| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add verification support to the agent https://review.openstack.org/120627 | 00:08 |
|---|---|---|
| *** harlowja_ has quit IRC | 00:25 | |
| *** yuanying has quit IRC | 00:25 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/120637 | 00:25 |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/114058 | 00:25 |
| *** yuanying has joined #openstack-ironic | 00:25 | |
| *** dlaube has quit IRC | 00:25 | |
| *** harlowja has joined #openstack-ironic | 00:28 | |
| *** dlaube has joined #openstack-ironic | 00:30 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/python-ironicclient: Updated from global requirements https://review.openstack.org/120651 | 00:31 |
| jroll | devananda: around? | 00:35 |
| jroll | .b 8 | 00:36 |
| jroll | oops | 00:36 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add docstrings for agent.py https://review.openstack.org/120660 | 00:41 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add docstrings for api/ tree https://review.openstack.org/120661 | 00:41 |
| jroll | whee. | 00:41 |
| *** Haomeng has quit IRC | 00:42 | |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add docstrings for api/ tree https://review.openstack.org/120661 | 00:44 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add docstrings for agent.py https://review.openstack.org/120660 | 00:44 |
| *** Haomeng has joined #openstack-ironic | 00:45 | |
| openstackgerrit | Jay Faulkner proposed a change to openstack/ironic-python-agent: Improve docstrings for cmd/agent.py https://review.openstack.org/120662 | 00:46 |
| openstackgerrit | Jay Faulkner proposed a change to openstack/ironic-python-agent: Improve/add docstrings for CommandResult classes https://review.openstack.org/120663 | 00:46 |
| JayF | rloo: ^ especially relevant to your interests :) although we still have a long way to go | 00:46 |
| JayF | Ironic folk ^^ please look over these and review them. I have a strong preference for people who haven't worked much with IPA to review the docstrings to ensure they're clear and complete. | 00:47 |
| jroll | IPA folk, quick review :) https://review.openstack.org/#/c/120586/4 | 00:47 |
| NobodyCam | JayF: can I loo it over in the morning | 00:48 |
| JayF | No rush whatsoever | 00:48 |
| JayF | we're just going through these adding docstrings | 00:48 |
| NobodyCam | :) | 00:48 |
| *** dlaube has quit IRC | 01:31 | |
| *** Hefeweizen has quit IRC | 01:33 | |
| *** rainya has quit IRC | 01:42 | |
| *** rainya has joined #openstack-ironic | 01:43 | |
| *** nosnos has joined #openstack-ironic | 01:47 | |
| *** scubacuda has quit IRC | 01:47 | |
| rloo | JayF: Thanks. I was looking for bedtime reading ;) | 01:47 |
| *** Nisha has quit IRC | 01:58 | |
| *** killer_prince is now known as lazy_prince | 02:05 | |
| *** rloo has quit IRC | 02:09 | |
| *** rainya has quit IRC | 02:31 | |
| *** rainya has joined #openstack-ironic | 02:36 | |
| openstackgerrit | Motohiro Otsuka proposed a change to openstack/ironic: Search line with awk itself and avoid grep https://review.openstack.org/118966 | 02:43 |
| *** ramineni has joined #openstack-ironic | 02:44 | |
| *** rushiagr_away is now known as rushiagr | 02:53 | |
| ramineni | JayF: Hi | 03:03 |
| *** lazy_prince is now known as killer_prince | 03:04 | |
| *** rushiagr is now known as rushiagr_away | 03:14 | |
| *** Poornima has joined #openstack-ironic | 03:29 | |
| *** rushiagr_away is now known as rushiagr | 03:46 | |
| *** harlowja is now known as harlowja_away | 04:05 | |
| *** krtaylor has joined #openstack-ironic | 04:11 | |
| *** rainya has quit IRC | 04:29 | |
| *** Poornima has quit IRC | 04:31 | |
| *** HenryG is now known as HenryG_zzzz | 04:40 | |
| *** rakesh_hs has joined #openstack-ironic | 04:49 | |
| *** chuckC has quit IRC | 04:57 | |
| *** Poornima has joined #openstack-ironic | 04:57 | |
| *** pcrews has quit IRC | 05:07 | |
| *** chuckC has joined #openstack-ironic | 05:27 | |
| *** killer_prince is now known as lazy_prince | 05:36 | |
| *** jcoufal has joined #openstack-ironic | 05:59 | |
| *** nikunj2512 has joined #openstack-ironic | 06:04 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/120701 | 06:12 |
| *** Nisha has joined #openstack-ironic | 06:23 | |
| *** k4n0 has joined #openstack-ironic | 06:34 | |
| *** ccit has joined #openstack-ironic | 06:41 | |
| *** ccit has quit IRC | 07:09 | |
| *** takadayuiko has joined #openstack-ironic | 07:09 | |
| *** dtantsur|afk is now known as dtantsur | 07:12 | |
| dtantsur | Morning Ironic | 07:12 |
| *** todd_dsm has joined #openstack-ironic | 07:12 | |
| openstackgerrit | A change was merged to openstack/ironic-specs: Add RSS feed https://review.openstack.org/120516 | 07:13 |
| openstackgerrit | A change was merged to openstack/ironic-specs: Remove templates from toctrees https://review.openstack.org/120517 | 07:13 |
| openstackgerrit | A change was merged to openstack/ironic-specs: Use the current date for the copyright statement https://review.openstack.org/120518 | 07:13 |
| *** ifarkas has joined #openstack-ironic | 07:14 | |
| openstackgerrit | Yuiko Takada proposed a change to openstack/python-ironicclient: Add unit tests for "ironic node-delete" shell cmd https://review.openstack.org/120710 | 07:22 |
| *** notq has quit IRC | 07:24 | |
| *** ndipanov has joined #openstack-ironic | 07:33 | |
| *** jistr has joined #openstack-ironic | 07:39 | |
| *** faizan has joined #openstack-ironic | 07:44 | |
| *** jcoufal has quit IRC | 07:46 | |
| *** faizan has quit IRC | 07:46 | |
| *** bluex-pl has joined #openstack-ironic | 07:55 | |
| *** andreykurilin has joined #openstack-ironic | 07:56 | |
| *** ifarkas has quit IRC | 08:00 | |
| *** ifarkas has joined #openstack-ironic | 08:01 | |
| *** MattMan has joined #openstack-ironic | 08:12 | |
| *** lucasagomes has joined #openstack-ironic | 08:20 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add list_by_node_id() to Port object https://review.openstack.org/120422 | 08:26 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove the objectify decorator https://review.openstack.org/120489 | 08:26 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add reserve() and release() to Node object https://review.openstack.org/120488 | 08:26 |
| *** derekh has joined #openstack-ironic | 08:29 | |
| *** athomas has joined #openstack-ironic | 08:30 | |
| *** igordcard has joined #openstack-ironic | 08:30 | |
| openstackgerrit | Szymon Wróblewski proposed a change to openstack/ironic-python-agent: Enhance decorators in agent and automate creation of command_map https://review.openstack.org/117529 | 08:49 |
| *** viktors|afk is now known as viktors | 08:53 | |
| *** andreykurilin has quit IRC | 08:55 | |
| *** chenglch|2 has joined #openstack-ironic | 09:04 | |
| *** dtantsur is now known as dtantsur|afk | 09:07 | |
| *** andreykurilin has joined #openstack-ironic | 09:08 | |
| *** bluex-pl has quit IRC | 09:19 | |
| *** notq has joined #openstack-ironic | 09:19 | |
| *** jcoufal has joined #openstack-ironic | 09:20 | |
| openstackgerrit | A change was merged to openstack/ironic: Search line with awk itself and avoid grep https://review.openstack.org/118966 | 09:22 |
| *** athomas has quit IRC | 09:30 | |
| *** athomas has joined #openstack-ironic | 09:39 | |
| *** pensu has joined #openstack-ironic | 09:40 | |
| *** pelix has joined #openstack-ironic | 09:52 | |
| *** faizan has joined #openstack-ironic | 09:58 | |
| *** yuanying has quit IRC | 10:00 | |
| *** rakesh_hs2 has joined #openstack-ironic | 10:06 | |
| *** rakesh_hs has quit IRC | 10:06 | |
| *** d0ugal has quit IRC | 10:08 | |
| *** d0ugal has joined #openstack-ironic | 10:08 | |
| *** Nisha has quit IRC | 10:09 | |
| *** athomas_ has joined #openstack-ironic | 10:17 | |
| *** athomas has quit IRC | 10:20 | |
| *** chenglch|2 has quit IRC | 10:20 | |
| *** Mike_D_wk has joined #openstack-ironic | 10:33 | |
| *** takadayuiko has quit IRC | 10:41 | |
| *** athomas_ has quit IRC | 10:43 | |
| *** athomas has joined #openstack-ironic | 10:43 | |
| *** athomas has quit IRC | 10:46 | |
| *** nikunj2512 has quit IRC | 10:47 | |
| pensu | Hi all, could you please share your views on this: http://lists.openstack.org/pipermail/openstack-dev/2014-September/045702.html | 10:48 |
| lucasagomes | pensu, hi there, just answered that | 10:53 |
| lucasagomes | pensu, basically docker.io is not present in precise :( and the -infra team is now using Trusty in gate | 10:54 |
| lucasagomes | so it won't be a problem | 10:54 |
| pensu | lucasagomes: thanks....:) | 10:55 |
| lucasagomes | we def could try to make the devstack+ironic work on precise thou (for e.g just try install docker if the IPA driver is enabled) | 10:55 |
| lucasagomes | pensu, if you think that's a problem, please file a bug about it and we can take a look | 10:55 |
| pensu | lucasagomes: so basically I am supposed to try that guide with trusty not precise, right? | 10:56 |
