined #openstack-ansible | 11:03 | |
| *** savvas has joined #openstack-ansible | 11:03 | |
| *** armaan has joined #openstack-ansible | 11:05 | |
| *** udesale__ has quit IRC | 11:20 | |
| openstackgerrit | Merged openstack/openstack-ansible stable/queens: Added monitoring tag to monitoring user creation task. https://review.openstack.org/559320 | 11:24 |
| openstackgerrit | Merged openstack/openstack-ansible master: Add periodic work for release preparations https://review.openstack.org/559031 | 11:24 |
| *** ansmith has quit IRC | 11:30 | |
| evrardjp | odyssey4me: want to review my pro english skillz? https://review.openstack.org/#/c/559688/3/releasenotes/notes/renamed_variable_python_ceph_package-9575466eb146e500.yaml | 11:30 |
| evrardjp | :p | 11:30 |
| admin0 | its not used for convenience is it evrardjp ? | 11:34 |
| evrardjp | some ppl might use it for convenience | 11:34 |
| evrardjp | some ppl might just define their own thing | 11:34 |
| odyssey4me | admin0 agreed, those words are superfluous | 11:35 |
| admin0 | if me, i would remove the " for convenience" | 11:35 |
| odyssey4me | admin0 yep, I added that to the review | 11:37 |
| odyssey4me | evrardjp was there ever a spec for the hyperconvergance? | 11:37 |
| Tahvok | Is limestone test mandatory? | 11:37 |
| odyssey4me | ah https://github.com/openstack/openstack-ansible-specs/blob/master/specs/queens/hyperconverged-containers.rst | 11:37 |
| admin0 | Tahvok, you can use charcoal :D | 11:37 |
| odyssey4me | Tahvok no, it's external CI | 11:37 |
| admin0 | sorry :D .. | 11:37 |
| evrardjp | odyssey4me: there was | 11:38 |
| Tahvok | Shouldn't this be merged then? https://review.openstack.org/#/c/559573/ | 11:38 |
| odyssey4me | Tahvok also, it's been failing generally lately - I think logan-'s been wanting to get back to figuring out why, but hasn't had the time | 11:38 |
| Tahvok | How much code reviews needed? | 11:38 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ceph_client master: Adapt the list of ceph packages for SUSE https://review.openstack.org/559688 | 11:39 |
| evrardjp | the rule is 2 +2 | 11:39 |
| evrardjp | odyssey4me: and I should do the trick for this patch for example :) | 11:40 |
| Tahvok | Oh, ok | 11:41 |
| Tahvok | Is there a way to set varibles for a group only? Such as storage_hosts or compute_hosts specific variables? | 11:42 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Use production like interfaces in the gate https://review.openstack.org/553636 | 11:42 |
| Tahvok | I want to override the configuration of rsyslog for this groups, but only on this groups | 11:42 |
| odyssey4me | Tahvok yes, using group_vars | 11:43 |
| odyssey4me | you can create a folder called /etc/openstack_deploy/group_vars and use it like any standard group_vars | 11:44 |
| Tahvok | will work for ocata as well? | 11:44 |
| odyssey4me | yes | 11:45 |
| Tahvok | Thanks! | 11:45 |
| *** kysse_ is now known as kysse | 11:48 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_sahara stable/queens: Update paste, policy and rootwrap configurations 2018-03-31 https://review.openstack.org/558087 | 11:49 |
| *** vakuznet has quit IRC | 11:54 | |
| openstackgerrit | Merged openstack/openstack-ansible-tests master: Add Ceph experimental job for roles https://review.openstack.org/559693 | 12:05 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible-os_cinder master: Add default to cinder_glance_api_servers variable https://review.openstack.org/559684 | 12:07 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible-os_nova master: Add default to nova_glance_api_servers variable https://review.openstack.org/559682 | 12:07 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible-os_ironic master: Move ironic_glance_api_servers variable into role https://review.openstack.org/549287 | 12:10 |
| *** hwoarang has joined #openstack-ansible | 12:12 | |
| *** vakuznet has joined #openstack-ansible | 12:15 | |
| Tahvok | odyssey4me: I wonder how to proceed with your comment on octavia variable. | 12:15 |
| Tahvok | I've taken it from here: https://review.openstack.org/#/c/549271/2/inventory/group_vars/all/octavia.yml | 12:15 |
| vakuznet | cloudnull | 12:17 |
| Tahvok | # this var below is referenced by neutron if Octavia is activated with V1 API | 12:17 |
| Tahvok | But looking at octavia, it's not defined in there as well | 12:17 |
| Tahvok | Should I introduce this variable in octavia? | 12:18 |
| Tahvok | odyssey4me: anyway, answered you here: https://review.openstack.org/#/c/549267/1/templates/neutron.conf.j2 | 12:20 |
| odyssey4me | Tahvok it looks to me like it's always referring back to the same URI, so I don't see why there needs to be more than one variable there? | 12:25 |
| odyssey4me | Is there ever a difference between the v1 URI and the v2 URI? | 12:25 |
| Tahvok | I don't know, but there is even a comment on this in inventory | 12:26 |
| Tahvok | So I didn't want to break anything by accident | 12:26 |
| evrardjp | vakuznet: just FYI a ping without comment is not really helpful | 12:26 |
| evrardjp | ask what you need to ask and other ppl might answer. | 12:27 |
| vakuznet | yeah, sorry | 12:27 |
| evrardjp | or at least don't expect me to answer if you do that to me :p | 12:27 |
| *** udesale has joined #openstack-ansible | 12:28 | |
| vakuznet | could this bug be revisited https://bugs.launchpad.net/openstack-ansible/+bug/1742206 fix applied to galere_server, which was not broken and nothing was done to galera_client, which is subject of the bug | 12:30 |
| openstack | Launchpad bug 1742206 in openstack-ansible "os-cinder-install.yml fails with conflicting mariadb packages on centos infra hosts" [Medium,Fix released] - Assigned to Markos Chandras (hwoarang) | 12:30 |
| openstackgerrit | Merged openstack/openstack-ansible-tests master: Extend linters time out https://review.openstack.org/559701 | 12:30 |
| *** savvas has quit IRC | 12:31 | |
| *** udesale has quit IRC | 12:31 | |
| *** udesale has joined #openstack-ansible | 12:32 | |
| *** nicolasbock has quit IRC | 12:33 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: [WIP] Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 12:35 |
| *** armaan has quit IRC | 12:36 | |
| *** armaan has joined #openstack-ansible | 12:36 | |
| *** nicolasbock has joined #openstack-ansible | 12:37 | |
| *** savvas has joined #openstack-ansible | 12:38 | |
| *** odyssey4me has quit IRC | 12:38 | |
| *** odyssey4me has joined #openstack-ansible | 12:38 | |
| *** dave-mccowan has joined #openstack-ansible | 12:41 | |
| *** savvas has quit IRC | 12:42 | |
| *** dave-mcc_ has joined #openstack-ansible | 12:44 | |
| *** dave-mccowan has quit IRC | 12:45 | |
| *** pcaruana has quit IRC | 12:46 | |
| *** savvas has joined #openstack-ansible | 12:47 | |
| *** savvas has quit IRC | 12:51 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Ensure package cache is updated appropriately https://review.openstack.org/559725 | 12:51 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: repo-use.yml: Scope host groups properly https://review.openstack.org/559726 | 12:51 |
| *** ansmith has joined #openstack-ansible | 12:54 | |
| *** savvas has joined #openstack-ansible | 12:56 | |
| openstackgerrit | Fabien Andrieux proposed openstack/openstack-ansible-specs master: Blueprint enhancements as per reviewers comments https://review.openstack.org/559730 | 12:56 |
| openstackgerrit | Niraj Singh proposed openstack/openstack-ansible-specs master: Add specs to create masakari role https://review.openstack.org/556756 | 12:57 |
| *** hwoarang has quit IRC | 12:57 | |
| *** hwoarang has joined #openstack-ansible | 12:57 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 13:00 |
| *** savvas has quit IRC | 13:00 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 13:02 |
| *** pcaruana has joined #openstack-ansible | 13:11 | |
| odyssey4me | evrardjp ^ added some cleanup for the hyperconvgerged container | 13:11 |
| *** udesale has quit IRC | 13:12 | |
| openstackgerrit | Fabien Andrieux proposed openstack/openstack-ansible-specs master: Blueprint specs for openstack fds/fdio/vpp integration https://review.openstack.org/559733 | 13:12 |
| *** udesale has joined #openstack-ansible | 13:13 | |
| *** RandomTech has joined #openstack-ansible | 13:15 | |
| RandomTech | morning everyone | 13:15 |
| openstackgerrit | Merged openstack/openstack-ansible stable/pike: Added monitoring tag to monitoring user creation task. https://review.openstack.org/559321 | 13:19 |
| openstackgerrit | Merged openstack/openstack-ansible master: [Docs] Restructure inventory documentation https://review.openstack.org/552081 | 13:19 |
| openstackgerrit | Merged openstack/openstack-ansible master: Fix parameter name for removing compute host https://review.openstack.org/559573 | 13:19 |
| *** lvdombrkr has joined #openstack-ansible | 13:23 | |
| RandomTech | hey odyssey4me i noticed --check was implamented for the inventory. Do you happen to know what version of Openstack ansible pike i will need to use it? | 13:25 |
| odyssey4me | RandomTech I think that was introduced in newton, so any pike version | 13:26 |
| RandomTech | okay so i just need to regrab the version i was using? | 13:27 |
| RandomTech | wait nevermind i misread | 13:27 |
| RandomTech | ill see if it works with the one i have | 13:27 |
| *** lbragstad has joined #openstack-ansible | 13:27 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 13:33 |
| *** markvoelker has joined #openstack-ansible | 13:34 | |
| *** hwoarang has quit IRC | 13:35 | |
| *** markvoelker_ has joined #openstack-ansible | 13:36 | |
| *** rodolof has quit IRC | 13:39 | |
| *** rodolof has joined #openstack-ansible | 13:39 | |
| *** markvoelker has quit IRC | 13:39 | |
| *** esberglu has joined #openstack-ansible | 13:39 | |
| *** idlemind has joined #openstack-ansible | 13:40 | |
| evrardjp | odyssey4me: have the groups like cinder_scheduler_container disappeared, or are they mapped to the metal now? | 13:41 |
| evrardjp | (or api in this case) | 13:41 |
| idlemind | ugh that was fun ... dnf autoremove uninstalled my grub2-efi package and caused my workstation to stop booting ... good way to start out a monday | 13:41 |
