| *** ducttape_ has quit IRC | 00:00 | |
| *** EricGonczer_ has quit IRC | 00:17 | |
| *** ducttape_ has joined #openstack-dns | 00:42 | |
| *** rudrajit has joined #openstack-dns | 00:55 | |
| *** EricGonczer_ has joined #openstack-dns | 01:10 | |
| *** ducttape_ has quit IRC | 01:16 | |
| *** EricGonc_ has joined #openstack-dns | 01:17 | |
| *** EricGonczer_ has quit IRC | 01:20 | |
| *** stanzgy has joined #openstack-dns | 01:23 | |
| *** stanzgy_ has joined #openstack-dns | 01:23 | |
| *** EricGonc_ has quit IRC | 01:25 | |
| *** ducttape_ has joined #openstack-dns | 01:26 | |
| *** EricGonczer_ has joined #openstack-dns | 01:27 | |
| *** ducttape_ has quit IRC | 01:37 | |
| *** chlong has quit IRC | 01:40 | |
| *** ducttape_ has joined #openstack-dns | 01:50 | |
| *** ducttape_ has quit IRC | 01:51 | |
| *** chlong has joined #openstack-dns | 01:53 | |
| *** ducttape_ has joined #openstack-dns | 01:53 | |
| *** ducttape_ has quit IRC | 01:55 | |
| *** 17WAAXO8P has joined #openstack-dns | 02:19 | |
| *** EricGonczer_ has quit IRC | 02:26 | |
| *** maestropandy has joined #openstack-dns | 03:03 | |
| *** ducttape_ has joined #openstack-dns | 03:09 | |
| *** 17WAAXO8P has quit IRC | 03:10 | |
| *** richm has quit IRC | 03:12 | |
| *** maestropandy has quit IRC | 03:23 | |
| *** puck has quit IRC | 03:23 | |
| openstackgerrit | Merged openstack/python-designateclient: Use osc_lib instead of cliff https://review.openstack.org/335912 | 03:25 |
|---|---|---|
| *** puck has joined #openstack-dns | 03:26 | |
| *** ducttape_ has quit IRC | 03:52 | |
| *** ducttape_ has joined #openstack-dns | 05:22 | |
| *** ducttape_ has quit IRC | 05:27 | |
| *** rudrajit has quit IRC | 05:47 | |
| *** rudrajit has joined #openstack-dns | 05:48 | |
| *** rudrajit has quit IRC | 05:53 | |
| *** rudrajit has joined #openstack-dns | 05:56 | |
| *** pcaruana has joined #openstack-dns | 06:05 | |
| *** ducttape_ has joined #openstack-dns | 06:24 | |
| -openstackstatus- NOTICE: All python 3.5 jobs are failing today, we need to build new xenial images first. | 06:27 | |
| *** ducttape_ has quit IRC | 06:30 | |
| *** rudrajit has quit IRC | 06:44 | |
| *** rudrajit has joined #openstack-dns | 06:45 | |
| *** rudrajit has quit IRC | 06:49 | |
| *** abalutoiu has joined #openstack-dns | 07:17 | |
| *** ReSam has joined #openstack-dns | 07:58 | |
| ReSam | good morning! | 07:59 |
| ReSam | I'm trying to setup designate with neutron, to register a new dns name when creating a new instance through the dashboard. so far designate works (tested with CLI). But neutron keeps telling me: "DNSDomainNotFound: Domain foobar.example.com. not found in external DNS service". | 08:05 |
| ReSam | But running "designate domain-list" shows it. | 08:05 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/designate: [WIP] Testing latest u-c https://review.openstack.org/318020 | 08:10 |
| *** en_austin has joined #openstack-dns | 08:15 | |
| *** ducttape_ has joined #openstack-dns | 08:27 | |
| *** nyechiel has joined #openstack-dns | 08:28 | |
| *** ducttape_ has quit IRC | 08:31 | |
| *** rudrajit has joined #openstack-dns | 09:24 | |
| *** rudrajit has quit IRC | 09:28 | |
| *** en_austin has quit IRC | 09:52 | |
| *** ducttape_ has joined #openstack-dns | 11:29 | |
| *** ducttape_ has quit IRC | 11:33 | |
| *** ducttape_ has joined #openstack-dns | 11:56 | |
| *** EricGonczer_ has joined #openstack-dns | 12:01 | |
| *** EricGonc_ has joined #openstack-dns | 12:03 | |
| *** EricGonczer_ has quit IRC | 12:07 | |
| *** gord0 has left #openstack-dns | 12:19 | |
| *** stanzgy_ has quit IRC | 12:21 | |
| *** stanzgy has quit IRC | 12:21 | |
| *** ducttape_ has quit IRC | 12:23 | |
| *** ducttape_ has joined #openstack-dns | 12:52 | |
| *** maestropandy has joined #openstack-dns | 13:10 | |
| *** ducttape_ has quit IRC | 13:11 | |
| mugsie | ReSam: and it is the same project for both the network and the dns zone? | 13:12 |