| lucasagomes | pensu, yeah, that's the version it's being tested on gate | 10:56 |
| pensu | lucasagomes: yeah, I did install docker separately on precise... | 10:56 |
| pensu | lucasagomes: but I guess the problem is that it is looking for 'docker.io'.....not sure where to get it.. | 10:56 |
| lucasagomes | pensu, I see... yeah but that would require you to tweak the devstack as well | 10:57 |
| lucasagomes | pensu, e.g remove the docker.io line from files/apts/ironic | 10:57 |
| lucasagomes | that would make devstack stop trying to install it | 10:57 |
| pensu | lucasagomes: hmmm.....that could work, let me try it....but won't it be a problem again when devstack try to use it.... | 10:58 |
| pensu | lucasagomes: I mean if it is trying to install docker.io, it will also try to use docker.io, right? | 10:58 |
| lucasagomes | pensu, so docker will only be used if you are testing IPA | 10:58 |
| lucasagomes | the IPA driver (which is not the default) | 10:58 |
| lucasagomes | afaiui | 10:58 |
| lucasagomes | pensu, not by default... that's why I think that, a fix for it would to have that conditional == for e.g just try install docker if the IPA driver is enabled | 10:59 |
| pensu | lucasagomes: sounds good.....let me check, I will get back to you...thanks again....:) | 10:59 |
| lucasagomes | pensu, no worries, yvw | 10:59 |
| *** yuanying has joined #openstack-ironic | 11:00 | |
| *** athomas has joined #openstack-ironic | 11:03 | |
| *** yuanying has quit IRC | 11:05 | |
| *** ramineni has quit IRC | 11:08 | |
| *** rakesh_hs has joined #openstack-ironic | 11:09 | |
| *** rakesh_hs2 has quit IRC | 11:09 | |
| *** Haomeng has quit IRC | 11:21 | |
| *** yuanying has joined #openstack-ironic | 11:21 | |
| *** Haomeng has joined #openstack-ironic | 11:26 | |
| *** dtantsur|afk is now known as dtantsur | 11:26 | |
| *** yuanying has quit IRC | 11:27 | |
| *** pensu has quit IRC | 11:50 | |
| NobodyCam | Goood Morning Ironic says the zombe walking in the door | 12:09 |
| openstackgerrit | Rakesh H S proposed a change to openstack/python-ironicclient: fixes help string for driver-list https://review.openstack.org/120764 | 12:10 |
| *** dprince has joined #openstack-ironic | 12:14 | |
| lucasagomes | morning NobodyCam | 12:18 |
| openstackgerrit | Rakesh H S proposed a change to openstack/python-ironicclient: handles keyboard interrupt https://review.openstack.org/120767 | 12:21 |
| *** nosnos has quit IRC | 12:22 | |
| *** aswadr has joined #openstack-ironic | 12:22 | |
| *** nosnos has joined #openstack-ironic | 12:22 | |
| *** yuanying has joined #openstack-ironic | 12:23 | |
| NobodyCam | morning lucasagomes | 12:25 |
| NobodyCam | lucasagomes: gots a second to take a look at : https://wiki.openstack.org/wiki/Ironic/NovaBaremetalIronicMigration | 12:26 |
| *** nosnos has quit IRC | 12:27 | |
| lucasagomes | NobodyCam, sure, btw wanted to talk to you about the proxy thing as well | 12:29 |
| lucasagomes | NobodyCam, are u planning to add unittest to it? | 12:29 |
| *** ramineni has joined #openstack-ironic | 12:29 | |
| NobodyCam | lol do i have to, do they actually want it | 12:29 |
| NobodyCam | I have yet to check the ml | 12:30 |
| lucasagomes | NobodyCam, yeah seems so dan smith was talking about it there | 12:30 |
| dtantsur | NobodyCam, morning :) aren't you early today? | 12:32 |
| NobodyCam | morning dtantsur , yep one of hte "kids" is not feeling well so its been up and down all night, this time I just made coffee and forgot about the down bits | 12:34 |
| dtantsur | oh sorry to hear :( | 12:35 |
| NobodyCam | :( how to hear back from the the vet today :( | 12:35 |
| NobodyCam | s/how/hope. | 12:35 |
| NobodyCam | so tempest tests too | 12:36 |
| NobodyCam | ugggg | 12:36 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Make the context parameter of create() mandatory on Objects https://review.openstack.org/120773 | 12:37 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: _from_db_object() to set the context to the RPC objects https://review.openstack.org/120774 | 12:37 |
| NobodyCam | i was thinking about unit test to see if the calls where in fact made, but temptest | 12:37 |
| NobodyCam | I have not messed with tempest | 12:37 |
| *** pensu has joined #openstack-ironic | 12:38 | |
| lucasagomes | NobodyCam, no tempest I think | 12:38 |
| lucasagomes | dan smith was against it, and I don't see reason for that as well | 12:38 |
| lucasagomes | well sean said something about it, but really? I think we may be able to push back on tempest | 12:39 |
| NobodyCam | other then the test, did you see anything needing changing on the code? | 12:40 |
| lucasagomes | NobodyCam, nop, since my last review | 12:41 |
| lucasagomes | lemme look the last patch changes | 12:41 |
| *** rloo has joined #openstack-ironic | 12:44 | |
| *** Poornima has quit IRC | 12:48 | |
| NobodyCam | dtantsur: while I'm can I get your thoughts on the migratation wiki thing: https://wiki.openstack.org/wiki/Ironic/NovaBaremetalIronicMigration | 12:53 |
| * dtantsur is looking | 12:53 | |
| NobodyCam | humm /me needs more coffee | 12:53 |
| NobodyCam | poors second cup, 6 more and I'll be good to go | 12:54 |
| lucasagomes | NobodyCam, posted some nits in the patch, no vote... see what you think | 12:54 |
| NobodyCam | ack | 12:54 |
| lucasagomes | NobodyCam, also lemme know if I can help with the unittests thing | 12:54 |
| lucasagomes | I would glad help with that | 12:54 |
| NobodyCam | looking.. Oh I always need help with tests... | 12:55 |
| * NobodyCam usaly writes poor tests | 12:55 | |
| dtantsur | NobodyCam, "Deploy-ironic kernel and ramdisk built (DIB)" maybe some link here? or is it supposed to be obvious? | 12:55 |
| *** HenryG_zzzz is now known as HenryG | 12:55 | |
| lucasagomes | NobodyCam, ack, I will work on that unittest then | 12:55 |
| dtantsur | NobodyCam, IMO "Deploy-ironic kernel and ramdisk built (DIB)" and " Empty Ironic database exists" should be in "High Level Walk Through" with appropriate commands | 12:56 |
| *** yuanying has quit IRC | 12:56 | |
| dtantsur | NobodyCam, nova flavor-key baremetal set "baremetal:deploy_kernel_id"="$dkid" <-- are we suggesting deprecated behavior? may make sense, but at least we need a note | 12:57 |
| NobodyCam | lucasagomes: awesome Thank you.. I'll agree with your comments and will get those pushed up | 12:57 |
| lucasagomes | NobodyCam, ack... I'll give the unittests a first stab here | 12:58 |
| NobodyCam | dtantsur: good catch will add build K&R and creating DB | 12:59 |
| lucasagomes | NobodyCam, btw, will add another comment about the exception we are raising there | 12:59 |
| lucasagomes | from webob | 12:59 |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Add uefi boot mode support in IloVirtualMediaIscsiDeploy https://review.openstack.org/116561 | 13:00 |
| lucasagomes | NobodyCam, oh no, forget... thought other contrib modules where using nova exceptions instead | 13:00 |
| NobodyCam | dtantsur: ahh thats just some code I had in my snibits file I tossed up there | 13:00 |
| NobodyCam | snippet even | 13:01 |
| NobodyCam | I got that execption from that file it self | 13:02 |
| NobodyCam | didn't want to make "MAJOR" change | 13:02 |
| openstackgerrit | Derek Higgins proposed a change to openstack/ironic: Revert "Search line with awk itself and avoid grep" https://review.openstack.org/120783 | 13:12 |
| *** Ugallu has joined #openstack-ironic | 13:13 | |
| dtantsur | one more proof we need tripleo ci voting ^^^ | 13:16 |
| derekh | lucasagomes: http://git.openstack.org/cgit/openstack-infra/tripleo-ci/tree/elements/testenv-worker/bin/ci_commands | 13:16 |
| dtantsur | derekh, is problem in 'grep' command? | 13:17 |
| rloo | I think we broke tripleo ci once before in a similar fashion :-( | 13:18 |
| derekh | dtantsur: yup, we have tight rules on what commands can be run over ssh, so the test environments have to be updated to allow any new commands to be run | 13:18 |
| NobodyCam | morning rloo | 13:18 |
| derekh | rloo: yup, this is the second time | 13:18 |
| *** linggao has joined #openstack-ironic | 13:19 | |
| NobodyCam | and derekh | 13:19 |
| rloo | sorry derekh | 13:19 |
| rloo | bonjour NobodyCam et tous les autres Ironickers | 13:19 |
| NobodyCam | key | 13:19 |
| NobodyCam | s/key/kay/ | 13:19 |
| NobodyCam | heheheh | 13:19 |
| derekh | rloo: no prob at all these things happen, once we have a new regex in http://git.openstack.org/cgit/openstack-infra/tripleo-ci/tree/elements/testenv-worker/bin/ci_commands I can do a redeploy, | 13:20 |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Add uefi boot mode support in IloVirtualMediaIscsiDeploy https://review.openstack.org/116561 | 13:20 |
| derekh | but unfortunatly that takes down CI for about an hour (assuming it all goes well) | 13:20 |
| rloo | derekh: so you need us to undo that change? | 13:21 |