| odyssey4me | evrardjp those groups are only in inventory.json because they were there before - in a new queens environment they no longer exist | 13:42 |
| odyssey4me | heh, every time I see dnf I think of 'did not finish' | 13:42 |
| idlemind | odyssey4me lol | 13:43 |
| *** CobHead has quit IRC | 13:43 | |
| evrardjp | odyssey4me: haha | 13:43 |
| evrardjp | that's the future of centos ! | 13:43 |
| evrardjp | (btw I see that too :p) | 13:43 |
| evrardjp | maybe an unpolished product? :p | 13:44 |
| evrardjp | idlemind: haha | 13:44 |
| idlemind | ya, i'm not sold on efi (uefi) either .. i understand the need to get away from legacy bios and the issues from it ... the process just seems rocky; of course the process to bios stability wasn't friendly from what i read in the history of that adventure either | 13:44 |
| evrardjp | I think uefi is nice | 13:45 |
| evrardjp | it's been too long we are with bios | 13:45 |
| idlemind | i should probably just update to a new 2018 edition mobo and get off my earlier implementation of it | 13:45 |
| evrardjp | like -- far too long | 13:45 |
| evrardjp | I think you have issues in bios and uefi implementations anyway | 13:45 |
| evrardjp | it's software AND hardware after all, what can go wrong? | 13:46 |
| idlemind | lol | 13:46 |
| evrardjp | odyssey4me: 16.0.5 to upgrade to Queens, I think that's a good example | 13:48 |
| evrardjp | :p | 13:48 |
| *** chhagarw has quit IRC | 13:48 | |
| *** mma has quit IRC | 13:49 | |
| odyssey4me | evrardjp what I did find a bit odd was that the neutron agents were all still in the neutron-agents container | 13:49 |
| odyssey4me | although I think that was by design - if they're there, leave them there | 13:49 |
| *** mma has joined #openstack-ansible | 13:49 | |
| odyssey4me | so this obviously means that a process to migrate where they are would need to be a bit different, and I haven't figured out how to do that yet | 13:50 |
| evrardjp | I think we should add a thing that checks if you have an apt lock | 13:50 |
| evrardjp | if you do something with apt, just don't start the upgrade | 13:51 |
| *** ansmith has quit IRC | 13:51 | |
| odyssey4me | what do you mean? | 13:51 |
| evrardjp | when I snapshotted my Pike, it was apparently doing an apt operation | 13:52 |
| odyssey4me | sounds like a pebkac problem ;) | 13:52 |
| evrardjp | any operation I do first with apt is failing because the apt lock is there | 13:52 |
| evrardjp | oh yes it starts as a pebkac :p | 13:52 |
| evrardjp | but my point is -- if there an apt lock file, we shouldn't try to upgrade | 13:53 |
| evrardjp | "Hey bro, you've something running on your machine, why don't you wait a little?" | 13:53 |
| evrardjp | ofc we can probably ignore it, and let apt crash/die | 13:54 |
| *** CobHead has joined #openstack-ansible | 13:54 | |
| evrardjp | but that's less user friendly | 13:54 |
| *** ansmith has joined #openstack-ansible | 13:54 | |
| evrardjp | also that only applies to deploy node | 13:54 |
| *** mma has quit IRC | 13:54 | |
| evrardjp | which is quite limited in scope vs all the upgrades in the world | 13:54 |
| evrardjp | yeah let's classify this not as wishlist, but pebkac. | 13:55 |
| odyssey4me | evrardjp sure, but then where do you draw the line? should we also check for disk space, available cpu, available ram - or maybe we should make coffee for the admin? | 13:56 |
| odyssey4me | we should perhaps execute the health check playbook right at the start | 13:56 |
| evrardjp | that's what I said -- let's classify as pebkac | 13:56 |
| odyssey4me | and perhaps add a check for an apt lock on all hosts in that playbook | 13:57 |
| evrardjp | yeah but healthchecks are hard | 13:57 |
| evrardjp | omg again | 13:58 |
| evrardjp | https://bugs.launchpad.net/openstack-ansible/+bug/1762421 | 13:58 |
| openstack | Launchpad bug 1762421 in openstack-ansible "openstack-ansible-galera_server role needlessly removes and reinstalls MariaDB-server" [Undecided,New] | 13:58 |
| evrardjp | let's merge galera_client and galera_server | 13:59 |
| *** tomtom001 has joined #openstack-ansible | 13:59 | |
| evrardjp | include_role galera_server/client.yml | 13:59 |
| evrardjp | that won't change a thing | 13:59 |
| *** markvoelker_ has quit IRC | 14:03 | |
| *** kstev has joined #openstack-ansible | 14:06 | |
| tomtom001 | Has anyone run into issues where the LXC cache would not finish in time? I changed the timeout, and checked the lxc-cache-prep-commands.log where the last line says setting up openssh-server, but then that's it, it doesn't go any further. Any ideas? | 14:07 |
| evrardjp | looks like you trouble building your things | 14:09 |
| evrardjp | maybe a lack of entropy? | 14:09 |
| *** cjloader has joined #openstack-ansible | 14:09 | |
| evrardjp | could you run havegd | 14:09 |
| evrardjp | ? | 14:09 |
| evrardjp | maybe that would help | 14:09 |
| evrardjp | adding it to the list of packages to install | 14:09 |
| evrardjp | or it's maybe something else | 14:09 |
| *** smorant has joined #openstack-ansible | 14:14 | |
| RandomTech | hey when setting up osa for the first time are there any advantages to pike over queens? | 14:18 |
| evrardjp | none | 14:18 |
| evrardjp | it's slower and older | 14:18 |
| evrardjp | :p | 14:19 |
| evrardjp | and you'll have to upgrade later | 14:19 |
| RandomTech | well i wish i knew that sooner :p. i may be starting again from cratch soon and might as well do queens then | 14:19 |
| *** rodolof has quit IRC | 14:20 | |
| evrardjp | we've never been shy to tell ppl to use latest stable in this channel ;) | 14:20 |
| *** rodolof has joined #openstack-ansible | 14:21 | |
| RandomTech | i guess i didnt undestand what the latest stable was then, sorry for the confusion. | 14:22 |
| Miouge | odyssey4me: Thanks for the tip last week. The setup-hosts.yml passed after checking out origin/stable/pike and re-running the scripts/bootstrap-ansible.sh | 14:23 |
| odyssey4me | RandomTech I think when you started out, queens was still in release candidate - so it made sense then to use pike | 14:23 |
| odyssey4me | but now that queens is released, may as well use it | 14:23 |
| RandomTech | ahh that makes sense | 14:23 |
| odyssey4me | the queens deployment/re-deployment is quite significantly faster... and we're hoping to make the next one even faster | 14:24 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: remove unused tunnel_bridge from all example configs https://review.openstack.org/559750 | 14:24 |
| cloudnull | mornings | 14:24 |
| cloudnull | evrardjp: what happened with networkd ? | 14:26 |
| cloudnull | something I need to take a look at? | 14:26 |
| RandomTech | would you say queens is any easier/harder to set up than pike? | 14:26 |
| odyssey4me | RandomTech absolutely no difference - the initial configuration mechanism has not changed | 14:27 |
| *** jonesba has joined #openstack-ansible | 14:27 | |
| *** throwsb1 has joined #openstack-ansible | 14:27 | |
| RandomTech | okay, sounds good, i may still try to trim down my configuration however. I have a feeling i overcomplicated it and tried to set up more than we needed | 14:28 |
| *** armaan has quit IRC | 14:28 | |
| RandomTech | i remember reading somewhere that ceph behaves alot better in cloud enviroments than swift. Is that still the case? | 14:29 |
| odyssey4me | RandomTech different use cases, one doesn't behave any better than the other | 14:29 |
| odyssey4me | ceph can be used for block storage, image storage and object storage... swift is object storage only | 14:30 |
| RandomTech | oh okay, i may have misunderstood what the page was trying to say | 14:30 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: Add mount options for better machinectl performance https://review.openstack.org/557970 | 14:30 |
| *** sar has quit IRC | 14:30 | |
| odyssey4me | swift is a specialised object storage system, ceph is more general purpose - ceph's object storage api is probably not the same as swift, but it'll cover many use-cases | 14:31 |
| cloudnull | odyssey4me: if you have a moment, mind having a look at https://review.openstack.org/#/c/558227 ? | 14:32 |
| odyssey4me | looking | 14:32 |
| cloudnull | hope is that we can pull in common roles into tests without having to map the dep throughout the stack | 14:32 |
| cloudnull | ^ regarding https://review.openstack.org/#/c/552185/13/tests/ansible-role-requirements.yml@64 | 14:32 |
| mnaser | if i write a playbook and just call openstack-ansible my-playbook.yml -- is that enough to access the inventory etc? | 14:35 |
| mnaser | i should probably test it out first but asking doesn't hurt i think :p | 14:35 |
| cloudnull | mnaser: yes | 14:36 |
| *** madorn_ has quit IRC | 14:36 | |
| *** DanyC has quit IRC | 14:36 | |
| mnaser | cool | 14:36 |
| * mnaser needs to do some haproxy stuff | 14:36 | |
| *** DanyC has joined #openstack-ansible | 14:37 | |
| cloudnull | cool | 14:37 |
| *** madorn has joined #openstack-ansible | 14:37 | |
| cloudnull | thanks odyssey4me! | 14:38 |
| *** savvas has joined #openstack-ansible | 14:38 | |
| odyssey4me | cloudnull yeah, I think that makes sense for the infrastructure roles (lxc, nspawn, openstack_hosts, galera, etc) - we can then allow the openstack roles to be more picky | 14:40 |
| cloudnull | ++ | 14:40 |
| *** DanyC has quit IRC | 14:41 | |
| odyssey4me | cloudnull with thanks to the initial work from fghaas, I pulled https://review.openstack.org/#/c/559721/ together - if you oculd take a peek that'd be grand | 14:41 |
| *** hw_wutianwei has joined #openstack-ansible | 14:41 | |
| odyssey4me | I'm building an env now with trove and ironic in the hopes to get those done too, then that is the hyperconvergence work done. | 14:42 |
| admin0 | during 17.0.0 to 17.0.1 i have something like this: {error,{inconsistent_cluster,"Node 'rabbit@c3-rabbit-mq-container-7ed627de' thinks it's clustered with node 'rabbit@c2-rabbit-mq-container-81e30416', but 'rabbit@c2-rabbit-mq-container-81e30416' disagrees"}} | 14:42 |
| admin0 | is it safe to nuke all rabbitmq clusters and re-create ? | 14:42 |
| odyssey4me | The missing piece after that is some tooling to automate implementing and moving where the neutron agents are. | 14:42 |
