17:01:19 <jaypipes> #startmeeting
17:01:20 <openstack> Meeting started Thu Jul 26 17:01:19 2012 UTC.  The chair is jaypipes. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:02:42 <jaypipes> davidkranz, rohitk, Ravikumar_hp: hello!
17:02:50 <Ravikumar_hp> hi
17:02:51 <rohitk> howdy!
17:02:54 <jaypipes> welcvome to devananda too :)
17:03:05 <devananda> hiya
17:03:21 <jaypipes> anyone from RAX here? or know where Daryl is?
17:03:49 <jaypipes> oh well, let's start without him...
17:04:13 <jaypipes> #topic devananda doing changes to tempest for tessting OpenVZ driver
17:04:29 <davidkranz> Hi all.
17:04:31 <annegentle> I'm here but I dunno where Daryl is, sorry
17:04:33 <jaypipes> devananda is wondering how many tests should he focus on for the first round of testing
17:04:59 <jaypipes> devananda: My suggestion would be to have all smoke tests passing against an openvz env
17:05:13 <davidkranz> jaypipes: Sorry, what is OpenVZ?
17:05:29 <jaypipes> devananda: in fact, I doubt that nova will accept the openvz patches without smoke tests passing entirely (it's the upstream gate for Nova now)
17:05:38 <devananda> some things I know don't work (yet), like volume-detach and image-create.
17:05:41 <jaypipes> davidkranz: OpenVZ is a virt driver like LXC
17:05:49 <davidkranz> jaypipes: k
17:05:54 <devananda> so, jaypipes, actually we're not submitting the OpenVZ driver upstream yet
17:05:58 <devananda> and won't be gating on it
17:06:02 <devananda> just smoke testing
17:06:24 <devananda> but to do smoke tests on openvz, we need devstack/tempest and devstack-gate to support it
17:06:25 <jaypipes> devananda: k. what we can do in that case is work with you to: a) create bugs for tracking those issues and b) mark the particular tests skipped when OpenVZ is the virt driver
17:06:50 <devananda> (b) yes, that's what I was looking for :
17:06:51 <devananda> :)
17:06:53 <jaypipes> kk
17:07:18 <jaypipes> devananda: so generally the way we are handling skips has been using the unittest2.skipIf() decorators.
17:07:35 <jaypipes> devananda: in the /tempest/tests/compute/__init__.py file, you will see a setup_module() function
17:08:04 <jaypipes> devananda: you can do some one-time querying of the environment and mark some module-level variable as False when OpenVZ is the virt driver.
17:08:15 <jaypipes> devananda: use the RESIZE_AVAILABLE variable as an example.
17:08:36 <jaypipes> devananda: and *add* a new skipIf to the appropriate smoke test methods for ones that fail with openvz
17:08:45 <jaypipes> devananda: so you would add a new skipper, like so:
17:08:56 <jaypipes> @unittest.skipIf(OPENVZ_INSTALLED)
17:09:03 <devananda> gotcha
17:09:10 <jaypipes> devananda: where OPENVZ_INSTALLED is a module-level variable in tempest.tests.compute
17:09:23 <jaypipes> devananda: that make sense?
17:10:05 <devananda> yep, though i have a similar idea
17:10:17 <jaypipes> k
17:10:41 <devananda> jaypipes: what do you think of, instead of tying all these skips to one variable, I copy the current format and add add'l variables, then have devstack-gate set those appropriately if it is testing openvz?
17:11:03 <devananda> i'm probably creating more work for myself, but it seems better to be flexible, and let devstack-gate manage the environment
17:11:18 <devananda> than have to come back and modify tempest when openvz starts passing some of the tests that I need to skip now
17:11:37 <jaypipes> devananda: I suppose I would have to see the code...
17:11:55 <jaypipes> devananda: I think you may still need to remove skips when/if you get the openvz tests passing
17:12:37 <devananda> jaypipes: for example, the CREATE_IMAGE_ENABLED option that is present now. this works fine and I am able to skip it when testing openvz
17:13:45 <jaypipes> devananda: well I have no problem if you want to default missing config file values to a value present in the shell ENV, if that's what you're suggesting?