| NobodyCam | lucasagomes: ok comments addresses | 13:21 |
| NobodyCam | addressed even | 13:21 |
| lucasagomes | NobodyCam, https://review.openstack.org/#/c/120783/ | 13:22 |
| lucasagomes | broke TripleO CI | 13:22 |
| derekh | rloo: ideally yes, then we add the new command into the list of approved command, then I'll reploy testenvs and finally remerge the change | 13:22 |
| *** dlaube has joined #openstack-ironic | 13:22 | |
| rloo | derekh: ok, approved by us humans. Pray to Jenkins now ;) | 13:24 |
| NobodyCam | lol :( so someone tried to fix that command mess :( | 13:24 |
| NobodyCam | gah your quick rloo | 13:24 |
| NobodyCam | lol | 13:24 |
| NobodyCam | :) | 13:24 |
| rloo | NobodyCam: ;) | 13:24 |
| derekh | lucasagomes: rloo thanks | 13:25 |
| lucasagomes | derekh, yvw, sorry for breaking you guys | 13:25 |
| rloo | so derekh, every time we add/change an ssh command, we ought to ...? inform you guys, or submit a tripleo patch to ...? | 13:26 |
| NobodyCam | rloo: infra locks down the ssh command that can be run | 13:27 |
| derekh | rloo: submit a patch to https://github.com/openstack-infra/tripleo-ci/blob/master/elements/testenv-worker/bin/ci_commands#L64-L71 , once merged we have to redeploy test envs, then you can make the change | 13:27 |
| rloo | derekh: so ideally, you want us to do the patch to openstack-infra. by 'we have to redeploy', that means tripleo folks or infrra folks? | 13:28 |
| derekh | rloo: tripleo folks | 13:29 |
| rloo | derekh: ok, so 1. patch to openstack-infra; 2. ask tripleo via irc? to redeploy test envs; 3. land patch-that-has-ssh-command | 13:30 |
| *** pcrews has joined #openstack-ironic | 13:31 | |
| derekh | rloo: yup, thats pretty much it, unfortunately we can't loosen up the "allowed" commands too much, because we're give out root to a rack of baremetal machine's and gotta be careful nothing malicious is run | 13:31 |
| rloo | thx derekh | 13:33 |
| derekh | rloo: no prob at all, thanks back :-) | 13:34 |
| dtantsur | folks what do we need to make this build stable and voting? | 13:34 |
| rloo | lucasagomes, NobodyCam, etc: so on the ironic side, do we ask the person that submitted the patch, to go through the process of submitting an infra patch etc? | 13:34 |
| lucasagomes | rloo, hmm yeah, or we could do that | 13:35 |
| lucasagomes | rloo, I'm core on tripleo CI | 13:35 |
| lucasagomes | so I can +2 it and merge it easier there | 13:35 |
| lucasagomes | to make easier to get it merged there* | 13:36 |
| rloo | lucasagomes: so it seems like the process is a bit more involved. First, someone submits the ironic patch, after it gets 2 +2s (but before approving), the other stuff needs to be done, then the ironic patch can be approved. | 13:36 |
| lucasagomes | rloo, exactly, the ideal would to be the TripleO voting for us | 13:36 |
| lucasagomes | rloo, so if it's broken we couldn't merge the patch anyway | 13:37 |
| lucasagomes | rloo, after the patch merge on tripleO CI the test env needs to be rebuilt | 13:37 |
| rloo | lucasagomes: right. if it is voting we'll know it will break. but we'll still need to do the song-and-dance. | 13:37 |
| NobodyCam | we just need to keep that in mind when reviewing patches that attempt to alter any of (virsh) the ssh commands | 13:37 |
| openstackgerrit | A change was merged to openstack/ironic: Add list_by_node_id() to Port object https://review.openstack.org/120422 | 13:37 |
| lucasagomes | rloo, yeah :( | 13:37 |
| rloo | so lucasagomes, if an infra patch is submitted, you can +2 it, so you shouldn't submit it, right? | 13:38 |
| lucasagomes | NobodyCam, +1, we broke it once in the past (in the midcycle after I added get_boot_device to virsh heh) | 13:38 |
| lucasagomes | rloo, yeah, tripleo-ci more specifically. | 13:38 |
| lucasagomes | rloo, do you think it's cool to one of us to submit the patch there instead of the author? | 13:39 |
| rloo | lucasagomes: honestly. I'd prefer that the author did it. Well, really honestly, as long as I don't have to do it ;) | 13:40 |
| rloo | lucasagomes: I'm almost positive that the author would love someone else to do it ;) | 13:40 |
| lucasagomes | rloo, heh right, I dunno the author IRC name | 13:40 |
| NobodyCam | lucasagomes: I would say yes, as not all dev know how the CI system works, so they just don't know to make the change there too, and some dev and not in channel to tell them to do it | 13:40 |
| rloo | lucasagomes: cuz if I were the author, I would not care to do it! | 13:41 |
| lucasagomes | NobodyCam, yeah | 13:41 |
| rloo | lucasagomes: especially since it has to be done after we've OK'd their ironic change. | 13:41 |
| NobodyCam | if we want the change, I'm ok with making the CI change | 13:41 |
| lucasagomes | rloo, NobodyCam yeah you guys are right... alright, so maybe we should do it | 13:41 |
| lucasagomes | NobodyCam, I think it's a cool change, it's not important but well fix a bug targeting RC | 13:42 |
| rloo | ah, NobodyCam, are you volunteering to do the tripleo-ci patch? | 13:42 |
| NobodyCam | the patch? we need a patch, we are reverting the ssh change? | 13:42 |
| lucasagomes | NobodyCam, yeah, we are reverting it until we fix in TripleO | 13:42 |
| rloo | cuz I was thinking of doing it but wondering... when the tripleoCI patch is done, it has to support both ssh commands, and then after the ironic patch is merged, do another tripleoci patch to unsupport the original ssh command? | 13:43 |
| lucasagomes | NobodyCam, so having a patch to the CI would be cool so we can re-revert it | 13:43 |
| *** Haomeng has quit IRC | 13:43 | |
| lucasagomes | rloo, yup, pretty much | 13:43 |
| NobodyCam | ahhh | 13:43 |
| lucasagomes | but the second one doesn't require the env to be rebuilt | 13:43 |
| lucasagomes | security people security heh | 13:43 |
| rloo | NobodyCam: you still want to volunteer for this? | 13:44 |
| NobodyCam | I've patched that file before :-p | 13:44 |
| *** igordcard has quit IRC | 13:44 | |
| lucasagomes | NobodyCam, ta much | 13:45 |
| * NobodyCam is not sure he is awake enough to hack regex commands atm, but will look ina bit | 13:45 | |
| lucasagomes | derekh, do you have the exact command that needs to be whitelisted handy? | 13:45 |
| * lucasagomes goes to the logs | 13:45 | |
| rloo | thx NobodyCam! | 13:46 |
| derekh | lucasagomes: looks like this is it | 13:46 |
| derekh | /usr/bin/virsh --connect qemu:///system dumpxml baremetalbrbm1_6 | awk -F "'" '/mac address/{print $2}'| tr -d ':' | 13:46 |
| lucasagomes | derekh, ta much | 13:46 |
| lucasagomes | NobodyCam, ^ | 13:46 |
| lucasagomes | we probably need a regex for that baremetalbrbm* | 13:47 |
| rloo | lucasagomes, NobodyCam: I think you can use [a-z0-9_]+ | 13:49 |
| lucasagomes | yup | 13:49 |
| rloo | cuz that's what is being used now in the ci_commands file ;) | 13:49 |
| *** Haomeng has joined #openstack-ironic | 13:50 | |
| derekh | NobodyCam: to test the ci_commands regex you can add something like this to you authorized_keys for the ssh user | 13:52 |
| derekh | no-port-forwarding,no-agent-forwarding,no-X11-forwarding,command="/usr/local/bin/ci_commands" ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDdWrAPZ/RcOE9gWFmww8MUs8bcmpKvqn+R6WKB9uWuFAMJeXHDoS9zTcCIA5pYXcsNmrbY5Wh/FfcYldeiLTcFb8CCE3SLSdEB8bbpLR6ibEXJ0jh4VgyTXkT7fjSa19ztwYSz9DGJuHxQ+R1B3MP+eGeiq+sRfI6yicnvz2KljWYWOtGX0rAJAuUaoj4kRHSwRxvmQk3fz1Kmxzp/zLWj3b0+J75PmRixf8/cF2laHwDzCJZYosKqXHDnjr9AeNZP6eoJ/n8pglc+oezCvB/QkAJR8MhKQKG/wEp5fJEdMcuO1RH/cCaPnNy9aPd4M | 13:52 |
| derekh | WT4yTXeW67rH/YkyS8Xf0Wz | 13:52 |
| *** yuanying has joined #openstack-ironic | 13:53 | |
| NobodyCam | derekh: yep been there done that... I think this is actually the thrid time we may have done this to TripleO :( | 13:55 |
| NobodyCam | this was the first time : https://github.com/openstack-infra/tripleo-ci/commit/532d56eac6cc4d95af2d94468c3273cd7ecf03bb#diff-e3e229c6b06324b3713822ab4e32b094 | 13:56 |
| derekh | NobodyCam: I forgot about that one | 13:57 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add docstrings for agent.py https://review.openstack.org/120660 | 13:58 |
| jroll | morning ironic :) | 13:58 |
| NobodyCam | morning jroll | 13:58 |
| jroll | oh, we broker tripleo, fun :( | 13:59 |
| jroll | broke* | 13:59 |
| rloo | morning jroll | 13:59 |
| NobodyCam | we're selling tripleo | 13:59 |
| NobodyCam | lol | 13:59 |
| jroll | derekh: is there a way to whitelist any virsh call? | 13:59 |
| jroll | NobodyCam: eH? | 13:59 |
| jroll | oh | 13:59 |
| jroll | lol | 13:59 |
| NobodyCam | broker | 13:59 |
| NobodyCam | lol | 13:59 |
| jroll | yeah | 13:59 |
| NobodyCam | sorry /me has had very little sleep | 13:59 |
| NobodyCam | stepping out for a somke then I get a patch for ci ready | 14:00 |
| jroll | no worries | 14:02 |
| derekh | jroll: Maybe, or possibly a subset of them, we still have to filter what its piped into | 14:03 |
| jroll | yeah | 14:03 |
| * jroll does a thing | 14:03 | |