| odyssey4me | admin0 the rpc messaging in rabbitmq is stateless - if your cluster is broken, you can wipe and replace - you'll lose some transactions which were in motion, but if your cluster's borked then I guess that's not too much of an issue | 14:43 |
| odyssey4me | telemetry data will be lost, if you're using that - so that's the sticking point | 14:43 |
| odyssey4me | admin0 you can also wipe all containers but one instead which might save your data | 14:44 |
| admin0 | odyssey4me, i will keep 1 and wipe 2 | 14:45 |
| openstackgerrit | Merged openstack/openstack-ansible master: Make the on metal job voting https://review.openstack.org/559404 | 14:48 |
| *** lvdombrkr has quit IRC | 14:48 | |
| admin0 | so i have 1 old and 2 new .. is there a special flag to run the playbook .. or just run the normal rabbitmq-install.yml and it will fix itself ? | 14:48 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-tests master: test-install-keystone: Do not hardcode path for keystone-manage binary https://review.openstack.org/559757 | 14:48 |
| dcdamien | Is it fine to propose change referencing to bug report that is already fixed but ineffectively or should I open a new one? | 14:50 |
| cloudnull | odyssey4me: will have a look | 14:50 |
| *** sar has joined #openstack-ansible | 14:51 | |
| odyssey4me | admin0 if you've wiped the containers, then you need to recreate them first, then you can run the rabbitmq install playbook to reinstall them | 14:53 |
| mnaser | dcdamien: you could use Related-Bug: #foo | 14:53 |
| admin0 | me running /opt/openstack-ansible/scripts/inventory-manage.py outputs a | 14:55 |
| admin0 | ImportError: No module named prettytable | 14:55 |
| admin0 | that multiple bootstraps were not able to fix . .17.0.0, 17.0.1 .. what i can do manually to fix it ? | 14:55 |
| evrardjp | cloudnull: it seems to fail -- but I haven't reverted it | 14:56 |
| evrardjp | and deploy of suse doesn't seem to work | 14:56 |
| evrardjp | admin0: go to your osa venv | 14:56 |
| evrardjp | pip install prettytable | 14:57 |
| odyssey4me | admin0 use /opt/ansible-runtime/bin/python to execute the script | 14:57 |
| dcdamien | mnaser: thx | 14:57 |
| admin0 | ImportError: No module named 'dictutils' | 14:57 |
| evrardjp | cloudnull: a* deploy of suse doesn't seem to work | 14:58 |
| cloudnull | it looks like it passed the gate? | 14:58 |
| evrardjp | cloudnull: yes I saw it passing this morning too | 14:58 |
| evrardjp | weirdly | 14:58 |
| evrardjp | so it's I guess a package that's not explicitly listed | 14:59 |
| odyssey4me | admin0 you could also do: source /opt/ansible-runtime/bin/activate | 14:59 |
| cloudnull | oh? | 14:59 |
| odyssey4me | then run the tools | 14:59 |
| evrardjp | admin0: do what odyssey4me just said :p | 15:00 |
| evrardjp | that's what I meant by enter the venv | 15:00 |
| admin0 | evrardjp, odyssey4me even with source and inside the venv, still the same .. no module named dictutils | 15:00 |
| evrardjp | pip install dictutils | 15:01 |
| odyssey4me | admin0 hmm, that's odd | 15:01 |
| *** chyka has joined #openstack-ansible | 15:01 | |
| evrardjp | are you sure it's dictutils? | 15:01 |
| evrardjp | isn't that oslo something? | 15:01 |
| odyssey4me | since when do we use dictutils? | 15:01 |
| evrardjp | What's the line? | 15:02 |
| evrardjp | I don't know | 15:02 |
| evrardjp | that's weird | 15:02 |
| evrardjp | without full log I won't consider this as a bug :p | 15:02 |
| *** jonesba has quit IRC | 15:03 | |
| *** armaan has joined #openstack-ansible | 15:03 | |
| admin0 | evrardjp, https://gist.github.com/a1git/568d917533d92298d6c6ee449de473d0 | 15:04 |
| *** MikeW has joined #openstack-ansible | 15:06 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: Use a sensible vlan range in the example configs https://review.openstack.org/559763 | 15:06 |
| *** rodolof has quit IRC | 15:07 | |
| evrardjp | odyssey4me: admin0 dictutils is releative | 15:08 |
| evrardjp | relative | 15:08 |
| evrardjp | cd | 15:08 |
| evrardjp | cd /opt/openstack-ansible | 15:08 |
| evrardjp | scripts/inventory-manage.py -l should work | 15:08 |
| evrardjp | poor packaging | 15:08 |
| *** qwebirc34220 has joined #openstack-ansible | 15:08 | |
| evrardjp | you can install the osa_toolkit in your venv it should work too but that's very ugly | 15:09 |
| MikeW | Hey have you guys had any issues with rabbitmq on a default deploy of 15.1.6? | 15:09 |
| admin0 | evrardjp, fails at prettytable | 15:10 |
| evrardjp | MikeW: could you be more explicit? | 15:10 |
| evrardjp | admin0: pip install prettytable | 15:10 |
| admin0 | 19 | 15:10 |
| qwebirc34220 | Hi guys! I try to change the ip address of management network but containers still use the previous network i re ran the setup-hosts.yml but nothing changed. could anyone help me please | 15:10 |
| admin0 | qwebirc34220 .. tricky | 15:10 |
| *** radeks has quit IRC | 15:11 | |
| MikeW | evrardjp Yeah sorry I'm sleep deprived... The server just randomly stops responding. I've been looking into it maybe being a haproxy/rabbitmq issue, because it's in no way resource constrained. | 15:11 |
| *** radeks has joined #openstack-ansible | 15:11 | |
| admin0 | MikeW, flaoting ip jumping ? | 15:11 |
| admin0 | check vip | 15:11 |
| admin0 | vip logs | 15:12 |
| admin0 | also conntrack | 15:12 |
| admin0 | maybe your tables are full | 15:12 |
| stuartgr | stable/queens bootstrap-ansible.sh picks python3 for ansible runtime if its available, but [pip_install : Drop pip config] is failing with "NameError: name 'unicode' is not defined" which appears to be caused when python2 code runs in python3? | 15:12 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-ceph_client master: vars: suse: Use correct package for python Ceph bindings https://review.openstack.org/559766 | 15:12 |
| *** hwoarang has joined #openstack-ansible | 15:14 | |
| MikeW | admin0 Cool thanks for the ideas. I've been circling on this issue for a while thinking it was memory or disk issues. | 15:14 |
| qwebirc34220 | admin0 so what should i do ? going back to previous setup is the only solution ? | 15:14 |
| admin0 | qwebirc34220, if you nuke the inventory, then it will create all brand new .. .. then you need to mysqldump and restore in the new cluster | 15:15 |
| admin0 | you have twice of everythign running | 15:15 |
| admin0 | unless | 15:16 |
| admin0 | qwebirc34220, what is the need to change the network ip ? | 15:16 |
| admin0 | i mean whats the need to change ips ? | 15:16 |
| admin0 | MikeW, if the servers are busy .. compute, controllers etc .. could be conntrack dropping new connections | 15:17 |
| evrardjp | admin0: make sure you're using the right "pip" binary too | 15:17 |
| admin0 | evrardjp, i tend to do bootstrap and "assume" all will be taken care in that proces | 15:18 |
| evrardjp | admin0: it is | 15:18 |
| evrardjp | it should be | 15:18 |
| evrardjp | let me check if pretty table is in our requirements | 15:18 |
| MikeW | admin0 Yeah it really feels like conntrack is what's going on... seems like the servers crash after getting busy | 15:18 |
| admin0 | how can i "un bootstrap" or force a override etc | 15:18 |
| admin0 | MikeW, what monitoring do you use ? | 15:19 |
| qwebirc34220 | admin0 there was an ip conflict between management and lxc bridges. i didn't find where the ip of lxc brdges set so i decided to change the network of management. finally i found where ip of lxc bridge sets . if i change management to previous ips and change the network for lxc . is the issue resolved? | 15:19 |
| MikeW | admin0 We're only using nagios right now, and I haven't implemented networking checks yet | 15:19 |
| admin0 | lxc is just used to provide NAT | 15:19 |
| admin0 | i think you can change lxc range | 15:19 |
| admin0 | but keep the mgmt as is | 15:20 |
| admin0 | its a hassle to change | 15:20 |
| admin0 | if you have a workload | 15:20 |
| admin0 | the default ips are good enough .. i do not understand why people need to change that :) | 15:20 |
| admin0 | its unroutable network range .. where 10.1 is as good as 100.1 :D | 15:20 |
| evrardjp | admin0: pretty table is in requirements | 15:20 |
| evrardjp | re-bootstrap your ansible | 15:20 |
| evrardjp | you have somethign wrong there | 15:21 |
| evrardjp | then run scripts/inventory-manage | 15:21 |
| evrardjp | it should work | 15:21 |
| admin0 | evrardjp, is there a --force --i-am-really-sure flag ? | 15:21 |
| evrardjp | /opt/ansible-runtime/bin/python scripts/inventory-manage.py -l | 15:21 |
| evrardjp | that works | 15:21 |
| evrardjp | for me | 15:22 |
| qwebirc34220 | admin0 thanks | 15:23 |
| admin0 | evrardjp, for me it gives the dictutils module failure | 15:23 |
| evrardjp | let me try with your version | 15:23 |
| evrardjp | 17.0.1 you said? | 15:23 |
| admin0 | yep | 15:23 |
| admin0 | last commit "2018-04-02 11:25 Logan V o <17.0.1> Bump pip_install SHA to include constraints fix | 15:24 |
| admin0 | " | 15:24 |
| *** armaan_ has joined #openstack-ansible | 15:25 | |
| *** armaan has quit IRC | 15:25 | |
| evrardjp | admin0: interesting I have the same issue when doing /opt/ansible-runtime/bin/python scripts/inventory-manage.py -l | 15:25 |
| evrardjp | but not when doing scripts/inventory-manage.py -l | 15:26 |
| evrardjp | I found the issue I think PrettyTable vs prettytable | 15:26 |
| admin0 | one is uglytable :D | 15:27 |
| *** chyka has quit IRC | 15:28 | |
| evrardjp | let me start from a clean machine | 15:28 |
| evrardjp | I'm lacking time right now admin0 | 15:29 |
| evrardjp | could you file a bug? | 15:29 |
| admin0 | bug on its way | 15:29 |
| evrardjp | I will work on it tomorrow | 15:29 |
| evrardjp | thanks | 15:29 |
| admin0 | when is our bug day ? | 15:29 |
| evrardjp | tuesday | 15:29 |
| admin0 | i want to file 30 mins before we start :D | 15:29 |
| evrardjp | lol no | 15:30 |
| admin0 | game the system | 15:30 |
| evrardjp | just od it | 15:30 |
| admin0 | ok | 15:30 |
| admin0 | will do | 15:30 |
| evrardjp | I will maybe fix it in the meantime | 15:30 |
| admin0 | everyone who will start new will start with 17.0.1 .. so they will all break | 15:31 |
| jrosser | our queens attempt is not getting very far either , see stuartgr comment ^^^^ | 15:31 |