| openstackgerrit | Merged openstack/designate: Updated from global requirements https://review.openstack.org/336171 | 13:23 |
| *** M00nr41n has quit IRC | 13:29 | |
| *** EricGonc_ has quit IRC | 13:42 | |
| *** richm has joined #openstack-dns | 13:43 | |
| openstackgerrit | Merged openstack/designate: Move APIv2 Docs to api-ref site https://review.openstack.org/329459 | 13:46 |
| *** eandersson_ has joined #openstack-dns | 14:27 | |
| eandersson_ | A silly question, but does each worker have a 1k greenlet thread pool, or do all workers share the 1k thread pool? | 14:27 |
| eandersson_ | *share | 14:27 |
| mugsie | each worker has their own pool eandersson_ | 14:34 |
| eandersson_ | So having 64 workers could potentially spawn 64k threads then? :D | 14:36 |
| *** mlavalle has joined #openstack-dns | 14:36 | |
| mugsie | well, they are greenlets | 14:37 |
| mugsie | oh | 14:38 |
| mugsie | wait | 14:38 |
| mugsie | the htreadpool is real | 14:38 |
| mugsie | so, yeah - it could | 14:38 |
| mugsie | eandersson_: why the need for 64 workers? | 14:38 |
| eandersson_ | You wouldn't :p | 14:39 |
| *** EricGonczer_ has joined #openstack-dns | 14:50 | |
| *** pglass has joined #openstack-dns | 15:01 | |
| *** abalutoiu has quit IRC | 15:03 | |
| *** pcaruana has quit IRC | 15:06 | |
| *** maestropandy has quit IRC | 15:12 | |
| *** maestropandy has joined #openstack-dns | 15:18 | |
| *** Kiall has quit IRC | 15:25 | |
| *** Kiall has joined #openstack-dns | 15:28 | |
| dmsimard | o/ have a question regarding the bind backend | 15:36 |
| dmsimard | puppet-designate drops a line in bind config to include /var/lib/designate/bind9/zones.config (and it creates that file but leaves it empty) | 15:37 |
| *** eandersson_ has quit IRC | 15:37 | |
| dmsimard | I don't have a lot of knowledge around designate but I don't see mentions of that in http://docs.openstack.org/developer/designate/getting-started.html | 15:37 |
| dmsimard | So should puppet-designate stop doing that or something ? | 15:38 |
| dmsimard | It otherwise involves getting bind to be able to read that configuration file (named-checkconf and afterwards include it) so that includes permissions (and on CentOS, selinux) considerations | 15:39 |
| elarson | dmsimard: I think puppet-designate doesn't need to do that | 15:40 |
| elarson | especially if it leaves it empty | 15:41 |
| dmsimard | I don't know what's the intended purpose of that file :( | 15:41 |
| elarson | me neither | 15:41 |
| dmsimard | It was added back in 2013! https://github.com/openstack/puppet-designate/commit/23f5954f97b3b5822f9b4131caf8e1cd3b5c7e49 | 15:42 |
| mugsie | dmsimard: there is not a lot of overlap between the designate devs and the puppet-designate dev unfoirtunatly | 15:43 |
| dmsimard | mugsie: a lot ? or not a lot ? | 15:44 |
| mugsie | not a lot | 15:44 |
| dmsimard | oh, yeah, I read that wrong | 15:44 |
| dmsimard | It's okay, I'm trying to bring it up to par :p | 15:44 |
| mugsie | I saw the email to the list :) | 15:44 |
| mugsie | thanks! it would be great to have it actually CI'd | 15:45 |
| dmsimard | so you might see me asking some questions here to try and bring it in line with upstream recommendations | 15:45 |
| dmsimard | so I'll get rid of that file then, if someone complains, he'll need a good reason :) | 15:45 |
| *** maestropandy has quit IRC | 15:46 | |
| mugsie | dmsimard: cool - ask away when ever | 15:46 |
| dmsimard | appreciate it, thanks :) | 15:47 |
| timsim | hey mugsie how do I even grenade | 15:49 |
| mugsie | heh | 15:49 |
| timsim | Like, get a devstack vm up | 15:49 |
| timsim | set some nebulous env vars somewhere | 15:49 |
| timsim | run ./grenade.sh ? | 15:50 |
| mugsie | yeah | 15:50 |
| mugsie | if you run the vagrant/contrib vms the files are all mounted in the right place | 15:50 |
| mugsie | timsim: #openstack-qa is a good place as well | 15:51 |