| jroll | oh, wtf, NobodyCam already has a comment in here | 14:04 |
| jroll | reviewers need to read da comments :( | 14:04 |
| NobodyCam | i have a comment? | 14:05 |
| lucasagomes | jroll, heh I think the idea of whitelisting is not allow _any_ calls :) | 14:05 |
| NobodyCam | where | 14:05 |
| lucasagomes | tho yeah, virsh should be problematic... but ah "virsh --help && rm -rf /" yeah :) | 14:05 |
| jroll | lucasagomes: right | 14:05 |
| jroll | NobodyCam: https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/ssh.py#L149-152 | 14:06 |
| lucasagomes | s/should/shouldn't/ | 14:06 |
| jroll | yeah, security is hard :P | 14:06 |
| NobodyCam | oh doh | 14:07 |
| NobodyCam | lol | 14:07 |
| jroll | I really wish tripleo CI was built on each run like devstack | 14:07 |
| rloo | thx for the comment NobodyCam! ;) | 14:07 |
| jroll | derekh: what's the level of effort required to rebuild your CI environment, out of curiousity? | 14:08 |
| *** andreykurilin has quit IRC | 14:08 | |
| *** andreykurilin_ has joined #openstack-ironic | 14:08 | |
| derekh | jroll: If I build a new test env image it would be about a day, but some something like this I would take an existing known "good" image and mount/edit the ci_commands file, then redeploy (should be around a hour or so if all goes well) | 14:10 |
| NobodyCam | jroll: I don't know the effort but it takes infra at least a day | 14:10 |
| jroll | wow | 14:10 |
| jroll | what is it that takes a day? | 14:10 |
| jroll | that seems a little absurd | 14:11 |
| NobodyCam | that the node pool stuff no? | 14:11 |
| jroll | I mean | 14:11 |
| jroll | I can't imagine a system that takes a day to deploy | 14:11 |
| NobodyCam | they use nova baremetal ...lol | 14:12 |
| derekh | jroll: the image build / deploy is fairly quick, but there are always regressions as its not a code path that is exercised often | 14:12 |
| NobodyCam | j/k | 14:12 |
| jroll | like, it takes us less than that to deploy a 40k node cloud | 14:12 |
| jroll | derekh: I see... if that was part of the CI job, it would get tested many times a day :) | 14:12 |
| derekh | jroll: yup, its on the list | 14:13 |
| derekh | jroll: that list that keeps getting longer... | 14:13 |
| NobodyCam | ok I'm thinking to patches on to add a new regex for the new ssh command and one to remove the old one, | 14:13 |
| *** chuckC has quit IRC | 14:13 | |
| NobodyCam | gah... two patches one ... | 14:13 |
| jroll | derekh: ok, I just wanted to make sure "deploy as little as possible" wasn't an explicit design decision | 14:14 |
| jroll | NobodyCam: +1 | 14:14 |
| jroll | anyone up for a quick IPA review? https://review.openstack.org/#/c/120586/ | 14:18 |
| *** Haomeng has quit IRC | 14:18 | |
| Shrews | jroll: looking | 14:20 |
| jroll | Shrews: this comes with a disclaimer that we need to refactor our exceptions in general :| | 14:21 |
| Shrews | ack | 14:21 |
| *** Haomeng has joined #openstack-ironic | 14:22 | |
| Shrews | jroll: lgtm | 14:22 |
| jroll | yay | 14:22 |
| jroll | thanks | 14:26 |
| Shrews | jroll: i really should spend some time with IPA at some point | 14:26 |
| jroll | you all should :P | 14:26 |
| * jroll wants to make it the default driver in kilo | 14:27 | |
| *** yuanying has quit IRC | 14:27 | |
| lucasagomes | NobodyCam, but on ur proxy patch | 14:27 |
| lucasagomes | bug | 14:27 |
| lucasagomes | added a comment | 14:27 |
| * lucasagomes can fix it if needed no worried | 14:27 | |
| lucasagomes | worries* | 14:27 |
| lucasagomes | NobodyCam, I will put up a new patch set fixing it + adding tests | 14:28 |
| Shrews | rloo: is your ironicclient bash-completion change supposed to mash the output up as one long line? | 14:28 |
| rloo | Shrews: my change shouldn't have modified what it did. but yes, bash-completion puts the stuff all on one line. | 14:29 |
| rloo | Shrews: similar to nova bash-completion. | 14:29 |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Fix exception that is not properly raised https://review.openstack.org/120586 | 14:29 |
| * Shrews tries nova | 14:29 | |
| rloo | Shrews: I never use that command, but the docs folks have a script that uses it to get the subcommands, to generate the CLI docn. | 14:30 |
| NobodyCam | lucasagomes: good catch... I was still mostly asleep on that rev | 14:30 |
| NobodyCam | brb walkies again | 14:30 |
| lucasagomes | NobodyCam, no bothers :) | 14:31 |
| Shrews | rloo: ah, ok. i was questioning the usefulness of that, but that makes sense. Perhaps we should note that in the commit msg? | 14:31 |
| rloo | Shrews: note what? that the response is all on one line? | 14:31 |
| Shrews | rloo: how it is used by docs | 14:31 |
| rloo | Shrews: it is mentioned in the bug. but the work of adding bash-completion was in another patch. | 14:32 |
| Shrews | rloo: ah! i missed the bug line in gerrit. hidden next to the commit-id | 14:33 |
| * Shrews wishes gerrit did a better job of highlighting those | 14:33 | |
| rloo | Shrews: those bugs like to hide... | 14:33 |
| jroll | lucasagomes: do you think we should have direct tests for reserve() and release()? https://review.openstack.org/#/c/120488/2 | 14:38 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: EXPERIMENTAL Implement hardware discovery in PXE driver https://review.openstack.org/110031 | 14:38 |
| lucasagomes | jroll, it's mostly calling the db methods (which has tests) but I can add it if you think it's important | 14:39 |
| jroll | meh, I don't know that it is | 14:39 |
| * jroll looks one more time | 14:40 | |
| rloo | dtantsur: checking, you don't like str1+str2+str3 (vs ''.join(strs)), but str1+str2 is OK? | 14:41 |
| dtantsur | rloo, right. I don't think I'm alone in it :) | 14:41 |
| rloo | dtantsur: ok, I see a couple more places where that hasn't been changed. | 14:42 |
| jroll | lucasagomes: found some other things, so might as well add tests while you're there | 14:42 |
| lucasagomes | jroll, ack will do | 14:43 |
| lucasagomes | dtantsur, about making the context mandatory... all RPC methods needs the first parameter to be the context | 14:44 |
| lucasagomes | dtantsur, and on the create the following patch is making use of the context to form the object | 14:44 |
| dtantsur | lucasagomes, can't we fix the decorator instead of inventing ugly code? | 14:44 |
| lucasagomes | dtantsur, that base.py for object is copied from nova | 14:45 |
| lucasagomes | we want to oslofy it at some point | 14:45 |
| lucasagomes | as it's mandatory for nova, I think we should not divert from it | 14:45 |
| lucasagomes | otherwise it will make it super hard in the future to put it into oslo | 14:45 |
| dtantsur | lucasagomes, it it from oslo? otherwise I don't see any problems changing it | 14:45 |
| lucasagomes | dtantsur, it's from nova | 14:46 |
| dtantsur | well, then we should assure that proper change lands to oslo | 14:46 |
| lucasagomes | we have been using their object-like | 14:46 |
| lucasagomes | it's not in oslo | 14:46 |
| dtantsur | then we can change it :) | 14:46 |
| jroll | what's the problem with having the context mandatory? | 14:46 |
| lucasagomes | dtantsur, on that create() the context have to be mandatory | 14:47 |
| dtantsur | jroll, it's already set on the same object | 14:47 |
| lucasagomes | then on the next sync with nova we can get rid of the context on the other methods | 14:47 |
| lucasagomes | like detroy | 14:47 |
| lucasagomes | destroy | 14:47 |
| dtantsur | lucasagomes, why not use self.context? | 14:47 |
| lucasagomes | dtantsur, https://github.com/openstack/nova/blob/master/nova/objects/base.py#L166-L208 | 14:47 |
| lucasagomes | but first I need to have all objects with a context inside it | 14:47 |
| lucasagomes | so I can do that ^ | 14:47 |
| lucasagomes | dtantsur, there's no self.context | 14:47 |
| jroll | right, it's not set on the object | 14:48 |
| * lucasagomes double check | 14:48 | |
| lucasagomes | yeah there's no self.context | 14:48 |
| NobodyCam | lucasagomes: I think this will do what we are looking for: https://review.openstack.org/#/c/120799 | 14:48 |
| lucasagomes | jroll, that's the idea of all the patches, to make sure all objects has a context | 14:48 |
| dtantsur | lucasagomes, https://github.com/openstack/ironic/blob/master/ironic/objects/base.py#L206 | 14:48 |
| jroll | lucasagomes: right, I'm asking dtantsur why there's a problem with mandatory context | 14:48 |
| dtantsur | ok sorry it's _context | 14:49 |
| jroll | right, but | 14:49 |
| lucasagomes | dtantsur, yeah... exactly and that's what I'm trying to make mandatory | 14:49 |
| jroll | if it isn't passed to create() | 14:49 |
| jroll | it will be None | 14:49 |
| jroll | and then we won't have it later when we need it | 14:49 |
| dtantsur | if you pass None to ctor - yes | 14:49 |
| dtantsur | but you don't | 14:49 |
| lucasagomes | jroll, yup | 14:49 |
| jroll | right now you can | 14:50 |
| jroll | and that sucks | 14:50 |