| evrardjp | I think we should run python2 on the host | 15:34 |
| evrardjp | destination host | 15:34 |
| evrardjp | jrosser: | 15:34 |
| admin0 | https://bugs.launchpad.net/openstack-ansible/+bug/1762459 | 15:34 |
| openstack | Launchpad bug 1762459 in openstack-ansible "inventory-manage.py broken on 17.0.x" [Undecided,New] | 15:34 |
| evrardjp | check if you have somethign in your inventory that is using something else? | 15:34 |
| evrardjp | ansible_python_interpreter? | 15:34 |
| evrardjp | else I suggest you to file a bug | 15:34 |
| *** qwebirc34220 has quit IRC | 15:34 | |
| evrardjp | jrosser: http://zuul.openstack.org/builds.html | 15:35 |
| * jrosser looks | 15:36 | |
| evrardjp | jrosser: sorry not full link | 15:36 |
| evrardjp | if you don't want to fill the blanks: | 15:36 |
| evrardjp | http://zuul.openstack.org/builds.html?pipeline=periodic&project=openstack%2Fopenstack-ansible&branch=stable%2Fqueens | 15:36 |
| evrardjp | you can see the latest stable/queens deploy worked | 15:36 |
| mnaser | admin0: i had that issue too | 15:36 |
| mnaser | yum -y install python-prettytable python-yaml fixed it | 15:37 |
| evrardjp | mnaser: yes but that shouldn't happen | 15:37 |
| evrardjp | it worked on my machine | 15:37 |
| mnaser | ubuntu or centos? | 15:37 |
| evrardjp | so I suspect that a new machine is missing some requirements | 15:37 |
| mnaser | this was a very minimal machine though | 15:37 |
| admin0 | evrardjp, https://image.slidesharecdn.com/devopsdayemcworld2015-memes-final-150501164657-conversion-gate02/95/devops-memes-emcworld-2015-83-638.jpg?cb=1430517087 | 15:37 |
| evrardjp | I think it's poor packaging of our lib | 15:38 |
| evrardjp | admin0: which os? | 15:38 |
| evrardjp | is that ubuntu or centos? | 15:38 |
| admin0 | ubuntu 16.04 | 15:38 |
| evrardjp | yeah | 15:38 |
| evrardjp | so | 15:38 |
| evrardjp | I will start with a clean ubuntu when I will have time | 15:38 |
| evrardjp | but I am sure it's something that's not properly installed in the venv | 15:38 |
| * mnaser made a docker image with OSA | 15:38 | |
| evrardjp | mnaser: haha | 15:38 |
| mnaser | and that is broken if we don't do that | 15:39 |
| evrardjp | ? | 15:39 |
| evrardjp | mnaser: what do you mean? | 15:39 |
| evrardjp | for me, we should have this tookit packaged separately | 15:39 |
| evrardjp | that's it | 15:39 |
| mnaser | evrardjp: https://gist.github.com/mnaser/30a66333072d3bf52f662397c1cd58a4 | 15:40 |
| mnaser | if you drop/comment out the "RUN yum -y install python-prettytable python-yaml", then the inventory manage won't work | 15:40 |
| mnaser | so many cause its minimal i guess | 15:40 |
| evrardjp | that shouldn't because prettytable and pyyaml are in the requirements.txt | 15:40 |
| admin0 | apt install python-prettytable python-yaml fixed it | 15:41 |
| evrardjp | https://github.com/openstack/openstack-ansible/blob/master/requirements.txt#L8-L10 | 15:41 |
| mnaser | wait so | 15:41 |
| mnaser | that means that maybe | 15:41 |
| evrardjp | yeah | 15:41 |
| mnaser | the inventory script is not using the virtualenv? | 15:41 |
| evrardjp | it means it's outside the venv | 15:41 |
| admin0 | but the point is its in there, but should have been installed | 15:41 |
| evrardjp | yeah | 15:41 |
| evrardjp | that's exactly what I am saying | 15:41 |
| mnaser | i'm slow :D | 15:41 |
| evrardjp | admin0: no | 15:41 |
| evrardjp | that's not what should be done | 15:41 |
| evrardjp | the packaging shuld be better | 15:41 |
| mnaser | https://raw.githubusercontent.com/openstack/openstack-ansible/master/scripts/inventory-manage.py | 15:41 |
| mnaser | #!/usr/bin/env python | 15:41 |
| evrardjp | basically changing the shebang shoudl help | 15:42 |
| evrardjp | YES | 15:42 |
| admin0 | when i see a new patch, i will apt autoremove --purge, try yoru patch and check | 15:42 |
| mnaser | should it be like | 15:42 |
| evrardjp | probably python3 | 15:42 |
| mnaser | /opt/openstack-ansible/bin/python or something | 15:42 |
| evrardjp | not sure | 15:42 |
| evrardjp | but it's probably a thing with python3 | 15:42 |
| mnaser | https://github.com/openstack/openstack-ansible/blame/master/scripts/inventory-manage.py | 15:42 |
| mnaser | broken since initial commit | 15:43 |
| mnaser | ha :P | 15:43 |
| Tahvok | Can anyone running ironic give me a screenshot of how it looks now in horizon? I need it for presentation | 15:43 |
| evrardjp | admin0: could you check what's the binary included in your /opt/openstack-ansible/bin/py* ? | 15:43 |
| evrardjp | ls -Fal would do the trick | 15:43 |
| evrardjp | mnaser: haha yes | 15:43 |
| evrardjp | I have a spec to change the ivnentory | 15:43 |
| evrardjp | and get rid of this | 15:44 |
| evrardjp | but here I am sure it's a python issue | 15:44 |
| evrardjp | ofc if we say we should /opt/openstack-ansible/bin/python in the shebang it would fix the issue | 15:44 |
| *** electrofelix has quit IRC | 15:44 | |
| evrardjp | mnaser: admin0 trust me, I will fix it. | 15:45 |
| evrardjp | :D | 15:45 |
| evrardjp | a quick hack and a long term cleanup :p | 15:46 |
| mnaser | more hardcoding woo :P | 15:46 |
| evrardjp | that's the quick hack :p | 15:46 |
| admin0 | that folder , /opt/openstack-ansible/bindep.txt has this text only that starts with bin | 15:46 |
| admin0 | i assume you meant runtime | 15:46 |
| evrardjp | admin0: could you answer my question plz? | 15:47 |
| evrardjp | did I paste the wrong thing | 15:47 |
| evrardjp | ofc I did | 15:47 |
| evrardjp | darn | 15:47 |
| admin0 | :D | 15:47 |
| evrardjp | ls -Fal /opt/ansible-runtime/bin/py | 15:47 |
| evrardjp | ls -Fal /opt/ansible-runtime/bin/py* | 15:47 |
| evrardjp | danr | 15:47 |
| admin0 | python2 | 15:47 |
| admin0 | python3 | 15:47 |
| admin0 | sorry 3 | 15:48 |
| admin0 | let me gist | 15:48 |
| admin0 | https://gist.github.com/a1git/c895736ff5f0a7ab76a01f7aff91536c | 15:48 |
| evrardjp | http://logs.openstack.org/46/559146/2/check/openstack-ansible-functional-centos-7/9753a3a/job-output.txt.gz#_2018-04-09_08_24_16_091754 | 15:50 |
| evrardjp | cloudnull: ^ | 15:50 |
| evrardjp | apparently that's blocking the tempest patch | 15:50 |
| admin0 | MikeW, wwhen you have time, check out check_mk | 15:50 |
| admin0 | for your openstack monitoring needs | 15:50 |
| cloudnull | evrardjp: thanks! | 15:51 |
| evrardjp | cloudnull: well no reason to thank me :p | 15:52 |
| evrardjp | I am sorry it's broken | 15:52 |
| evrardjp | :p | 15:52 |
| cloudnull | meh, it happens :) | 15:52 |
| *** nicolasbock has left #openstack-ansible | 15:52 | |
| admin0 | taking a quick drive home .. will reconnect back later | 15:52 |
| evrardjp | cloudnull: fyi it's an ovh node | 15:52 |
| admin0 | my 17.0.0 -> 17.0.1 upgrade is running | 15:53 |
| evrardjp | so we do some kind of munging there | 15:53 |
| evrardjp | admin0: ok I did something wrong in the release of 17.0.1 | 15:53 |
| evrardjp | please use stable/queens :p | 15:53 |
| admin0 | :D | 15:53 |
| admin0 | will it break a running cluster ? | 15:54 |
| evrardjp | nope but it will build things weirdly on your repo build | 15:54 |
| evrardjp | cleanup your repo and you should be fine | 15:54 |
| admin0 | will check only if somethign breaks .. if the upgrade goes in fine and all is fine, will wait to address it in 17.0.2 | 15:54 |
| admin0 | repo is in setup-infra right .. those playbooks already passed fine | 15:55 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: Add IP addresses to infra node br-vxlan in the examples https://review.openstack.org/559775 | 15:56 |
| *** sep has quit IRC | 15:57 | |
| *** admin0 has quit IRC | 15:57 | |
| *** sep has joined #openstack-ansible | 15:58 | |
| *** smorant has quit IRC | 15:59 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: Remove spurious VxLAN bridge IP from ceph osd hosts https://review.openstack.org/559776 | 16:01 |
| evrardjp | d34dh0r53: we have a few upgrade jobs that are failing, could you have a look? http://zuul.openstack.org/builds.html?pipeline=periodic&project=openstack%2Fopenstack-ansible | 16:02 |
| d34dh0r53 | evrardjp: yep | 16:03 |
| evrardjp | thanks | 16:03 |
| evrardjp | I'd start with master :) | 16:03 |
| evrardjp | cloudnull: do you check on the gating issue for centos/suse for that tempest patch? | 16:03 |
| *** DanyC has joined #openstack-ansible | 16:03 | |
| evrardjp | else I can work on it | 16:03 |
| cloudnull | I have not yet | 16:04 |
| evrardjp | ok I will tackle this | 16:04 |
| cloudnull | though i will be looking at suse in a min | 16:04 |
| cloudnull | well as soon as my lab is back online | 16:04 |
| evrardjp | ok found it | 16:06 |
| evrardjp | will fix the bug real quick | 16:06 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 16:06 |
| *** DanyC has quit IRC | 16:08 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-tests master: Use the right hypervisor name for ovh hosts https://review.openstack.org/559778 | 16:09 |
| evrardjp | odyssey4me: cloudnull ^ | 16:09 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 16:09 |
| evrardjp | that should unblock nova role tests | 16:09 |
| evrardjp | see also http://logs.openstack.org/46/559146/2/check/openstack-ansible-functional-centos-7/9753a3a/logs/ara-report/result/75fa2c7e-6aa6-40df-a1f0-60fe5c97b65c/ | 16:10 |
| evrardjp | thanks | 16:10 |
| openstackgerrit | Merged openstack/openstack-ansible stable/queens: Developer docs refactor https://review.openstack.org/559059 | 16:11 |
| odyssey4me | evrardjp I think we also need https://review.openstack.org/559146 in for the nova gate, because there is only one image provided in the lxd test isn't there? | 16:12 |
| *** DanyC has joined #openstack-ansible | 16:13 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-tests stable/queens: Add Ceph experimental job for roles https://review.openstack.org/559781 | 16:13 |
| *** DanyC_ has joined #openstack-ansible | 16:15 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ceph_client master: Adapt the list of ceph packages for SUSE https://review.openstack.org/559688 | 16:16 |
| *** DanyC has quit IRC | 16:18 | |
| evrardjp | odyssey4me: yes, it's documented :) | 16:19 |
| evrardjp | you should see it on your patch | 16:19 |
| evrardjp | I mean | 16:19 |