| mugsie | if they actually respond | 15:51 |
| timsim | Why can't there just be a goddam Makefile | 15:51 |
| *** abalutoiu has joined #openstack-dns | 15:51 | |
| mugsie | there is | 15:51 |
| mugsie | tox.ini | 15:51 |
| timsim | No | 15:51 |
| mugsie | just grenade doesnt use it | 15:52 |
| mugsie | :P | 15:52 |
| timsim | weeeeeeeee | 15:52 |
| *** penick has joined #openstack-dns | 15:57 | |
| *** dxu has joined #openstack-dns | 16:06 | |
| dxu | Is there any way to change the name of a recordset after it's been created? | 16:22 |
| dxu | I dug through the source a little and it looks like it's set as immutable | 16:23 |
| mugsie | it is | 16:27 |
| mugsie | we don't currently support rename | 16:27 |
| dxu | :( alright. Thanks | 16:29 |
| *** penick has quit IRC | 16:51 | |
| *** nyechiel has quit IRC | 17:07 | |
| *** gnuoy is now known as jamespage_shadow | 17:17 | |
| *** jamespage_shadow is now known as gnuoy | 17:18 | |
| *** gnuoy is now known as jamespage_shadow | 17:23 | |
| *** jamespage_shadow is now known as gnuoy | 17:25 | |
| *** gnuoy is now known as jamespage_shadow | 17:29 | |
| *** jamespage_shadow is now known as gnuoy | 17:30 | |
| *** gnuoy is now known as jamespage_shadow | 17:35 | |
| *** jamespage_shadow is now known as gnuoy | 17:36 | |
| *** gnuoy is now known as jamespage_shadow | 17:45 | |
| *** jamespage_shadow is now known as gnuoy | 17:48 | |
| *** rudrajit has joined #openstack-dns | 17:50 | |
| *** rudrajit has quit IRC | 17:55 | |
| *** ducttape_ has joined #openstack-dns | 18:15 | |
| dmsimard | I'm trying to test designate with tempest, typically we run smoke but I see that no designate tests are tagged as such. | 18:28 |
| dmsimard | What would be a fitting regex to run a viable amount of tests for designate ? | 18:28 |
| timsim | I believe there are tests marked as smoke in designate-tempest-plugin | 18:29 |
| *** ducttape_ has quit IRC | 18:29 | |
| dmsimard | oh, I might be looking at the wrong place | 18:30 |
| *** ducttape_ has joined #openstack-dns | 18:30 | |
| dmsimard | ah, there's https://github.com/openstack/designate-tempest-plugin .. I was looking in in-tree functional tests | 18:30 |
| *** penick has joined #openstack-dns | 18:31 | |
| mugsie | dmsimard: yeah - we moved to an out of tree plugin | 18:52 |
| mugsie | just pip install it, then running `tox -e all-plugin -- designate` gets our tests | 18:53 |
| dmsimard | mugsie: testing it with RDO packaging and with designate installed with puppet-designate right now ;) | 18:54 |
| mugsie | sweet :) | 18:55 |
| *** penick has quit IRC | 18:56 | |
| *** penick has joined #openstack-dns | 18:59 | |
| dxu | I was reading the documentation, and if I'm reading it right, it says there's currently no way to implement access controls. Is that right? | 19:02 |
| dxu | By access control I mean something like | 19:02 |
| dxu | Only the people in a certain project X can delete records created by people in project X | 19:02 |
| timsim | dxu: There is. If you're using keystone, and the designate keystone middleware you have to authenticate with a tenant/project before you do anything and that will separate things by tenant | 19:03 |
| timsim | You can do it without keystone too by passing certain headers | 19:04 |
| timsim | But it's generally assumed that you're using keystone | 19:04 |
| dxu | timsim: Cool! Is there documentation I can take a look at? | 19:05 |
| openstackgerrit | Graham Hayes proposed openstack/designate-tempest-plugin: Add functional testing for the v2 API quotas endpoint https://review.openstack.org/337321 | 19:06 |
| openstackgerrit | Graham Hayes proposed openstack/designate: Add v2/quotas https://review.openstack.org/335100 | 19:06 |
| timsim | dxu: I think you just need to configure the keystone_authtoken section of the config to make it work with your existing keystone | 19:07 |