17:14:10 <jaypipes> devananda: but devstack/tools/configure_tempest.sh is the thing that would need to be updated as well, not just devstack-gate
17:14:39 <devananda> jaypipes: yes, i was just about to say that :)
17:14:51 <jaypipes> devananda: sure, no problems there...
17:15:50 <devananda> jaypipes: so tempest would need new config options to control some new skipIf() decorators, and devstack/tools/configure_tempest would need to pull those in from the ENV or default them to True
17:15:54 <devananda> jaypipes: does that cover it? :)
17:16:07 <jaypipes> devananda: yuppers.
17:16:54 <jaypipes> devananda: just remember when submitting the tempest patch to include mods to BOTH the tempest.cnf.sample and tempest.cnf.tpl! :)
17:17:07 <devananda> jaypipes: awesome. will start on that after I finish jeblair's suggestion for IMAGE_URLS in devstack
17:17:17 <devananda> jaypipes: hah, thanks! i probably would have overlooked that ;)
17:17:23 <jaypipes> devananda: the sample is useful for folks looking for what info goes in the config. the .tpl file is used by devstsack's configure_tempest.sh script
17:17:42 <jaypipes> OK, devananda, any further questions from your end?
17:18:04 <devananda> not at the moment, jaypipes. thanks again
17:18:09 <jaypipes> coolio.
17:18:19 <jaypipes> rohitk, shall we move on to discuss whitebox tests?
17:18:27 <rohitk> jaypipes: sure
17:18:51 <jaypipes> rohitk: I've read your email from yesterday evening... should be able to get to that immediately after this meeting
17:19:44 <rohitk> jaypipes: do you also recommend to submit a change request to the devstack/tools/configure_tempest.sh file to set the shell ENV variables needed?
17:19:52 <rohitk> by WB tests?
17:20:22 <jaypipes> rohitk: I already have a branch ready for that :)
17:20:37 <rohitk> jaypipes: ok, gr8
17:20:37 <jaypipes> rohitk: not the ENV defautls, but the whitebox variables inserted into the configure_tempest.sh
17:20:48 <rohitk> jaypipes: yes
17:20:51 <jaypipes> k
17:20:57 <davidkranz> We still have a change that is waiting for https://review.openstack.org/#/c/10215/1
17:21:33 <jaypipes> dtroyer: if you have a chance :) ^^
17:22:14 <davidkranz> jaypipes: Is Tempest gating itself at this point?
17:22:24 <jaypipes> rohitk: one thing I have not mentioned in this particular patch code review on the whitebox tests was that I was hoping we can get rid of the /tempest/tests/compute/whitebox directory and just have the whitebox test files in /tempest/tests/compute?
17:22:45 <jaypipes> davidkranz: yes it is. that's partly the reason I spent all yesterday getting the gate working again...
17:22:57 <davidkranz> jaypipes: :)
17:22:59 <rohitk> jaypipes: The class names should be made self explanatory then, which they are currently for WB
17:22:59 <jaypipes> davidkranz: because random changes in Nova/Glance/Keystone can now break the tempest gate
17:23:10 <rohitk> jaypipes: sure
17:23:16 <davidkranz> jaypipes: Good. Great work.
17:24:42 <jaypipes> donaldngo, Ravikumar_hp: soemthing you guys might be interested in, but the CI team (clarkb specifically) recently added support for Selenium testing in the upstream CI system. Might be worth checking out.
17:25:19 <Ravikumar_hp> jaypipes: That is great
17:25:29 <jaypipes> rohitk, davidkranz: I don't have any futher topics to discuss this week. We've kind of fallen behind on the "QA team member attached to blueprint" goal, but I've personally been swamped with other hthings. :(
17:25:45 <clarkb> https://jenkins.openstack.org/job/gate-horizon-selenium/ is the job
17:25:48 <davidkranz> jaypipes: Me too.
17:25:51 <jaypipes> clarkb: cheers
17:25:52 <donaldngo> jaypipes: cool I'll take a look is it checked into tempest trunk?
17:26:05 <jaypipes> donaldngo: not part of tempest... :)
17:26:06 <Ravikumar_hp> jaypipes: our team is looking into blurprints and starting QA comments
17:26:14 <rohitk> jaypipes: me too, but I have a blueprint for Cinder tests in folsom at which I should be able to follow those goals
17:26:24 <jaypipes> cool.