| jroll | this is fixing that | 14:50 |
| dtantsur | then let's drop it from c-tor | 14:50 |
| lucasagomes | yeah... exactly before objects was malformed with no context | 14:50 |
| jroll | assuming ctor is create? | 14:50 |
| jroll | wait | 14:50 |
| jroll | what is c-tor | 14:50 |
| dtantsur | no, it's not | 14:50 |
| dtantsur | __init__ | 14:50 |
| jroll | oh, constructor | 14:51 |
| jroll | dunno if we call that directly | 14:51 |
| jroll | (or just from create()) | 14:51 |
| jroll | and if we drop it from __init__, then we'll never get a context, right? | 14:51 |
| dtantsur | folks, please have a look at the code :-/ | 14:51 |
| dtantsur | https://review.openstack.org/#/c/120773/1/ironic/api/controllers/v1/node.py | 14:51 |
| dtantsur | lines 751-753 | 14:51 |
| NobodyCam | derekh: fyi: https://review.openstack.org/#/c/120799 | 14:52 |
| lucasagomes | we form objects using the constructor... but that doesnt save it to the db | 14:52 |
| jroll | sorry, multitasking with an omelette :| | 14:52 |
| dtantsur | context is now passed twice | 14:52 |
| lucasagomes | create() creates to the db | 14:52 |
| dtantsur | lucasagomes, ^^^ | 14:52 |
| lucasagomes | dtantsur, will take a look | 14:52 |
| jroll | oh, hm | 14:52 |
| lucasagomes | hmm I will take another look | 14:53 |
| jroll | yeah, read this wrong | 14:53 |
| lucasagomes | I was trying to make nova-ish | 14:53 |
| lucasagomes | https://github.com/openstack/nova/blob/master/nova/objects/block_device.py#L93 | 14:53 |
| lucasagomes | dtantsur, cheers, will take a look at it | 14:53 |
| dtantsur | ok :) | 14:53 |
| jroll | lucasagomes: yeah, maybe make it mandatory for __init__ | 14:54 |
| derekh | NobodyCam: thanks, will look it over in a little bit, just gotta finished something first | 14:54 |
| lucasagomes | jroll, ack, will look into it | 14:55 |
| lucasagomes | dtantsur, jroll cheers | 14:55 |
| jroll | cool :) thanks dtantsur | 14:55 |
| dtantsur | np) | 14:56 |
| jroll | (for bringing this up) | 14:56 |
| lucasagomes | NobodyCam, tests added https://review.openstack.org/120433 | 14:56 |
| NobodyCam | lucasagomes: w00 h00 | 14:59 |
| NobodyCam | ty | 14:59 |
| lucasagomes | no bothers :) | 14:59 |
| *** chuckC has joined #openstack-ironic | 15:00 | |
| *** k4n0 has quit IRC | 15:01 | |
| ramineni | JayF : Hi | 15:02 |
| jroll | ramineni: don't think jay is online yet, is there something maybe I can help you with? | 15:03 |
| ramineni | jroll , hi , ya | 15:03 |
| ramineni | jroll , we would like to submit script for building iso from kernel and initrd , how should we go about it? | 15:03 |
| ramineni | jroll , can be part of ironic or dib? | 15:04 |
| jroll | ramineni: for ironic PXE driver or for IPA? | 15:04 |
| ramineni | jroll , it's generic, even for iloiscsidriver , so that user can make use of it , even for building iso if required | 15:06 |
| jroll | ok, for iloiscsi driver, should go in DIB I think | 15:07 |
| ramineni | jroll, is it not same , if its for ipa ? | 15:07 |
| jroll | IPA repo holds IPA build scripts right now | 15:07 |
| openstackgerrit | A change was merged to openstack/ironic: Revert "Search line with awk itself and avoid grep" https://review.openstack.org/120783 | 15:08 |
| jroll | but you could also do it in DIB | 15:08 |
| jroll | doesn't matter much either way | 15:08 |
| jroll | if DIB helps you build the ISO, do it there | 15:08 |
| jroll | derekh: that revert landed | 15:08 |
| *** jcoufal has quit IRC | 15:08 | |
| *** andreykurilin_ is now known as andreykurilin | 15:08 | |
| derekh | jroll: ok, thanks | 15:09 |
| openstackgerrit | Rakesh H S proposed a change to openstack/python-ironicclient: ironic client to use os_region_name if provided https://review.openstack.org/120807 | 15:10 |
| ramineni | jroll , ok, thanks | 15:11 |
| jroll | :) | 15:12 |
| NobodyCam | lucasagomes: great tests TY | 15:15 |
| lucasagomes | NobodyCam, yw | 15:15 |
| lucasagomes | hope that's enough to satisfy the nova ppl | 15:15 |
| NobodyCam | will see, I still half beleive they end up not wanting lol... | 15:16 |
| lucasagomes | NobodyCam, yea | 15:16 |
| NobodyCam | s/they/they'll/ | 15:16 |
| jroll | ugh, Shrews, still dunno what to do about this rebuild test with IPA :( | 15:19 |
| lucasagomes | at least it shows good faith from our part | 15:20 |
| lucasagomes | that's alright | 15:20 |
| Shrews | jroll: with IPA, what is ephemeral size set to? 0 or N/A? | 15:21 |
| jroll | we don't support ephemeral yet :( | 15:22 |
| NobodyCam | lucasagomes: ++ | 15:22 |
| Shrews | jroll: i know, but what is in the flavor? | 15:22 |
| jroll | oh uh | 15:22 |
| jroll | right now we share flavors | 15:22 |
| jroll | so I guess I could change that | 15:22 |
| jroll | yay, more devstack changes to push through :( | 15:22 |
| lucasagomes | NobodyCam, you know if they want it merged till tuesday? | 15:22 |
| lucasagomes | because next tuesday is the final decision | 15:22 |
| lucasagomes | whether we integrate or not right? | 15:23 |
| Shrews | jroll: oh, so the ephemeral size is ignored by IPA then? | 15:23 |
| NobodyCam | i'm not sure. | 15:23 |
| *** yuanying has joined #openstack-ironic | 15:23 | |
| jroll | oh, just need to set the variable, hm | 15:23 |
| jroll | Shrews: yeah, that's probably not ideal either | 15:23 |
| NobodyCam | we have it if they want it | 15:23 |
| NobodyCam | I believe that is the last bit of code we need | 15:23 |
| jroll | and default size is 0 | 15:23 |
| Shrews | jroll: yeah, so if you set the variable (when using IPA) to zero, then change the test to skip the verify_partition stuff if it's zero... | 15:23 |
| NobodyCam | whats left now is just docs I think | 15:24 |
| jroll | Shrews: yeah, ok | 15:24 |
| jroll | this makes more sense now | 15:24 |
| jroll | ty | 15:24 |
| Shrews | jroll: and skip the test file create... and existence after rebuild.... | 15:24 |
| Shrews | heck, i wonder if it's even worth testing rebuild at that point.... | 15:24 |
| NobodyCam | oh question... do we have command line examples of the migration script? | 15:24 |
| *** rakesh_hs has quit IRC | 15:25 | |
| jroll | Shrews: right, probably not | 15:25 |
| jroll | Shrews: if IPA supports rebuild right now, it's by accident :P | 15:25 |
| jroll | Shrews: is #openstack-qa the channel to bug people about tempest changes? | 15:25 |
| Shrews | yeah | 15:26 |
| jroll | cool | 15:26 |
| *** rainya has joined #openstack-ironic | 15:26 | |
| * jroll will put these up later today | 15:26 | |
| lucasagomes | NobodyCam, yeah fingers crossed | 15:27 |
| jroll | Shrews: I'll try to keep the rebuild test in there though :P | 15:27 |
| Shrews | jroll: yeah, i think that should be tested even w/o ephemeral | 15:27 |
| jroll | agree | 15:28 |
| Shrews | now that i think about it | 15:28 |
| Shrews | jroll: so, one thing you could do is add a compute_feature for rebuild and check that *inside* the test: if CONF.compute_feature.rebuild: ... | 15:31 |
| Shrews | you'd have to set that feature to False when using IPA | 15:31 |
| *** rainya has quit IRC | 15:31 | |
| jroll | Shrews: meh :/ | 15:31 |
| jroll | Shrews: I don't see why rebuild wouldn't work, should be fine | 15:32 |
| Shrews | jroll: oh, i thought you said that's not supported with IPA | 15:32 |
| jroll | I'm not sure if it is | 15:32 |
| jroll | I think if it's not, it should be | 15:32 |
| Shrews | ++ | 15:32 |
| jroll | because from the deploy driver perspective, it's just a deploy | 15:32 |
| jroll | rebuild just checks some extra things | 15:32 |
| jroll | before deploying | 15:33 |
| *** rainya has joined #openstack-ironic | 15:33 | |
| jroll | Shrews: https://review.openstack.org/120814 | 15:35 |
| *** todd_dsm has quit IRC | 15:36 | |
| Shrews | jroll: you need to move the test file creation under the if | 15:39 |
| Shrews | well, you *should*, logically | 15:39 |
| jroll | ah, yeah | 15:39 |
| jroll | thanks | 15:39 |
| jroll | Shrews: probably fine to leave preserve_ephemeral=True, yes? | 15:40 |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Support for setting boot mode in pxe_ilo driver https://review.openstack.org/118578 | 15:40 |
| Shrews | jroll: hrm, good question | 15:41 |
| jroll | from ironic's perspective, it's fine | 15:41 |
| jroll | I think | 15:41 |
| * jroll looks again | 15:41 | |
| jroll | ah yeah | 15:41 |
| jroll | that's just an instance_info thing | 15:41 |
| * Shrews wonders what the disk partitioning code does for that case.... | 15:42 | |
| * Shrews looks | 15:42 | |
| *** todd_dsm has joined #openstack-ironic | 15:43 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: EXPERIMENTAL Implement hardware discovery in PXE driver https://review.openstack.org/110031 | 15:43 |
| jroll | Shrews: updated because doing it right today | 15:43 |
| jroll | :P | 15:43 |
| *** lazy_prince is now known as killer_prince | 15:43 | |
| Shrews | jroll: comment needs updated :-P | 15:45 |
| Shrews | # Rebuild and preserve the ephemeral partition | 15:45 |