| timsim | That's in step 6 of "configure designate" here http://docs.openstack.org/developer/designate/install/ubuntu-liberty.html | 19:07 |
| mugsie | dxu: we scope zones to projects - so only people in project x can even see the records exist | 19:08 |
| mugsie | we do not allow zones to be shared across projects yet | 19:08 |
| mugsie | (it is on the list, but not a priority yet - it is also difficult to do well, and with decent performance) | 19:09 |
| dxu | Ok, thanks. I'll take a look | 19:10 |
| ducttape_ | mugsie: question on that ux review. is everything going zones? I don't see the term domains anywhere etc | 19:13 |
| ducttape_ | apologies if that is an extra derpy question | 19:14 |
| timsim | ducttape_: Yeah in the designate v2 api, domains -> zones | 19:19 |
| timsim | Similarly, records -> recordsets, where recordsets are actually sets of "records" | 19:20 |
| mugsie | ducttape_:yeah keystone won that battle | 19:22 |
| ducttape_ | if keystone domains (v3) was the winner - not sure what the loser looks like ;) | 19:23 |
| mugsie | haah | 19:25 |
| mugsie | yeah | 19:25 |
| *** penick has quit IRC | 19:27 | |
| *** penick has joined #openstack-dns | 19:36 | |
| *** rudrajit has joined #openstack-dns | 19:57 | |
| timsim | Is that really why we did that? | 19:57 |
| *** rudrajit has quit IRC | 20:02 | |
| *** rudrajit has joined #openstack-dns | 20:02 | |
| mugsie | timsim: yes | 20:23 |
| mugsie | and it is technically more correct | 20:23 |
| timsim | Yeah like how recordsets are more correct. | 20:30 |
| * timsim can't count the number of times he's had to explain zones/recordsets to people :( | 20:31 | |
| openstackgerrit | Emilien Macchi proposed openstack/designate-tempest-plugin: Remove 'smoke' decorator usage https://review.openstack.org/338549 | 20:43 |
| openstackgerrit | Emilien Macchi proposed openstack/designate-tempest-plugin: Remove 'smoke' decorator usage https://review.openstack.org/338549 | 20:45 |
| mugsie | timsim: well it is more correct | 20:56 |
| mugsie | :P | 20:56 |
| mugsie | RAX were the ones who wanted v2/zones/id/recordsets/id/records dropped ;) | 20:56 |
| timsim | Turns out no one cares what's correct | 20:57 |
| timsim | Much to my chagrin | 20:57 |
| timsim | RAX != me in that scenario | 20:57 |
| * mugsie mumbles something about golang | 20:57 | |
| * timsim mumbles something about api v3 | 20:57 | |
| * timsim in golang | 20:57 | |
| mugsie | well, you worked there at the time - good enough for me :) | 20:57 |
| mugsie | but, yea - it is a bit complicated | 20:57 |
| mugsie | and having domain / records as well as zone / recordsets makes it all much harder | 20:58 |
| *** EmilienM has joined #openstack-dns | 20:59 | |
| EmilienM | hello! | 20:59 |
| EmilienM | I'm working with dmsimard on testing Designate in upstream Puppet CI and we would need this patch to be reviewed: https://review.openstack.org/#/c/338549/ - thanks! | 20:59 |
| mugsie | EmilienM: I am looking at that now | 20:59 |
| EmilienM | w00t awesome | 21:00 |
| mugsie | not going to lie, it looks like a -1 at this point. We cannot remove it entirely, as people may be using it already | 21:01 |
| dmsimard | mugsie: so could we settle for something that is more smoke-ish than "let's run most of the tests" ? | 21:02 |
| dmsimard | smoke is meant to try and make sure the service is not horribly broken | 21:02 |
| mugsie | cutting it back could be considered, but I would like to get pglass and Kiall to feed in on it, as they have done a lot of the inital work | 21:02 |
| dmsimard | designate has more smoke tests than, like, nova and neutron combined :p | 21:04 |
| pglass | yeah. i would prefer a smaller set of smoke tests, rather than lots of cheap tests. we should also have a few of the smoke tests actually query the nameservers. | 21:04 |
| pglass | so then the smoke tests by themselves are a reliable indicator that your designate is good to go, without taking a long time to run. | 21:05 |