| evrardjp | on jimmy's patch | 16:19 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 16:19 |
| evrardjp | odyssey4me: I will backport the ovh things back to I don't know where | 16:19 |
| odyssey4me | evrardjp just queens | 16:20 |
| evrardjp | I will | 16:20 |
| evrardjp | :p | 16:20 |
| odyssey4me | well, actually I can't recall if it ported back to pike - but I doubt it | 16:20 |
| evrardjp | i will check | 16:20 |
| evrardjp | I am talking about the tests | 16:21 |
| evrardjp | you did for the integrated | 16:21 |
| evrardjp | will check what you did and mirror | 16:21 |
| evrardjp | you backported to O | 16:22 |
| odyssey4me | evrardjp already checked, but due to the implementation I don't think its suffering the same problem | 16:22 |
| evrardjp | did we miss P? | 16:22 |
| odyssey4me | the reason the problem is seen in tests is due to https://github.com/openstack/openstack-ansible-tests/blob/master/common-tasks/test-set-nodepool-vars.yml#L45 - and the integrated build doesn't do that | 16:22 |
| evrardjp | yeah I know it's a different implementation | 16:22 |
| evrardjp | I will have trouble to remember where we did it if we backport it differently in tests and in the integrated | 16:23 |
| evrardjp | I just want to have the same kind of hacks at the same place | 16:23 |
| odyssey4me | only did master/queens: https://review.openstack.org/#/q/I4f679815ac55095000dc1017cebbd1a8bf0f6c5f | 16:23 |
| logan- | thanks for putting the upgrade plays together odyssey4me | 16:23 |
| openstackgerrit | Merged openstack/openstack-ansible-ceph_client master: Add ceph integrated repo testing https://review.openstack.org/559695 | 16:23 |
| odyssey4me | evrardjp for the integrated I think I did it further back | 16:23 |
| evrardjp | odyssey4me: I am talking about https://review.openstack.org/#/q/I54681092adf9c0647bf65583176ddc21c99494f3 | 16:23 |
| evrardjp | yeah | 16:24 |
| evrardjp | I don't see P there | 16:24 |
| odyssey4me | ah yeah for that one the p patch was merged with the pip_install role sha update | 16:24 |
| evrardjp | oh ok | 16:24 |
| evrardjp | yeah I remember now | 16:24 |
| evrardjp | cool | 16:24 |
| evrardjp | so backport all the way to O | 16:24 |
| evrardjp | for the tests implementation | 16:24 |
| evrardjp | if we can | 16:24 |
| odyssey4me | logan- it was kinda fun actually, oddly enough | 16:24 |
| odyssey4me | been a while since I actually had to use the openstack cli | 16:25 |
| RandomTech | does anyone have an example of what a custom network dedicated to floating ips looks like? | 16:25 |
| logan- | odyssey4me: yeah its nice, the -f json output makes the output really easy to work with in ansible | 16:30 |
| logan- | although I noticed you aren't using it, but next time it might be easier for you to use -f json in openstack CLI and then consume it in ansible with '| from_json' | 16:31 |
| odyssey4me | logan- yeah, that is handy - but I didn't really need it here | 16:31 |
| logan- | ya | 16:31 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-tests stable/queens: Use the right hypervisor name for ovh hosts https://review.openstack.org/559785 | 16:32 |
| odyssey4me | logan- that's most useful when you're getting a large bunch of data you want to work with - this was much more surgical | 16:32 |
| logan- | yea | 16:32 |
| odyssey4me | logan- I was looking at resource usage as well and you're right, Queens is a bit of a hog compared to pike. I think we need to tune a bunch of defaults down a bit. | 16:34 |
| odyssey4me | I suspect that we might be a little too generous on the uwgi threads/processes settings. | 16:34 |
| logan- | thats interesting.. I haven't compared pike vs queens yet, but i did notice a large improvement once i deleted all of the duplicate nova/cinder/heat services | 16:35 |
| odyssey4me | yes, that definitely helps | 16:35 |
| odyssey4me | I did also find some cinder services hogging cpu | 16:35 |
| odyssey4me | I suspect hat reloading the service instead of restarting it is still an issue. | 16:35 |
| logan- | ouch | 16:36 |
| odyssey4me | it all relates back to this patch: https://review.openstack.org/#/c/528140/ | 16:37 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-tests stable/pike: Run with KVM if we can in tests https://review.openstack.org/559787 | 16:37 |
| odyssey4me | I've got a couple more test environments up now to validate my final patch set and build plays for trove and ironic. | 16:37 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ceph_client stable/queens: Add ceph integrated repo testing https://review.openstack.org/559788 | 16:39 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-ceph_client stable/queens: Add ceph integrated repo testing https://review.openstack.org/559788 | 16:41 |
| *** epalper has quit IRC | 16:42 | |
| evrardjp | odyssey4me: it seems the upgrade has worked for me | 16:43 |
| evrardjp | with your patch | 16:43 |
| evrardjp | I still have lots of containers | 16:44 |
| evrardjp | we can discuss later, but at least it looks nice | 16:44 |
| evrardjp | thanks for the owrk | 16:44 |
| *** udesale has quit IRC | 16:46 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-tests stable/pike: Run with KVM if we can in tests https://review.openstack.org/559787 | 16:48 |
| *** armaan_ has quit IRC | 16:49 | |
| *** armaan has joined #openstack-ansible | 16:50 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Remove in-development warning from run-upgrade https://review.openstack.org/559791 | 16:50 |
| *** MikeW has quit IRC | 16:50 | |
| -openstackstatus- NOTICE: zuul was restarted to update to the latest code; please recheck any changes uploaded within the past 10 minutes | 16:50 | |
| *** armaan has quit IRC | 16:54 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible stable/queens: Have zuul check out ansible for devel AIO job https://review.openstack.org/559793 | 16:55 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible stable/queens: Run the on metal job https://review.openstack.org/559794 | 16:55 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible stable/queens: Make the on metal job voting https://review.openstack.org/559795 | 16:55 |
| *** savvas has quit IRC | 16:57 | |
| evrardjp | if ppl not logan could vote on https://review.openstack.org/#/c/559781/ | 17:01 |
| *** savvas has joined #openstack-ansible | 17:03 | |
| odyssey4me | evrardjp logan- Two backport patches which improve usability and speed: https://review.openstack.org/559725 & https://review.openstack.org/559726 | 17:03 |
| tomtom001 | Should I be using python3 for pike? is python2.7 still usable? | 17:06 |
| evrardjp | I'm off for today | 17:07 |
| evrardjp | I will read that later tomorrow | 17:07 |
| *** savvas has quit IRC | 17:07 | |
| evrardjp | d34dh0r53: if you could also help on 558082 that would be nice | 17:08 |
| evrardjp | https://review.openstack.org/#/c/558082/ | 17:09 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-tests stable/queens: Extend linters time out https://review.openstack.org/559797 | 17:09 |
| *** RandomTech has quit IRC | 17:09 | |
| d34dh0r53 | evrardjp: yep | 17:10 |
| evrardjp | odyssey4me: haha I was doing it | 17:10 |
| evrardjp | beat me to it | 17:10 |
| evrardjp | oh you just changed that | 17:10 |
| evrardjp | thought you were planning to do the big backport :) | 17:10 |
| odyssey4me | evrardjp I looked, then realised that was more complicated and didn't think that'd be a good idea | 17:11 |
| evrardjp | yup | 17:11 |
| evrardjp | same for me :) | 17:11 |
| evrardjp | haha | 17:11 |
| *** savvas has joined #openstack-ansible | 17:12 | |
| *** savvas has quit IRC | 17:17 | |
| tomtom001 | cloudnull I'm looking at: /etc/ansible/roles/lxc_hosts/tasks/lxc_cache_preparation.yml @ http://paste.openstack.org/show/718759/ Do you have any wisdom on what this process is doing? From what I can tell it runs a machine in chroot and does an upgrade, but my ansible environment fails on this everytime. Is it just waiting for the machine to upgrade? | 17:18 |
| sar | Running openstack-ansible repo-build.yml suddenly doesn't work anymore. I get this error while building gnocchi: "error: error in setup.cfg: command 'build_py' has no such option 'pre_hook.build_config'" Anyone else having the same issue?: | 17:18 |
| odyssey4me | tomtom001 it downloads a pre-build image from lxccontainers.org, then extracts it, then runs the script in vars/<distro> inside that chroot, then packages it for re-use when building the other containers | 17:19 |
| tomtom001 | ok, thank you | 17:19 |
| sar | (i'm on ocata 15.1.13) | 17:19 |
| odyssey4me | sar you're hitting the pbr/pip issues which have been resolved in the most recent releases - update your checkout to the latest release, bootstrap-ansible again, then do the rest of the minor upgrade process | 17:19 |
| sar | minor upgrade process? | 17:20 |
| tomtom001 | odyssey4me do you mean linuxcontainers.org? | 17:21 |
| *** savvas has joined #openstack-ansible | 17:21 | |
| tomtom001 | i could not find lxccontainers.org | 17:21 |
| odyssey4me | tomtom001 yep, apologies | 17:22 |
| tomtom001 | thanks | 17:22 |
| odyssey4me | tomtom001 at the stage you're at, the download's already done | 17:22 |
| odyssey4me | it's failing inside the script in the chroot | 17:22 |
| odyssey4me | tomtom001 which release are you working with? | 17:23 |
| tomtom001 | Pike | 17:23 |
| tomtom001 | git checkout stable/pike | 17:23 |
| tomtom001 | Unfortunately the lxc_cache_prep_commands log doesn't reveal any info as to what it's failing on from what I can tell except a random invoke-rc.d can't determine runlevel error. | 17:24 |
| tomtom001 | I've google'd it, but didn't find anything on it except some incompatibilites with systemd vs upstart | 17:25 |
| odyssey4me | tomtom001 I think that's more likely just a chroot funny | 17:25 |
| *** savvas has quit IRC | 17:25 | |
| odyssey4me | tomtom001 this is what runs - can you see any clues regarding where it got to? https://github.com/openstack/openstack-ansible-lxc_hosts/blob/stable/pike/vars/ubuntu-16.04.yml#L63-L95 | 17:26 |
| *** savvas has joined #openstack-ansible | 17:26 | |
| tomtom001 | Yes it gets to apt-get install -y that much I can see from the cache prep logs | 17:28 |