| jroll | ugh | 15:45 |
| jroll | thanks | 15:45 |
| jroll | updated | 15:45 |
| *** jistr has quit IRC | 15:47 | |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add docstrings for api/ tree https://review.openstack.org/120661 | 15:49 |
| * jroll afk for a bit | 15:51 | |
| *** andreykurilin has quit IRC | 15:51 | |
| *** ifarkas has quit IRC | 15:51 | |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Support for setting boot mode in pxe_ilo driver https://review.openstack.org/118578 | 15:52 |
| NobodyCam | brb... /me runs to cornor store for a pack of smokes... | 15:55 |
| dtantsur | calling it a day, see you tomorrow | 15:55 |
| *** dtantsur is now known as dtantsur|afk | 15:55 | |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Support for setting boot mode in pxe_ilo driver https://review.openstack.org/118578 | 15:55 |
| *** yuanying has quit IRC | 15:57 | |
| *** faizan has quit IRC | 15:59 | |
| *** aswadr has quit IRC | 16:03 | |
| NobodyCam | night dtantsur|afk | 16:03 |
| lucasagomes | dtantsur|afk, g'night | 16:18 |
| *** rushiagr is now known as rushiagr_away | 16:21 | |
| *** rwsu has quit IRC | 16:24 | |
| *** ndipanov is now known as ndipanov_gone | 16:24 | |
| *** rwsu has joined #openstack-ironic | 16:29 | |
| JayF | ramineni: you pinged me? | 16:31 |
| JayF | jroll: ramineni: I would prefer any DIB elements created to build IPA to go under imagebuild/ in the IPA repo | 16:32 |
| JayF | jroll: ramineni: AFAICT, the only repo for DIB elements is 'tripleo-elements" and while the existing Ironic pxe ramdisk is there, I don't think it makes sense to continue overloading that repo | 16:33 |
| *** rwsu has quit IRC | 16:33 | |
| JayF | jroll: ramineni: might be valuable do do something like write the library for dib to build ISOs, then put a minimal set of stuff in IPA imagebuild/ that utilizes that to build the image | 16:34 |
| *** rainya has quit IRC | 16:35 | |
| *** rwsu has joined #openstack-ironic | 16:35 | |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Add 'bash-completion' to 'ironic help' response https://review.openstack.org/120122 | 16:38 |
| *** killer_prince is now known as lazy_prince | 16:39 | |
| derekh | NobodyCam: regex lgtm, just looking for a small comment then I'm happy | 16:40 |
| derekh | NobodyCam: gotta grab a train will check it later | 16:40 |
| *** derekh has quit IRC | 16:40 | |
| lucasagomes | ok folks I'll call it a day too | 16:40 |
| lucasagomes | gotta take the train back home yet | 16:40 |
| lucasagomes | have a good night everybody! | 16:40 |
| *** lucasagomes has quit IRC | 16:41 | |
| *** r-daneel has joined #openstack-ironic | 16:45 | |
| *** rushiagr_away is now known as rushiagr | 16:46 | |
| openstackgerrit | A change was merged to openstack/python-ironicclient: handles keyboard interrupt https://review.openstack.org/120767 | 16:46 |
| jroll | JayF: +1 | 16:48 |
| jroll | sounds reasonable | 16:49 |
| *** yuanying has joined #openstack-ironic | 16:53 | |
| *** lazy_prince is now known as killer_prince | 16:58 | |
| *** rainya has joined #openstack-ironic | 17:00 | |
| *** harlowja_away is now known as harlowja | 17:09 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/120637 | 17:14 |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/114058 | 17:14 |
| *** athomas has quit IRC | 17:16 | |
| *** athomas has joined #openstack-ironic | 17:17 | |
| *** athomas has quit IRC | 17:18 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/python-ironicclient: Updated from global requirements https://review.openstack.org/120651 | 17:20 |
| *** yuanying has quit IRC | 17:27 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add verification support to the agent https://review.openstack.org/120627 | 17:42 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add versioning to Agent decommission https://review.openstack.org/106859 | 17:43 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding support for decommissioning https://review.openstack.org/104379 | 17:43 |
| *** rainya has quit IRC | 17:44 | |
| *** wanyen has joined #openstack-ironic | 17:44 | |
| *** openstackgerrit has quit IRC | 17:46 | |
| *** rainya has joined #openstack-ironic | 17:47 | |
| *** pelix has quit IRC | 17:47 | |
| *** openstackgerrit has joined #openstack-ironic | 17:48 | |
| *** ramineni has quit IRC | 18:03 | |
| *** rushiagr is now known as rushiagr_away | 18:09 | |
| *** rushiagr_away is now known as rushiagr | 18:14 | |
| *** yuanying has joined #openstack-ironic | 18:23 | |
| *** mikedillion has joined #openstack-ironic | 18:27 | |
| *** mikedillion has quit IRC | 18:30 | |
| devananda | o/ | 18:37 |
| Shrews | \o | 18:37 |
| *** Ugallu has quit IRC | 18:39 | |
| NobodyCam | morning devananda | 18:44 |
| NobodyCam | ok I think thats it for all the conf calls this morning | 18:52 |
| *** yuanying has quit IRC | 18:57 | |
| *** rushiagr is now known as rushiagr_away | 19:00 | |
| * NobodyCam had some errands to run... back a bit... will have phone with me if I am needed | 19:03 | |
| *** bigjools has quit IRC | 19:03 | |
| *** dprince has quit IRC | 19:05 | |
| *** russellb has quit IRC | 19:05 | |
| *** JayF has quit IRC | 19:06 | |
| *** adam_g has quit IRC | 19:06 | |
| *** pensu has quit IRC | 19:06 | |
| *** russellb has joined #openstack-ironic | 19:07 | |
| * devananda lunches | 19:12 | |
| *** Guest75250 has quit IRC | 19:14 | |
| *** russellb has quit IRC | 19:15 | |
| *** jroll has quit IRC | 19:19 | |
| *** dprince has joined #openstack-ironic | 19:20 | |
| *** JayF has joined #openstack-ironic | 19:33 | |
| *** stevebaker has quit IRC | 19:33 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/120637 | 19:40 |
| *** ryanpetrello has quit IRC | 19:44 | |
| *** ryanpetrello has joined #openstack-ironic | 19:44 | |
| *** ryanpetrello is now known as ryanpetre | 19:44 | |
| *** mgagne has joined #openstack-ironic | 19:51 | |
| *** jroll has joined #openstack-ironic | 19:51 | |
| *** bigjools has joined #openstack-ironic | 19:52 | |
| *** yuanying has joined #openstack-ironic | 19:53 | |
| *** jjmanzer has joined #openstack-ironic | 19:59 | |
| jjmanzer | learning about ironic here, I am a network automation guy and I wanted to know how/if ironic would actually talk to switches like cisco/arista gear to provision bare metal servers. | 20:00 |
| jroll | jjmanzer: hey :) | 20:02 |
| jjmanzer | hello:) | 20:02 |
| jroll | it didn't make it upstream this cycle, but us folks at rackspace have done some work on that front | 20:03 |
| jroll | as far as isolating provisioning networks from tenant networks etc | 20:03 |
| jjmanzer | is mapping of the local l2 topology done? i.e. does it notice that if a VLAN is needed for a server, that it might need to be added to the uplinks/downlinks between access and distribution layer? | 20:04 |
| jroll | it's actually just some neutron code to talk to real switches, and ironic just tells it to switch vlans | 20:04 |
| jjmanzer | ahh so do a 'switch trunk vlan allowed ...' on a certian port? | 20:04 |
| jroll | yeah, that's all on the neutron side iirc | 20:04 |
| jroll | yep | 20:04 |
| jjmanzer | so if i want to use my own port_provision app that I wrote and exposed through a REST/JSON API how might I link that in? | 20:05 |
| jjmanzer | our network is somewhat complicated so lots of provision logic is needed hence the port_provision app | 20:05 |
| jroll | right, so | 20:05 |
| jroll | like I said, this isn't upstream yet | 20:06 |
| jroll | I'd like to make that layer pluggable | 20:06 |
| jroll | so you could write your own little plugin or whatever to talk to your app | 20:06 |
| jroll | instead of neutron | 20:06 |
| jjmanzer | is ironic primarily about the baremetal server side of thing? get it booted, OS installed that kind of thing? I'm still feeling my way around here. I will check out neutron as well | 20:07 |
| jroll | yes, exactly. neutron is for controlling networks | 20:11 |
| jjmanzer | tyvm | 20:12 |
| Shrews | https://gigaom.com/2014/09/11/hp-buys-eucalyptus-as-cloud-consolidation-commences-for-real/ | 20:13 |
| Shrews | and i have no idea what this means, fwiw | 20:13 |
| jroll | huh. | 20:14 |
| rloo | Shrews: hmm | 20:22 |
| *** yuanying has quit IRC | 20:26 | |
| * NobodyCam is back | 20:27 | |
| *** wanyen has quit IRC | 20:30 | |
| *** dprince has quit IRC | 20:30 | |
| *** wanyen has joined #openstack-ironic | 20:30 | |
| *** andreykurilin has joined #openstack-ironic | 20:37 | |
| wanyen | new patches for uefi support for iLO virtual media iscsi and pxe-ilo driver were uploaded this morning https://review.openstack.org/#/c/116561/ and https://review.openstack.org/#/c/118578/. Please review them so that we can address as many comments as possible in our next patch. FFE for these two submissions is 09/12. Thanks! | 20:37 |
| *** igordcard has joined #openstack-ironic | 20:41 | |
| jroll | wanyen: just uploaded some comments I made this morning that didn't sync | 20:41 |