| EmilienM | whatever you decide please go over 5 tests, that's already too much | 21:06 |
| mugsie | OK, look through what nova / neutron etc do, we can (and should cut back) | 21:06 |
| EmilienM | our puppet CI aims to validate designate is well configured and is actually working (API, etc) | 21:06 |
| mugsie | so, senario tests then? | 21:06 |
| pglass | 5 tests is too few | 21:06 |
| mugsie | that will include tests that query the nameservers | 21:06 |
| pglass | imo. but I'll follow nova's lead. | 21:06 |
| EmilienM | pglass: our CI runs more than designate | 21:06 |
| EmilienM | pglass: see https://github.com/openstack/puppet-openstack-integration#description for coverage | 21:07 |
| pglass | well, you should filter out more tests then if you don't want to run all of them. | 21:07 |
| mugsie | the smoke tests that seem to get run by others are for checking the API root, and "discovery" (which is static content) | 21:07 |
| EmilienM | the more you add in smoke, more time our CI will take and maybe timeout | 21:07 |
| mugsie | so, neutron has 15 - 20 smoke tests in the api folder alone | 21:09 |
| EmilienM | fwiw, other projects don't use the tag for lot of tests (ceilometer, gnocchi, aodh, etc) | 21:11 |
| pglass | EmilienM: do you know roughly how long an individual test for nova or neutron takes? e.g. are they spinning up servers in those tests? | 21:11 |
| EmilienM | really, please don't use it for too may tests | 21:11 |
| EmilienM | yes | 21:11 |
| EmilienM | a sec | 21:11 |
| EmilienM | http://logs.openstack.org/98/337898/4/check/gate-puppet-openstack-integration-3-scenario003-tempest-centos-7/1f6904a/logs/stackviz/#/testrepository.subunit/timeline | 21:12 |
| EmilienM | see the graph | 21:12 |
| EmilienM | pglass: scenarios take 8 min each | 21:13 |
| EmilienM | they spin up VM, create volumes, etc | 21:13 |
| mugsie | so you run senarios as well as smoke? | 21:14 |
| mugsie | have you looked at increasing your worker count? | 21:14 |
| *** mtreinish has joined #openstack-dns | 21:14 | |
| pglass | even with testing propagation to nameservers, none of our scenario tests takes longer than two minutes. | 21:14 |
| mugsie | yeah | 21:14 |
| dmsimard | mugsie: we're ram constrained, there's a lot of stuff going on on these VMs | 21:15 |
| dmsimard | upping the workers would mean having to split into a fourth scenario | 21:15 |
| dmsimard | which, while nice, is a heavier burden on -infra | 21:15 |
| mugsie | so, dropping the number of smoke API tests, and running a few of our senarios might be a good split | 21:15 |
| dmsimard | I have to brb | 21:16 |
| *** dmsimard is now known as dmsimard|afk | 21:16 | |
| EmilienM | mtreinish: we're discussing about the number of smoke tests to set | 21:16 |
| EmilienM | I would argue for the minimum | 21:16 |
| mugsie | what is the current thought on "minimum" ? | 21:17 |
| mugsie | we should test both API versions | 21:17 |
| mtreinish | well our guidline in tempest for smoke is that a full smoke run should not exceed 10min | 21:17 |
| EmilienM | I would ask for > 5 and < 10 | 21:17 |
| mtreinish | which the selection on tempest is pretty close to | 21:17 |
| mugsie | really? even with all the neutron ones? | 21:18 |
| mtreinish | for plugins I don't have a rule of thumb but I'd almost argue that they shouldn't overload that tag | 21:18 |
| EmilienM | right but some CIs (like Puppet) run more than Designate service, we have timeout constraint | 21:18 |
| EmilienM | mtreinish: just designate, max 10 | 21:18 |
| mugsie | and that is fine, but if we are getting commits to remove smoke entirely - I would expect the same to be happening to LBaaS / VPN etc | 21:19 |
| EmilienM | I have to brb too, I'll catch up on Gerrit and here. Thanks guys - please feel free to take over my patch to select what you want to keep | 21:19 |