| tomtom001 | it seems in the middle of it that it just dies. | 17:28 |
| cloudnull | tomtom001: it sounds like the base image is messed up. assuming you've not done this already, I'd nuke the cache and allow the playbooks to rebuild it. | 17:29 |
| tomtom001 | I also see this error right before it starts unpacking a bunch of packages: E: Can not write log (Is /dev/pts mounted?) - posix_openpt (2: No such file or directory) | 17:29 |
| odyssey4me | that's a chroot funny and normal I think | 17:30 |
| odyssey4me | cloudnull what's the way to wipe the cache again? I forget | 17:30 |
| cloudnull | `machinectl remove $image_name; rm -rf /var/cache/lxc/download/*` | 17:30 |
| tomtom001 | oh, I've never seen that before. ok thanks! I'll try it. | 17:30 |
| cloudnull | if you have the image in machinectl you can list them | 17:31 |
| cloudnull | machinectl list-images | 17:31 |
| odyssey4me | cloudnull maybe that'd be a good entry for somewhere in here: https://docs.openstack.org/openstack-ansible/latest/admin/troubleshooting.html | 17:31 |
| cloudnull | the base cache will be named something like $distro-$arch-$release | 17:31 |
| cloudnull | odyssey4me: agreed. I can toss something in that | 17:31 |
| odyssey4me | awesome | 17:32 |
| *** kstev has quit IRC | 17:32 | |
| tomtom001 | Yeah to be honest this is the first time I've seen a problem relating to the lxc containers not being built or the process that actually builds the containers, sorry for my ignorance. | 17:32 |
| cloudnull | no need to be sorry, I'm honestly not sure why that would be failing for you like this. | 17:33 |
| cloudnull | ive seen similar issues when the cache image is downloaded from lxc and fails prep, ive also seen a bad image cache from the upstream image build process cause something similar. | 17:34 |
| cloudnull | I generally try and fix an issue for a few minutes, if i cant make it go, i nuke things | 17:34 |
| odyssey4me | https://media.giphy.com/media/U0xH9PiljwlS8/giphy.gif | 17:41 |
| *** savvas has quit IRC | 17:42 | |
| *** chhagarw has joined #openstack-ansible | 17:42 | |
| odyssey4me | cloudnull got a gap to take a peek at https://review.openstack.org/559725 which will help folks out a bit also | 17:43 |
| mnaser | is there a way to override specific things out of `haproxy_default_services` ? | 17:46 |
| mnaser | ex: add frontend options to an existing service | 17:46 |
| logan- | not really mnaser, but you could add a service in extra_services with the same name | 17:46 |
| logan- | so it would overwrite the previous service | 17:46 |
| mnaser | dang, at the expense of possibly getting out of sync later | 17:47 |
| logan- | yup | 17:47 |
| *** savvas has joined #openstack-ansible | 17:47 | |
| * mnaser is trying to get all services to respond on port 80/443 with dns lookups using https://www.haproxy.com/blog/web-application-name-to-backend-mapping-in-haproxy/ | 17:47 | |
| mnaser | logan-: actually i wonder if that would even work, because it looks like it's just merging two arrays | 17:48 |
| mnaser | so it would just be doubled up, it doesnt look like its a map | 17:48 |
| logan- | yeah the extra services are appended | 17:48 |
| logan- | so the extra service would overwrite the previous service before the assemble task for /etc/haproxy/haproxy.conf runs | 17:48 |
| logan- | it is really hacky | 17:48 |
| * mnaser thinks | 17:49 | |
| *** kstev has joined #openstack-ansible | 17:53 | |
| cloudnull | odyssey4me: its more like https://media1.giphy.com/media/BT4ygwV9vgwAU/giphy.gif | 18:00 |
| * cloudnull looking at 559725 | 18:01 | |
| odyssey4me | hahaha | 18:01 |
| cloudnull | odyssey4me: nice change! | 18:01 |
| *** poopcat has joined #openstack-ansible | 18:02 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Ensure both nova and ironic have access to ironic vars https://review.openstack.org/559804 | 18:06 |
| odyssey4me | d34dh0r53 prometheanfire ^ That'll be at least one reason why Ironic on Queens doesn't work. | 18:06 |
| prometheanfire | odyssey4me: cool, so that's for master/queens? | 18:08 |
| odyssey4me | prometheanfire ja, we did some rearrangement of group_vars for queens which clearly broke some things | 18:09 |
| odyssey4me | I happen to be doing a p->q upgrade test right now with ironic in the mix and ironic wouldn't even complete deployment due to that. | 18:09 |
| prometheanfire | heh | 18:11 |
| *** eumel8 has quit IRC | 18:13 | |
| *** kstev has quit IRC | 18:17 | |
| *** Administrator__ has joined #openstack-ansible | 18:21 | |
| mnaser | question | 18:21 |
| mnaser | decision behind running haproxy on is_metal.. any context? | 18:21 |
| odyssey4me | mnaser logan- runs it in containers | 18:22 |
| odyssey4me | and there is no real reason other than for rackspace it was never a 'supported' configuration, not until newton | 18:23 |
| odyssey4me | so basically until logan- put the effort in to implement it containerised, no-one else did | 18:23 |
| *** ellopunk has joined #openstack-ansible | 18:23 | |
| *** gokhan is now known as Guest94774 | 18:23 | |
| mnaser | okay i see, any reason in particular the switch to containerized by default has not been done? | 18:24 |
| *** Guest53306 has quit IRC | 18:24 | |
| odyssey4me | mnaser nope, no-one's ever put a patch up to do it I guess - possibly due to the extended consideration for documentation, upgrades, etc | 18:24 |
| mnaser | yeah i can imagine it being a wild thing that might cause an outage | 18:25 |
| odyssey4me | similar to the ->q upgrade stuff in https://review.openstack.org/559721 changing a default can be complicated | 18:25 |
| odyssey4me | you'd probably have to add the containers, move the vip to one of those, then remove the old containers | 18:25 |
| odyssey4me | sorry, then remove the haproxy from the hosts it was on | 18:26 |
| odyssey4me | something like that I guess | 18:26 |
| logan- | it is a little tricky in the gate too | 18:26 |
| logan- | because of how the collapsed networking is, juggling the vips to get haproxy containerized can be a little challenge, but its probably doable now | 18:27 |
| logan- | i think https://github.com/openstack/openstack-ansible-lxc_container_create/commit/bb76ea23f5682242df66648c5afe0c34c7c49988 fixed one of the big blockers for containerized haproxy that i was hitting in https://review.openstack.org/#/c/429019/ | 18:29 |
| *** kstev has joined #openstack-ansible | 18:29 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Add nova/ironic/glance cross-service vars to group_vars/all https://review.openstack.org/559804 | 18:39 |
| *** chhagarw has quit IRC | 18:45 | |
| *** vincent has joined #openstack-ansible | 18:47 | |
| *** vincent is now known as Guest77189 | 18:47 | |
| *** hamza21 has joined #openstack-ansible | 18:47 | |
| *** Administrator__ has quit IRC | 18:50 | |
| openstackgerrit | Merged openstack/openstack-ansible master: Remove spurious VxLAN bridge IP from ceph osd hosts https://review.openstack.org/559776 | 18:51 |
| *** radeks has quit IRC | 18:51 | |
| *** radeks has joined #openstack-ansible | 18:51 | |
| openstackgerrit | Merged openstack/openstack-ansible-tests master: Use the right hypervisor name for ovh hosts https://review.openstack.org/559778 | 18:52 |
| *** kstev has quit IRC | 18:56 | |
| *** kstev has joined #openstack-ansible | 18:56 | |
| *** chhagarw has joined #openstack-ansible | 18:58 | |
| *** adcpdk has joined #openstack-ansible | 18:59 | |
| adcpdk | HI Team! I have a quick question regarding the supported hypervisor | 19:00 |
| *** sep has quit IRC | 19:02 | |
| *** sep has joined #openstack-ansible | 19:03 | |
| odyssey4me | logan- odd, I'm getting a failure importing dictutils, but only when running via ansible | 19:04 |
| odyssey4me | adcpdk what's the question | 19:04 |
| adcpdk | I was trying to add a compute host that would run on xen hypervisor 4.4, and selected nova-virt-type: kvm in opetnstack_user_config.yml. I was able to run the setup-hosts.yml playbook just fine, however when I tried to run # openstack-ansible setup-openstack.yml --skip-tags nova-key-distribute --limit compute-04 it failed | 19:05 |
| adcpdk | 19:05 | |
| adcpdk | fatal: [compute-04]: FAILED! => {"changed": false, "failed": true, "msg": "Unsupported Virt Type Provided [u'qemu', u'kvm', u'lxd', u'ironic', u'powervm']"} | 19:05 |
| odyssey4me | adcpdk and your question is? | 19:06 |
| adcpdk | does xen is a supported type of hypervisor for openstack? | 19:06 |
| odyssey4me | adcpdk no, as the error stated | 19:07 |
| odyssey4me | no-one has contributed any patches to make that work | 19:07 |
| adcpdk | I found some documentation that says that xen is supported | 19:07 |
| odyssey4me | by openstack sure, but not in OSA | 19:07 |
| adcpdk | https://books.google.ca/books?id=ZulODwAAQBAJ&pg=PA54&lpg=PA54&dq=nova_virt_type+xen&source=bl&ots=xGzaYAzB5M&sig=UKZIyMKNjSWZ__WElJs61iVjyPY&hl=en&sa=X&ved=0ahUKEwi0_v3llabaAhXD44MKHQXGB7AQ6AEIMTAB#v=onepage&q&f=false | 19:07 |
| odyssey4me | all it takes is figuring out the right config options, then pushing a patch up | 19:07 |
| adcpdk | hm, so I can't use this book as a reference I beleive? | 19:08 |
| odyssey4me | adcpdk it's a reference for openstack - but OSA is a deployment system which deploys whatever its contributors have made to work | 19:09 |
| odyssey4me | you should be able to make it work with the config_override mechanism | 19:09 |
| *** marwel has quit IRC | 19:10 | |
| adcpdk | I will try to read about config_override | 19:11 |
| adcpdk | I have never done it yet, and don't have much of exerience with patching | 19:11 |
| odyssey4me | adcpdk https://docs.openstack.org/project-deploy-guide/openstack-ansible/pike/app-advanced-config-override.html and https://evrard.me/group-and-host-variables-overriding-in-openstack-ansible.html should give you enough to figure it out | 19:12 |
| *** radeks has quit IRC | 19:12 | |
| *** marwel has joined #openstack-ansible | 19:12 | |
| *** radeks has joined #openstack-ansible | 19:13 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 19:13 |
| *** mbuil has quit IRC | 19:19 | |
| *** Administrator__ has joined #openstack-ansible | 19:26 | |
| *** Guest77189 has quit IRC | 19:29 | |