| wanyen | jroll, thanks! we will take a look at your comments and address them in the next patch. | 20:43 |
| jroll | thanks :) | 20:44 |
| jroll | it looks pretty close | 20:44 |
| *** linggao has quit IRC | 20:45 | |
| *** rainya_ has joined #openstack-ironic | 20:48 | |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Add sort_key and sort_dir parameters to *-list https://review.openstack.org/114555 | 20:48 |
| *** rainya has quit IRC | 20:48 | |
| openstackgerrit | A change was merged to openstack/python-ironicclient: List resources with detail https://review.openstack.org/119825 | 20:53 |
| *** rainya_ has quit IRC | 20:54 | |
| NobodyCam | humm anyone happen to have seen this error b4? http://logs.openstack.org/33/120433/7/check/gate-nova-python26/df2f742/console.html#_2014-09-11_16_48_15_935 | 20:55 |
| *** stevebaker has joined #openstack-ironic | 20:55 | |
| jroll | nope | 20:56 |
| jroll | that's weird | 20:56 |
| NobodyCam | ya | 20:57 |
| NobodyCam | looks like its comming up on a recheck too | 20:57 |
| NobodyCam | so now comes the BUG hunt phase | 20:57 |
| jroll | "site:bugs.launchpad.net" ftw | 20:59 |
| NobodyCam | :-p | 20:59 |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Remove bypassing of H302 for gettextutils markers https://review.openstack.org/120910 | 21:07 |
| *** rainya has joined #openstack-ironic | 21:22 | |
| *** yuanying has joined #openstack-ironic | 21:23 | |
| *** ChuckC_ has joined #openstack-ironic | 21:24 | |
| *** adam_g has joined #openstack-ironic | 21:27 | |
| *** adam_g has joined #openstack-ironic | 21:27 | |
| * NobodyCam cannt take it any more ... I need to reboot... brb | 21:30 | |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Add uefi boot mode support in IloVirtualMediaIscsiDeploy https://review.openstack.org/116561 | 21:35 |
| *** wanyen has quit IRC | 21:43 | |
| NobodyCam | update done! /me is back | 21:48 |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Add uefi boot mode support in IloVirtualMediaIscsiDeploy https://review.openstack.org/116561 | 21:48 |
| NobodyCam | slagle: Ty for recheck on 120433 :) | 21:50 |
| *** rainya has quit IRC | 21:51 | |
| *** rainya has joined #openstack-ironic | 21:52 | |
| NobodyCam | doh darn tab... sorry slagle ping was for sdague :-p | 21:53 |
| *** yuanying has quit IRC | 21:56 | |
| openstackgerrit | Kyle Stevenson proposed a change to openstack/ironic: Add implementation of pluggable distributed sync https://review.openstack.org/115148 | 22:04 |
| *** andreykurilin has quit IRC | 22:14 | |
| openstackgerrit | greghaynes proposed a change to openstack/ironic: Add conductor_hostname property to nodes https://review.openstack.org/109688 | 22:25 |
| greghaynes | ok, I am in need of advice on how to remove the # TODO I added on https://review.openstack.org/#/c/109688/12/ironic/api/controllers/v1/node.py | 22:26 |
| greghaynes | that is a pretty nasty thing to do, but im having a hard time finding where else I could put that code... digging into rpcapi (which is where the hash ring manager currently lives) seems nasty and not necessary | 22:26 |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Support for setting boot mode in pxe_ilo driver https://review.openstack.org/118578 | 22:27 |
| NobodyCam | oh ouch | 22:31 |
| openstackgerrit | Kyle Stevenson proposed a change to openstack/ironic: Add implementation of ZooKeeper SyncManager https://review.openstack.org/117576 | 22:34 |
| greghaynes | NobodyCam: That doesnt sound good :( | 22:34 |
| greghaynes | Im up for adding another patch to pull the hash ring manager out of rpcapi if that sounds sane... just wanted some other input before I start charging ahead | 22:35 |
| NobodyCam | greghaynes: where is it going to go | 22:36 |
| *** jjmanzer has quit IRC | 22:36 | |
| greghaynes | yep, that is the question | 22:36 |
| *** openstackgerrit has quit IRC | 22:38 | |
| *** openstackgerrit_ has joined #openstack-ironic | 22:39 | |
| *** openstackgerrit_ is now known as openstackgerrit | 22:40 | |
| greghaynes | this is one of those things where in $other_language I would just say "clearly we just need a singleton" | 22:40 |
| devananda | greghaynes: why is this being python mean a singleton not the right answer? | 22:45 |
| greghaynes | It doesnt, I just have noticed that pattern being more frowned upon in python land | 22:46 |
| devananda | greghaynes: we need a process-wide lock when creating singletons, due to greenthreads, but otherwise, I don't have a problem with them | 22:47 |
| greghaynes | sweet | 22:47 |
| greghaynes | Ill play with that, then | 22:47 |
| NobodyCam | where is check-grenade-dsvm-ironic-sideways ? | 22:48 |
| devananda | greghaynes: this is an API change - https://review.openstack.org/#/c/109688/12 | 22:49 |
| devananda | greghaynes: without a blueprint/spec, and after feature freeze | 22:49 |
| devananda | greghaynes: the data model change is one thing -- risky this late, but maybe justifiable. the REST API change is raising red flags for me | 22:50 |
| greghaynes | ok, I think the REST part can get pulled out | 22:50 |
| greghaynes | also, I wont be offended if it doesnt go in until k | 22:51 |
| devananda | greghaynes: I don't see why the REST change is related to the bug fix at all | 22:51 |
| devananda | there are two problems I see with the hash ring: rebalancing (what this patch starts) and redistribution percentage (which Nejc was working on, i think?) | 22:52 |
| *** rainya has quit IRC | 22:52 | |
| greghaynes | Agreed, that part of it isnt needed | 22:52 |
| greghaynes | the redistribution patch lifeless posted | 22:52 |
| *** rainya has joined #openstack-ironic | 22:53 | |
| greghaynes | https://review.openstack.org/#/c/118932/ | 22:53 |
| *** yuanying has joined #openstack-ironic | 22:53 | |
| devananda | ah, right | 22:53 |
| lifeless | devananda: I don't think we can land the redistribution fix patch until rebalancing is landed | 22:54 |
| lifeless | devananda: so we should get rebalancing in place, which we know will do too much work at first; then land the redistribution patch straight after, which will fix the amount of work. | 22:54 |
| lifeless | greghaynes: oh! | 22:55 |
| lifeless | greghaynes: we need to make sure that the column add in your patch either sets everything NULL and assumes NULL=needs rebalancing | 22:55 |
| lifeless | greghaynes: or | 22:55 |
| *** athomas has joined #openstack-ironic | 22:55 | |
| lifeless | greghaynes: we need to populate it with the /old/ hash algorithm, otherwise folk will end up with the right eventual host, but the content still on the old one | 22:55 |
| greghaynes | NULL sounds like the way to go | 22:56 |
| greghaynes | Less code to perform less operations | 22:57 |
| jroll | yes, NULL is the right thing to do | 22:58 |
| jroll | you'll need to restart conductors when deploying this change anyhow | 22:58 |
| lifeless | greghaynes: and so the rebalance code probably needs to be smart enough to check disk and make it roughly a no-op if everything is already in place | 22:58 |
| jroll | which would cause a rebalance | 22:58 |
| lifeless | jroll: that won't cause a rebalance | 22:58 |
| jroll | rolling restart of conductors won't rebalance? | 22:59 |
| lifeless | jroll: a conductor that misses a checkpoint will | 22:59 |
| jroll | ah, true | 22:59 |
| jroll | regardless | 22:59 |
| lifeless | jroll: rolling shouldn't miss checkpoints | 22:59 |
| jroll | as someone running ironic in production | 22:59 |
| jroll | a rebalance on this deploy would be ok with me | 22:59 |
| jroll | not optimal, but ok | 22:59 |
| lifeless | sure | 22:59 |
| lifeless | thanks | 22:59 |
| JayF | jroll: we also use IPA | 22:59 |
| jroll | ah fffffff | 22:59 |
| JayF | jroll: which conductor is attached to a node matters significantly less | 22:59 |
| jroll | yeah | 22:59 |
| lifeless | it will be fine | 22:59 |
| jroll | and rebalance less expensive for us | 23:00 |
| lifeless | greg is working with non-IPA Ironic deploys, I'm sure he'll test well :) | 23:00 |
| jroll | read: not expensive | 23:00 |
| jroll | heh | 23:00 |
| jroll | ok | 23:00 |
| greghaynes | haha | 23:00 |
| lifeless | rebalance onto the same node should be nearly free. | 23:00 |
| *** rainya has quit IRC | 23:00 | |
| devananda | lifeless: i believe a rolling restart /will/ rebalance everythign today, because ironic/common/service will call self.manager.del_host() which will unregister the conductor | 23:02 |
| devananda | I think we need a HUP handler to avoid that | 23:02 |
| jroll | ah, right. that sounds reasonable | 23:03 |
| jroll | I mean | 23:03 |
| jroll | shutting down a conductor should maybe trigger things | 23:03 |
| jroll | so you don't need to wait $timeout for it to failover and whatnot | 23:03 |
| devananda | jroll: it triggers manager.del_host() | 23:03 |
| * jroll looks | 23:04 | |
| jroll | devananda: I think that's okay behavior | 23:04 |