17:26:27 <donaldngo> gotcha
17:26:31 <Ravikumar_hp> we are focussing on Nova, swift
17:26:37 <jaypipes> good.
17:27:00 <Ravikumar_hp> jaypipes: I want to get an update from daryl on Swift test checkin. Can you please follow up
17:27:35 <jaypipes> Ravikumar_hp: I can try :) you can also email jose.idar@rackspace.com... Jose is the QA person working on the swift tests
17:27:56 <Ravikumar_hp> jaypipes: ok
17:28:20 <jaypipes> OK folks, anything else before we wrap up?
17:28:36 <Ravikumar_hp> nothing from me
17:28:43 <davidkranz> Nope.
17:29:08 <rohitk> me neither
17:29:27 <rohitk> oh wait
17:29:49 <rohitk> i was just if anyone are seeing failures in the volumes tests?
17:29:55 <rohitk> *wondering
17:30:02 <jaypipes> rohitk: with cinder or with nova-volumes?
17:30:07 <jaypipes> jgriffith: ping
17:30:09 <rohitk> tempest gate just run smoketests i think
17:30:13 <rohitk> with cinder
17:30:23 <jgriffith> jaypipes: pong
17:30:30 * jgriffith is reading to catch up
17:30:51 <jaypipes> rohitk: no, I have not seen any failures (and I'm running cinder + devstack locally)
17:31:11 <rohitk> where the nova endpoint forwards this via the python cinderclient
17:31:25 <jaypipes> rohitk: but, that reminds me, we should set the devstack gate to set cinder instead of nova-volumes if that hasn't been done already
17:31:33 <jgriffith> jaypipes: done
17:31:33 <rohitk> hmm, ok, i'll run up my env again
17:31:39 <jaypipes> jgriffith: ah, k.
17:31:46 <jaypipes> rohitk: remember to do:
17:31:46 <rohitk> jaypipes: yes that should be done going forward
17:31:47 <jgriffith> jaypipes: That change landed yesterday afternoon
17:32:14 <jgriffith> rohitk: Depending on what you're seeing... exercises/volume.sh *should* work
17:32:21 <jaypipes> ENABLED_SERVICES .. -n-vol,cinder,c-api,c-sch,c-vol
17:32:58 <rohitk> well, I have the cinder env setup alright, just that the tests in the current tempest suite (test_volumes_get) fail
17:33:04 <jgriffith> Don't forget it's very picky: ENABLED_SERVICES+=,-n-vol,cinder,c-vol,c-api,c-sch
17:33:07 <jgriffith> tbe
17:33:12 <rohitk> specifically create_delete_volume
17:33:16 <jaypipes> jgriffith: beat ya to it :)
17:33:21 <jgriffith> rohitk: Ahhhh
17:33:42 <jgriffith> rohitk: I patched nova/api/openstack/__init yesterday as well :)
17:33:47 <rohitk> Nova sends a ComputeFault when cinderclient returns a 404 after delete vol
17:33:48 <jgriffith> rohitk: I suspect that's the issue
17:33:53 <jaypipes> rohitk: https://jenkins.openstack.org/job/gate-tempest-devstack-vm/4960/testReport/tempest.tests.compute.test_volumes_get/ shows passing in the gate...
17:33:54 <jgriffith> rohitk: Yep, that's the one
17:33:56 <rohitk> jgriffith: hmm
17:34:12 <rohitk> jgriffith: thanks, noted that
17:34:50 <jgriffith> rohitk: https://review.openstack.org/#/c/10200/
17:35:20 <rohitk> jgriffith: cool, i'll pull in the latest code
17:35:25 <jgriffith> rohitk: I patched nova, cinder, cinderclient all yesterday
17:35:29 <jaypipes> rohitk: perhaps it was an "old" version of nova? ;) old == more than an hour old at times!
17:36:01 <rohitk> jaypipes: I can't catch up with the code changes ! :D
17:37:27 <jaypipes> rohitk: heh, I hear ya :)
17:37:36 <jaypipes> OK y'all, let's wrap this up! :)
17:37:38 <jaypipes> #endmeeting