| adcpdk | odyssey4me thank you I am reading upon that | 19:30 |
| *** radeks has quit IRC | 19:31 | |
| *** marwel has quit IRC | 19:31 | |
| *** marwel has joined #openstack-ansible | 19:31 | |
| *** poopcat has quit IRC | 19:31 | |
| *** radeks has joined #openstack-ansible | 19:31 | |
| *** armaan has joined #openstack-ansible | 19:34 | |
| *** marwel has quit IRC | 19:37 | |
| *** marwel has joined #openstack-ansible | 19:39 | |
| *** mwe1 has quit IRC | 19:44 | |
| *** mwe1 has joined #openstack-ansible | 19:45 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Fix osa_toolkit dictutils import https://review.openstack.org/559823 | 19:46 |
| odyssey4me | admin0 evrardjp ^ found where the bug was causing the import fail | 19:47 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-tests stable/queens: Use the right hypervisor name for ovh hosts https://review.openstack.org/559785 | 19:48 |
| adcpdk | odyssey4me I have prepared the config, but still got the same error https://paste.ubuntu.com/p/xwPT5qrhP6/ | 20:00 |
| adcpdk | fatal: [compute-04]: FAILED! => {"changed": false, "failed": true, "msg": "Unsupported Virt Type Provided [u'qemu', u'kvm', u'lxd', u'ironic', u'powervm']"} | 20:00 |
| *** DanyC_ has quit IRC | 20:01 | |
| *** DanyC has joined #openstack-ansible | 20:01 | |
| *** DanyC has quit IRC | 20:02 | |
| *** DanyC has joined #openstack-ansible | 20:02 | |
| *** vincent has joined #openstack-ansible | 20:09 | |
| *** vincent is now known as Guest2010 | 20:09 | |
| odyssey4me | adcpdk can you share the right configuration with me and I'll prep a patch to add support for it | 20:10 |
| *** Administrator__ has quit IRC | 20:11 | |
| *** mhayden has quit IRC | 20:13 | |
| *** Guest2010 has quit IRC | 20:15 | |
| *** mhayden has joined #openstack-ansible | 20:15 | |
| *** Guest2010 has joined #openstack-ansible | 20:15 | |
| *** b1tsh1ft3r has joined #openstack-ansible | 20:20 | |
| *** poopcat has joined #openstack-ansible | 20:20 | |
| *** vakuznet has quit IRC | 20:23 | |
| adcpdk | you mean this? /etc/openstack_deploy/openstack_user_config.yml | 20:25 |
| adcpdk | odyssey4me | 20:25 |
| adcpdk | https://paste.ubuntu.com/p/HXhmdXzsg3/ | 20:27 |
| adcpdk | here is my openstack_user_config.yml configuration | 20:27 |
| *** esberglu has quit IRC | 20:27 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Add container/service clean up for nova/cinder/heat https://review.openstack.org/559721 | 20:28 |
| *** esberglu has joined #openstack-ansible | 20:28 | |
| odyssey4me | adcpdk no, I mean if you can work out the right nova.conf settings and whatever needs to be installed to get xeniapi to work, then we can add those in to patch the nova role to accept them | 20:29 |
| odyssey4me | at the moment, OSA does not accept the xen hypervisor as a supported hypervisor because we have no such configuration | 20:29 |
| b1tsh1ft3r | Hey, extremely unsure where or what to check here. Im receiving the following error which is roadblocking me from getting a horizon lxc-container up and running again | 20:29 |
| b1tsh1ft3r | failed: [infra1_horizon_container-f731b502] (item=/openstack/venvs/horizon-15.1.13/bin/horizon-manage.py compress --force) => {"changed": false, "failed": true, "item": "/openstack/venvs/horizon-15.1.13/bin/horizon-manage.py compress --force", "module_stderr": "Shared connection to 10.250.0.200 closed.\r\n", "module_stdout": "/bin/sh: fork: retry: No child processes\r\n/bin/sh: fork: retry: No child processes\r\n/bin/sh: f | 20:29 |
| *** esberglu_ has joined #openstack-ansible | 20:30 | |
| odyssey4me | b1tsh1ft3r looks like ansible is struggling - got enough ram on your deploy node? | 20:30 |
| odyssey4me | are you trying to use a lot of forks? | 20:30 |
| b1tsh1ft3r | Not on purpose heh. Just running the play and receiving that message | 20:31 |
| *** esbergl__ has joined #openstack-ansible | 20:31 | |
| b1tsh1ft3r | checking resources on the deploy host though now that you mention it. | 20:31 |
| *** esbergl__ is now known as esberglu__ | 20:32 | |
| adcpdk | odyssey4me it's done using libvirt and not xenpapi. From my inderstanding of the article I have sent you earlier only the following needs to be added to the nova.conf: | 20:32 |
| adcpdk | https://docs.openstack.org/ocata/config-reference/compute/hypervisor-xen-libvirt.html | 20:32 |
| *** hw_wutianwei1 has joined #openstack-ansible | 20:32 | |
| odyssey4me | b1tsh1ft3r you'll need quite a bit of ram to fit all the inventory stuff into memory | 20:32 |
| adcpdk | https://docs.openstack.org/ocata/config-reference/compute/hypervisor-xen-libvirt.html#configuring-xen-with-libvirt | 20:32 |
| *** esberglu has quit IRC | 20:33 | |
| adcpdk | odyssey4me in here I also see the same iconfig for nova.conf https://wiki.xenproject.org/wiki/OpenStack | 20:33 |
| *** hw_wutianwei has quit IRC | 20:34 | |
| *** hw_wutianwei1 is now known as hw_wutianwei | 20:34 | |
| odyssey4me | adcpdk updated guide here: https://docs.openstack.org/nova/latest/admin/configuration/hypervisor-xen-libvirt.html | 20:34 |
| *** esberglu_ has quit IRC | 20:34 | |
| *** savvas_ has joined #openstack-ansible | 20:34 | |
| adcpdk | ok | 20:35 |
| odyssey4me | ok, I can push up a patch to try out - but can't promise anything - you'll have to test and provide feedback | 20:35 |
| adcpdk | this is my user_variables file: https://paste.ubuntu.com/p/7CNByXRCC9/ | 20:36 |
| *** pcaruana has quit IRC | 20:36 | |
| adcpdk | that is amazing | 20:37 |
| adcpdk | thank you | 20:37 |
| *** esberglu__ is now known as esberglu | 20:37 | |
| adcpdk | let me know when it's rready and what I need to do | 20:37 |
| *** savvas has quit IRC | 20:37 | |
| adcpdk | e.g. download some git repo | 20:38 |
| evrardjp[m] | odyssey4me yes I know, that's the first thing I said. But I think we should properly package this | 20:38 |
| evrardjp[m] | (the dictutils thing) | 20:38 |
| b1tsh1ft3r | @odyssey4me deploy host is fine with cpu and ram. Only seems to happen on the following task: [os_horizon : Collect and compress static files] | 20:39 |
| b1tsh1ft3r | also only on one lxc-container out of the 3 total horizon containers across 3 controllers | 20:40 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_nova master: [WIP] Add xen-libvirt support https://review.openstack.org/559834 | 20:41 |
| evrardjp[m] | omg pandoras box | 20:42 |
| odyssey4me | adcpdk open https://review.openstack.org/559834 -> download (top right) -> click on the 'copy to clipboard' for 'cherry pick' -> paste that into your os_nova role which should be in /etc/ansible/roles/ | 20:42 |
| evrardjp[m] | what have you done ? ;) | 20:42 |
| odyssey4me | evrardjp[m] package it or get rid of it ;) | 20:43 |
| odyssey4me | I mean seriously.... | 20:43 |
| evrardjp[m] | yes, i am writing the spec that will bring consensus i guess. | 20:45 |
| *** chhagarw has quit IRC | 20:45 | |
| evrardjp[m] | 3rd one is the deal. | 20:45 |
| odyssey4me | evrardjp[m] sure, so for now we can help the queens users out by merging my patch :p | 20:45 |
| odyssey4me | without that patch, the upgrade patch I submitted totally fails. :/ | 20:46 |
| evrardjp[m] | yes i will. i think it should be done differently but the gist is there. but i thinl there is another issue for some ppl. | 20:46 |
| odyssey4me | jmccrory you around perhaps? need another vote for https://review.openstack.org/559785 | 20:46 |
| evrardjp[m] | odyssey4me it didn't fail for me | 20:47 |
| odyssey4me | evrardjp[m] yeah, wierdly not for me the first time either | 20:47 |
| odyssey4me | but the second time on a clean install it did | 20:47 |
| evrardjp[m] | not surprised | 20:48 |
| odyssey4me | there seems to be some weird system pathing issue going on there, but the patch definitely does away with the problem as osa_toolkit is added to the system path | 20:48 |
| odyssey4me | also, every other lib in osa_toolkit uses the same from os_toolkit ... mechanism | 20:48 |
| evrardjp[m] | well that's the thing | 20:48 |
| evrardjp[m] | let's discuss that tomorrow please | 20:49 |
| odyssey4me | oh sure, of course | 20:49 |
| evrardjp[m] | yes we should have that | 20:49 |
| evrardjp[m] | that's the minimum | 20:49 |
| evrardjp[m] | better is to avoid from osa_toolkit import x | 20:49 |
| evrardjp[m] | but import osa_toolkit. | 20:49 |
| evrardjp[m] | x | 20:49 |
| evrardjp[m] | proves one more bad practice | 20:50 |
| *** ellopunk has quit IRC | 20:50 | |
| *** ansmith has quit IRC | 20:50 | |
| evrardjp[m] | and the cli should be separated and loading this, but tangled... anyway. tomorrow ! | 20:50 |
| evrardjp[m] | and we don't exercise this in tests... | 20:51 |
| adcpdk | odyssey4me this is what I got https://paste.ubuntu.com/p/xfmYXb8pXv/ | 20:52 |
| odyssey4me | adcpdk yes, you'll need to setup git | 20:53 |
| odyssey4me | it gives you clear enough intstructions | 20:53 |
| adcpdk | ok | 20:53 |
| *** b1tsh1ft3r has quit IRC | 20:54 | |
| adcpdk | I ave git configured with me email address https://paste.ubuntu.com/p/YjyY9XDyyn/ | 20:55 |
| odyssey4me | adcpdk what branch/tag of OSA are you trying to patch this onto? | 20:55 |
| odyssey4me | oh, and I see you've got a dirty stash | 20:56 |
| odyssey4me | so something's been customised there | 20:56 |
| *** hamza21 has quit IRC | 20:56 | |
| adcpdk | I can just updated the defaults/main.yml manually, no? | 20:56 |
| odyssey4me | git status to see what's changed, then git checkout --yeah, you could | 20:56 |
| adcpdk | git checkout -- defaults/main.yml worked | 20:58 |
| adcpdk | ok done | 20:59 |
| adcpdk | what should I do now? re-run the script? | 20:59 |
| *** Administrator__ has joined #openstack-ansible | 21:03 | |
| odyssey4me | adcpdk: cd /opt/openstack-ansible; openstack-ansible os-nova-install.yml | 21:04 |
| odyssey4me | although now I realise there are other missing bits | 21:04 |
| *** Guest2010 has quit IRC | 21:06 | |
| odyssey4me | adcpdk are you doing this on an ubuntu or some other distro host, or is this installed from the the cd/dvd source? | 21:12 |
| adcpdk | ubuntu 16.04.3 | 21:12 |
| adcpdk | ubuntu server 16.04.3 | 21:12 |
| odyssey4me | ok, I'm setting one up to see if I can get it working a little quicker | 21:13 |