| russell_h | that doesn't actually trigger a rebalance in other conductors today though does it? | 23:04 |
| devananda | russell_h: there's no mechanism today to *trigger* a rebalance | 23:05 |
| russell_h | right | 23:05 |
| jroll | unless you explicitly want a restart to *not* rebalance; with the caveat that it will take $timeout for rebalancing to happen if a conductor gets killed | 23:05 |
| lifeless | unregistering the conductor is bad | 23:06 |
| lifeless | because it changes the ring shape immediately | 23:06 |
| devananda | russell_h: the API service, aiui, models the hash ring at every request | 23:06 |
| lifeless | a rolling deploy shouldn't cause a whole bunch of churn - regardless of driver | 23:06 |
| lifeless | devananda: there's a cache in there I believe | 23:06 |
| russell_h | nope | 23:06 |
| russell_h | there should be, but there isn't | 23:06 |
| devananda | so unregistering the conductor immediately changes the ring appearance and the routing of messages | 23:07 |
| lifeless | russell_h: class HashRingManager(object): | 23:07 |
| lifeless | russell_h: is that object constructed per request? | 23:07 |
| russell_h | lifeless: yeah | 23:07 |
| lifeless | well nuts | 23:07 |
| russell_h | IIRC I originally instantiated one that was shared across requests | 23:07 |
| lifeless | anyhow | 23:07 |
| russell_h | and would periodicially refresh | 23:08 |
| lifeless | its really bad in that case | 23:08 |
| lifeless | because two requests for node X | 23:08 |
| lifeless | which hash onto conductor Y normally | 23:08 |
| devananda | lifeless: HashRingManager is instantiated within RPCAPI, which in turn is instantuated by api/hooks.py for every request | 23:08 |
| devananda | :( | 23:08 |
| lifeless | can be split onto different conductors, if a third conductor comes/goes. | 23:08 |
| russell_h | but we felt like we didn't fully understand our consistency model at the time, so I was asked to remove it to preserve existing behavior | 23:08 |
| lifeless | russell_h: ack | 23:08 |
| devananda | I don't see a singleton or module-scoped instance caching it anywhere | 23:08 |
| devananda | but maybe i'm missing it | 23:08 |
| lifeless | so I think its pretty important that we differentiate between things restarting and things failing by going away permanently | 23:09 |
| devananda | russell_h: ah, right. i recall that now | 23:09 |
| lifeless | e.g. I'd like conductors to never unregister. | 23:09 |
| lifeless | let operators delete them | 23:09 |
| devananda | lifeless: differentiate between ... ++ ; operators delete them -- | 23:09 |
| russell_h | lifeless: do you want a distinction between a conductor being "registered" and it being "active"? | 23:10 |
| lifeless | devananda: note that operators deleting doesn't imply operators having to care about failover stuff | 23:10 |
| lifeless | russell_h: exactly. | 23:10 |
| russell_h | (where active is probably a terribly overloaded term) | 23:10 |
| lifeless | russell_h: 'has checkpointed within X seconds' <- registered and alive | 23:10 |
| russell_h | right | 23:10 |
| lifeless | russell_h: 'has checkpointed more than X seconds' <- registered and not alive | 23:10 |
| * devananda has no brainsugar left and needs food, bbl | 23:11 | |
| lifeless | if it isn't alive we can remove it from the hash ring *in the new implementation*. | 23:11 |
| lifeless | [because the new implementation uses hashed partitions, it won't destablise things to do so]. | 23:11 |
| russell_h | but given that the conductor currently (according to devananda, I haven't verified) deletes itself on a graceful shutdown | 23:12 |
| russell_h | is there much point in adding code to make the old hash ring less broken instead of just going to the new one? | 23:13 |
| russell_h | the old one is super broken at this point | 23:13 |
| russell_h | like if a conductor dies, there is no auto-rebalancing | 23:13 |
| russell_h | for that matter, in order to start multiple conductors you have to do multiple passes at it | 23:14 |
| russell_h | start them each up | 23:14 |
| russell_h | then go back and restart them | 23:14 |
| russell_h | (thats not specific to the hash ring, but rather how the conductor uses it) | 23:14 |
| *** Guest54173 is now known as antonym | 23:15 | |
| *** antonym is now known as Guest35655 | 23:15 | |
| lifeless | russell_h: you raise good points, I need to cogitate. | 23:18 |
| lifeless | russell_h: and context switch, sorry. | 23:18 |
| lifeless | russell_h: I don't see how restarting can help | 23:18 |
| lifeless | russell_h: because of the deleting behaviour | 23:19 |
| lifeless | russell_h: we need to refresh the rings without restarting, because restarting deletes the conductor which means restarting two nodes at once is a race | 23:19 |
| lifeless | russell_h: which is bad news for ops. | 23:19 |
| lifeless | russell_h: so - I think its worth making things more stable, how we do that - sure. | 23:19 |
| *** athomas has quit IRC | 23:26 | |
| *** r-daneel has quit IRC | 23:33 | |
| russell_h | lifeless: yeah, my brain is pretty fried at this point. | 23:36 |
| russell_h | the restarting thing helps when done serially, because the conductor notices its peers | 23:36 |
| russell_h | so say I start 3 conductors | 23:36 |
| russell_h | conductor 0 builds a hash ring where it owns every node | 23:36 |
| chuckC | using http://docs.openstack.org/developer/ironic/dev/dev-quickstart.html#deploying-ironic-with-devstack to try ironic/devstack for the first time, I see stack.sh fail because of a bad url: http://tarballs.openstack.org/ironic-python-agent/coreos/files/coreos_production_pxe-oem.cpio.gz | 23:36 |
| russell_h | conductor 1 builds a hash ring where it owns some random 1/2 of the nodes | 23:36 |
| chuckC | Is something wrong or missing in the doc, or should I fix devstack? | 23:36 |
| russell_h | and conductor 2 builds a hash ring where it owns some random 1/3 of the nodes | 23:37 |
| russell_h | when I restart conductor 0, it unregisters itself on shutdown, re-registers itself on startup, then rebuilds the hash ring | 23:37 |
| russell_h | and now believes it owns a different 1/3 of the nodes than conductor 2 | 23:37 |
| jroll | chuckC: looking | 23:38 |
| russell_h | restart conductor 1, it does the same thing, and now everything is good | 23:38 |
| *** dlaube has quit IRC | 23:38 | |
| chuckC | jroll: thanks | 23:38 |
| russell_h | (until a conductor dies, then you're boned) | 23:38 |
| jroll | chuckC: can you paste the error? | 23:38 |
| russell_h | we actually had a conductor die in production, and for absurd reasons we didn't notice for days | 23:38 |
| * chuckC looking | 23:39 | |
| russell_h | because the API rebuilds the hashring on every request | 23:39 |
| jroll | chuckC: never mind | 23:39 |
| JayF | chuckC: there's a fix for that url | 23:39 |
| russell_h | and conductors actually will handle any request they receive, regardless of whether they own the node | 23:39 |
| jroll | I thought that was correct in devstack, wth | 23:39 |
| JayF | chuckC: http://tarballs.openstack.org/ironic-python-agent/coreos/files/coreos_production_pxe_image-oem.cpio.gz | 23:39 |
| JayF | chuckC: is the correct url | 23:39 |
| russell_h | so all you really lose is periodic tasks | 23:39 |
| JayF | chuckC: and should be correct in devstack master, if not please let us know and we'll stab at it again | 23:39 |
| russell_h | which still sucks | 23:39 |
| jroll | chuckC: I'm going to throw up a fix to devstack, but for now set in localrc: IRONIC_AGENT_RAMDISK_URL=http://tarballs.openstack.org/ironic-python-agent/coreos/files/coreos_production_pxe_image-oem.cpio.gz | 23:40 |
| NobodyCam | adam_g: happen to be around? | 23:40 |
| jroll | JayF: I thought it was but it's not :( | 23:40 |
| chuckC | jroll: ok. I had already coded the fix, but wasn't sure if something in the doc might have been involved | 23:41 |
| JayF | I swear we merged a patch that fixed that <.< | 23:41 |
| jroll | chuckC: nah, it's devstack code, but I swear I double-checked that :( | 23:41 |
| jroll | I did too | 23:41 |
| JayF | let me see if I has it | 23:41 |
| JayF | I don't have a patchset up for it | 23:43 |
| JayF | wth | 23:43 |
| JayF | thanks for reporting this chuckC | 23:43 |
| JayF | now to go find what leprechaun ate our patchset | 23:43 |
| chuckC | JayF: np | 23:43 |
| JayF | jroll: you going to put the fix back up? I'd appreciate it as I'm in cloud-init hell right now | 23:43 |
| jroll | yep | 23:43 |
| *** rainya has joined #openstack-ironic | 23:47 | |
| *** rainya has quit IRC | 23:47 | |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add docstrings for agent.py https://review.openstack.org/120660 | 23:51 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic-python-agent: Add docstrings for api/ tree https://review.openstack.org/120661 | 23:53 |
| *** igordcard has quit IRC | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!