| pglass | okay, so what I would do is 8 tests - c/u/d zones, c/u/d recordsets, and then a zone import and a zone export. | 21:20 |
| EmilienM | sounds cool | 21:20 |
| pglass | with one or two of those being a scenario that tests propagation out to the nameserver. | 21:20 |
| mugsie | pglass: we need to cover v1 as well | 21:20 |
| EmilienM | pglass: you can push over my patch, I really don't mind | 21:20 |
| mugsie | I know it is going away, but until i fully rm -rf /v1/ we need to test it, as horizon + cli still use it | 21:21 |
| pglass | mugsie: okay. so we can exchange the recordset tests for c/u/d v1 domains | 21:21 |
| pglass | and maybe instead of a zone or domain update we do a recordset or record create. | 21:22 |
| mugsie | yeah, that could work | 21:23 |
| mugsie | mtreinish: when you say "shouldnt overload that tag" do you mean use it? | 21:23 |
| *** _ducttape_ has joined #openstack-dns | 21:24 | |
| mtreinish | mugsie: right, have an equiv definition of what smoke is for the designate plugin, but don't use the tag 'smoke' | 21:24 |
| mtreinish | people tagging everything smoke in plugins has come up before and causes issues for people. Especially since everyone seems to have a different definition for it | 21:25 |
| mugsie | OK, I have bigger problems with that statement, but that is a discussion for another day. | 21:25 |
| mugsie | so, tempest team documenting a definition might be a step forward | 21:25 |
| mugsie | instead of "to do a quick test to check you clkoud do smoke test, unless you use projects who are not in the special club, then run $other thing as well" | 21:26 |
| *** ducttape_ has quit IRC | 21:28 | |
| mtreinish | it doesn't apply to plugins, it can't. It's a selection of things in the projects tempest supports in tree. We have our own guidelines for that selection. But, because there is a temporal component in that guideline as you add more plugins you can't enforce that | 21:28 |
| mtreinish | overloading the tag in plugins just breaks users expectations, especially when they accidently install plugins (which happens more often then you'd think) | 21:28 |
| mugsie | i would argue *not* using it breaks users expectations | 21:29 |
| mugsie | "I installed the designate tempest tests, the smoke says everythig was OK, but I can't create a zone. Are your tests broken?" | 21:30 |
| mugsie | if your guideline can't match the technical design tempest went for, the guideline should change | 21:30 |
| mugsie | or change the design, but something needs to change. | 21:32 |
| mtreinish | that's assuming the tag is something we support externally, which I'm saying it isn't. The tag only applies to what's in tree, which is why plugins overloading it causes issues. | 21:32 |
| mtreinish | but plugins can do whatever they want, that's kinda the point | 21:32 |
| mtreinish | and if you want to tag things as smoke go ahead | 21:32 |
| mtreinish | I'm just saying what the definition of the tag in tempest means, and plugins adding on to that will break that definition | 21:33 |
| mugsie | can I suggest that this gets documented somewhere | 21:33 |
| mtreinish | choosing to extend it like you did here will go against what people are expecting, which is why you have patches like EmilienM's because it ran too long and consumed too many resources | 21:34 |
| pglass | we didn't make that decision deliberately. | 21:34 |
| mugsie | and we followed what most other plugins had done | 21:34 |
| mtreinish | sure, I wasn't saying it was intentional | 21:34 |
| mtreinish | it just gets cargo culted | 21:34 |
| mtreinish | a couple of the neutron plugins did the same thing | 21:34 |
| mugsie | which circles back to the documentation of it | 21:34 |
| pglass | yeah. i'm fine following a rule. | 21:35 |
| mugsie | because asking questions in #openstack-qa is like shouting into the wind | 21:35 |