| adcpdk | fatal: [compute-04]: FAILED! => {"failed": true, "msg": "'nova_firewall_driver' is undefined"} | 21:13 |
| adcpdk | just so you know, I do really appreciate the work you do for the community | 21:14 |
| adcpdk | thanks | 21:14 |
| adcpdk | this is the command I ran | 21:14 |
| adcpdk | ``` | 21:14 |
| adcpdk | ccn@kvm01:/opt/openstack-ansible/playbooks$ sudo openstack-ansible os-nova-install.yml --skip-tags nova-key-distribute --limit compute-04 | 21:14 |
| adcpdk | ``` | 21:14 |
| odyssey4me | adcpdk hmm, what branch/tag are you basing this from? | 21:14 |
| adcpdk | I believe 16.0.5 | 21:15 |
| odyssey4me | ah, pike - ok | 21:15 |
| adcpdk | yes | 21:15 |
| odyssey4me | similar to this, add that var to the defaults: https://github.com/openstack/openstack-ansible-os_nova/blob/stable/pike/defaults/main.yml#L148 | 21:15 |
| odyssey4me | that variable doesn't exist for queens/rocky | 21:16 |
| adcpdk | should I add it to xen hypervisor config as weell? | 21:18 |
| odyssey4me | adcpdk not sure what you mean by that? | 21:18 |
| adcpdk | /etc/ansible/roles/os_nova/defaults/main.yml should I change this file? | 21:19 |
| odyssey4me | yes, add that line I highlighted to the same new structure for xen | 21:19 |
| adcpdk | and just add nova_firewall_driver: nova.virt.firewall.NoopFirewallDriver line for xen: | 21:20 |
| odyssey4me | yes | 21:20 |
| adcpdk | ok | 21:20 |
| odyssey4me | fyi, the likelihood of this patch ever making it into pike is very, very low - nor queens... maybe rocky | 21:20 |
| odyssey4me | which means if you need this, you'll need to fork the repo and keep it updated - it's a bit of technical debt, but it is not uncommon to do this sort of thing | 21:21 |
| adcpdk | I see | 21:21 |
| adcpdk | so that means that OSA community is not going to add xen functionality infuture deployements as well? | 21:22 |
| odyssey4me | that said, we can try and make it work - then there will be support from rocky onwards | 21:22 |
| odyssey4me | you're the first to want xen support ;) | 21:22 |
| adcpdk | ))) | 21:22 |
| adcpdk | sorry | 21:22 |
| adcpdk | I think I will need to add this one as well nova_scheduler_use_baremetal_filters | 21:23 |
| odyssey4me | and OSA is a community of contributors who have their own needs to meet | 21:23 |
| odyssey4me | ah yes, that seems sensible | 21:23 |
| odyssey4me | good thinking :) | 21:23 |
| adcpdk | I see now | 21:24 |
| adcpdk | XEN makes a lot of sense from my POV, it provides almost as native hardware performance | 21:25 |
| adcpdk | there is another one | 21:25 |
| adcpdk | fatal: [compute-04]: FAILED! => {"failed": true, "reason": "the file_name '/opt/openstack-ansible/playbooks/common-playbooks/drivers/xen/nova_compute_xen.yml' does not exist, or is not readable"} | 21:25 |
| odyssey4me | yeah, I thought that might happen | 21:26 |
| adcpdk | is it related to my user_variables.yml? | 21:27 |
| odyssey4me | in the os_nova role again, try symlink drivers/xen to the drivers/kvm folder | 21:27 |
| odyssey4me | I don't think it'll work, but it's worth a try | 21:27 |
| adcpdk | wait is it drivers to interact with lbvirt-xen | 21:27 |
| odyssey4me | one of the problems with adding xen support is that we have no way to really test it in a cloud, which is what all openstack tests on | 21:28 |
| tomtom001 | cloudnull, hey thanks for the info! I watched the log as it was running and the process still just seems to die in the middle of the apt-get install -y. Disk space does not seem to be an issue, but this process does seem to take more than an hour to "complete" (which means just die). I tried to adjust the timeout and deleted ansible facts and all that but no matter what I do I can't seem to | 21:29 |
| tomtom001 | get past this issue. I'm stumped unless you have some other advice. What else do you nuke to get this working? | 21:29 |
| odyssey4me | well, what's happening now is that ansible is looking for the tasks to do the hypervisor-specific things it needs... and we haven't added any | 21:29 |
| odyssey4me | tomtom001 do you perhaps have a really, really slow mirror? | 21:29 |
| odyssey4me | or bad connection or whatever | 21:29 |
| odyssey4me | if so, you can override the timeout: https://github.com/openstack/openstack-ansible-lxc_hosts/blob/master/defaults/main.yml#L170-L174 | 21:30 |
| odyssey4me | 20 mins is a long time though | 21:30 |
| adcpdk | I don't have the drivers folder under /opt/openstack-ansible/playbooks/common-playbooks/ | 21:30 |
| odyssey4me | adcpdk no, not there - it's in the /etc/ansible/roles/os_nova/tasks folder | 21:31 |
| tomtom001 | Maybe, but I'm not like selecting a mirror, I'm just using whatever came with the git pull stable/pike. How is the mirror chosen? | 21:31 |
| odyssey4me | tomtom001 the default will be to use whatever's configured on the host | 21:31 |
| tomtom001 | Ok, so it's in /etc/apt? | 21:32 |
| odyssey4me | tomtom001 /etc/apt/sources.list | 21:32 |
| *** vincent has joined #openstack-ansible | 21:33 | |
| tomtom001 | ok thanks I'll check that | 21:33 |
| *** vincent is now known as Guest14521 | 21:33 | |
| *** Administrator__ has quit IRC | 21:35 | |
| openstackgerrit | Merged openstack/openstack-ansible-tests stable/queens: Use the right hypervisor name for ovh hosts https://review.openstack.org/559785 | 21:37 |
| adcpdk | odyssey4me like this https://paste.ubuntu.com/p/HjxJ4QZh7N/ | 21:40 |
| adcpdk | ? | 21:40 |
| odyssey4me | adcpdk yeah, that's worth a try | 21:40 |
| adcpdk | still get this | 21:41 |
| adcpdk | fatal: [compute-04]: FAILED! => {"failed": true, "reason": "the file_name '/opt/openstack-ansible/playbooks/common-playbooks/drivers/xen/nova_compute_xen.yml' does not exist, or is not readable"} | 21:41 |
| odyssey4me | I have a host busy prepping to try the same out. | 21:41 |
| adcpdk | np | 21:41 |
| adcpdk | so that file /opt/openstack-ansible/playbooks/common-playbooks/drivers/xen/nova_compute_xen.yml | 21:41 |
| adcpdk | actually is not there | 21:41 |
| odyssey4me | Ah yes, you'll need another symlink from tasks/drivers/kvm/nova_compute_xen.yml to tasks/drivers/kvm/nova_compute_kvm.yml | 21:42 |
| adcpdk | ok | 21:42 |
| adcpdk | thanks | 21:42 |
| odyssey4me | yes, of course it's not because we're busy doing a new implementation for it ;) | 21:42 |
| adcpdk | looks good now | 21:47 |
| adcpdk | ran fine | 21:47 |
| *** ansmith has joined #openstack-ansible | 21:47 | |
| odyssey4me | so once it's done, the real question would be whether it works as expected | 21:50 |
| odyssey4me | cloudnull heh, I just did the aio bootstrap on a physical host and the dns is gone | 21:51 |
| odyssey4me | networking fine, but dns is kaput | 21:51 |
| odyssey4me | how should it be resolving now? | 21:51 |
| adcpdk | odyssey4me weird, I don't have the new compute node in openstack | 21:52 |
| odyssey4me | adcpdk check the nova logs on the node | 21:53 |
| adcpdk | https://docs.openstack.org/openstack-ansible/newton/developer-docs/ops-add-computehost.html that's the playbooks I used | 21:53 |
| openstackgerrit | Merged openstack/openstack-ansible master: Use a sensible vlan range in the example configs https://review.openstack.org/559763 | 21:57 |
| odyssey4me | if any cores are around, the earlier discussions related to dictutils import failures are fixed by https://review.openstack.org/559823 | 21:58 |
| *** kstev has quit IRC | 21:59 | |
| *** Guest14521 has quit IRC | 22:00 | |
| *** Guest14521 has joined #openstack-ansible | 22:00 | |
| adcpdk | should I restart the controller? | 22:05 |
| adcpdk | nova is not running on cumpute node | 22:07 |
| adcpdk | Restarted after I ran all the playbooks | 22:07 |
| adcpdk | odyssey4me look at the driver section, number 12 and 13 https://books.google.ca/books?id=ZulODwAAQBAJ&pg=PA54&lpg=PA54&dq=nova_virt_type+xen&source=bl&ots=xGzaYAzB5M&sig=UKZIyMKNjSWZ__WElJs61iVjyPY&hl=en&sa=X&ved=0ahUKEwi0_v3llabaAhXD44MKHQXGB7AQ6AEIMTAB#v=onepage&q&f=false | 22:09 |
| adcpdk | page55 | 22:09 |
| *** ianychoi has quit IRC | 22:10 | |
| *** ianychoi has joined #openstack-ansible | 22:11 | |
| odyssey4me | adcpdk I have no access to it :/ | 22:12 |
| *** DanyC has quit IRC | 22:14 | |
| *** DanyC has joined #openstack-ansible | 22:15 | |
| *** Administrator__ has joined #openstack-ansible | 22:15 | |
| *** savvas_ has quit IRC | 22:17 | |
| *** Guest14521 has quit IRC | 22:19 | |
| adcpdk | Should I use juno release that mentioned in that book maybe? | 22:19 |
| *** DanyC has quit IRC | 22:19 | |
| adcpdk | hold on | 22:19 |
| adcpdk | I have the new compute host | 22:19 |
| odyssey4me | adcpdk oh no, definitely not juno | 22:21 |
| odyssey4me | and yes, it may take some time for a new compute to show | 22:21 |
| odyssey4me | I'm out for the night - let me know if things are working right, nor not. | 22:21 |
| odyssey4me | I'll update the patch tomorrow based on your feedback. | 22:21 |
| adcpdk | how can I choose between hypervisors? | 22:21 |
| adcpdk | I can see the new compute host, but I will test it and let you know | 22:22 |
| odyssey4me | I expect you may need to do some image setup, and maybe hypervisor | 22:22 |
| odyssey4me | not sure how it all works, you'll have to figure it out and find docs | 22:22 |
| odyssey4me | night all | 22:22 |
| adcpdk | good night | 22:23 |
| *** esberglu has quit IRC | 22:27 | |
| adcpdk | the instance failed due to some issues with networking: https://paste.ubuntu.com/p/Rdkrwkc973/ | 22:28 |
| *** esberglu has joined #openstack-ansible | 22:28 | |
| *** esberglu_ has joined #openstack-ansible | 22:29 | |
| *** esberglu has quit IRC | 22:33 | |
| *** esberglu_ has quit IRC | 22:34 | |
| *** dave-mcc_ has quit IRC | 22:49 | |
| *** esberglu has joined #openstack-ansible | 22:50 | |
| *** throwsb1 has quit IRC | 22:58 | |
| *** lbragstad has quit IRC | 23:17 | |
| *** radeks has quit IRC | 23:22 | |
| *** radeks has joined #openstack-ansible | 23:30 | |
| *** radeks has quit IRC | 23:34 | |
| *** lbragstad has joined #openstack-ansible | 23:36 | |
| *** cjloader has quit IRC | 23:39 | |
| *** mwe1 has quit IRC | 23:46 | |
| *** lbragstad has quit IRC | 23:47 | |
| *** mwe1 has joined #openstack-ansible | 23:47 | |
| *** exodusftw has quit IRC | 23:58 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