| pglass | but like how we had to discover which tempest modules are stable, we have to discover this as well via trial and error. | 21:35 |
| mugsie | I just want clarity, and a way for us to avoid hoops, and other issues. | 21:36 |
| mtreinish | I thought we had a doc for smoke somewhere, but I can't find it | 21:36 |
| mtreinish | mugsie: fwiw, the stable interfaces thing was documented. It just wasn't clearly worded. (it was in the middle of a paragraph) Kiall pushed a patch to clean that up, which makes it obvious now: http://docs.openstack.org/developer/tempest/plugin.html#stable-tempest-apis-plugins-may-use | 21:37 |
| pglass | also, fwiw even with all of those tests tagged as smoke tests, they still run under 10 minutes. | 21:39 |
| mugsie | mtreinish: yeah, thats fine (now) but finding out how these things are supposed to work can be very time cosuming, and involve a ton of trial and error | 21:39 |
| mtreinish | pglass: by themselves right? what happens if you add 4 or 5 other plugins into the mix? | 21:40 |
| pglass | i mean: `Ran: 115 tests in 104.0000 sec.` | 21:40 |
| mtreinish | pglass: or include tests in tempest itself? | 21:40 |
| mugsie | mtreinish: the problem is, this is creating 2 tiers of projects again. if plugins cannot get in to the smoke tests, they will not be tested in real world envs. | 21:42 |
| pglass | mtreinish: again, if "smoke" means that all tests that ever might be run by tempest including with all plugins enabled needs to run in less than X minutes - that needs to be documented. | 21:42 |
| pglass | that is a really strict requirement. | 21:42 |
| mugsie | unless there is a clear "this is how plugins add smoke like tests, and this is how operators can run them in a single command" | 21:42 |
| mtreinish | that's what I'm trying to say, it's not a strict requirment (it's only enforcable that way in tempest) | 21:42 |
| mtreinish | but because it's taht way in tempest that's the expecation from users | 21:43 |
| mtreinish | like I said as a plugin you're able to do what you want | 21:43 |
| pglass | mtreinish: right, but there are real restrictions like job timeouts | 21:43 |
| mugsie | but do they not have the expectation of testing all the services in the cloud? | 21:43 |
| *** EricGonczer_ has quit IRC | 21:44 | |
| mtreinish | pglass: right, but that's normally not an issue (well except for puppet jobs) because no one besides puppet runs >1 plugin in a ci job. Normally the selection regex is $service_name | 21:44 |
| mtreinish | mugsie: running with smoke as your selection regex isn't how I expect most people to run tempest, and I normally don't recommend it except for a quick sanity check after making a change. I think you're overvaluing it's importance | 21:46 |
| mtreinish | for example in the gate, except for puppet we only use it on grenade | 21:46 |
| mugsie | so, important for us then, as we are writing our grenade test - and we want to make sure that say a keystone update doesn't break our API, or the neutron interface for floating IP lists are still working | 21:48 |
| mtreinish | we've discussed adding an option in the grenade plugin interface to add onto the tempest selection regex in the past. That seems like totally valid thing to add to a grenade | 21:50 |
| mtreinish | so a grenade plugin can export a regex for additional tests they want for running things and because by default testr ORs the regexes that would just work | 21:51 |
| openstackgerrit | Paul Glass proposed openstack/designate-tempest-plugin: Reduce the number of "smoke" tests https://review.openstack.org/338549 | 21:59 |
| *** _ducttape_ has quit IRC | 22:03 | |
| *** ducttape_ has joined #openstack-dns | 22:04 | |
| pglass | EmilienM: ^ | 22:12 |
| *** ducttape_ has quit IRC | 22:17 | |
| *** pglass has quit IRC | 22:17 | |
| *** penick has quit IRC | 22:38 | |
| *** mlavalle has quit IRC | 22:43 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!