Monday, 2017-03-20

openstackgerritzhangyanxian proposed openstack/ceilometer-powervm master: Optimize the link address for fetching git code
*** tjakobs has joined #openstack-powervm05:39
*** tjakobs has quit IRC05:51
*** k0da has joined #openstack-powervm08:30
*** openstackgerrit has quit IRC09:02
*** a1fisher has quit IRC09:52
*** a1fisher has joined #openstack-powervm09:53
*** smatzek has joined #openstack-powervm11:17
*** kylek3h has quit IRC11:55
*** svenkat has joined #openstack-powervm12:04
*** edmondsw has joined #openstack-powervm12:14
*** kylek3h has joined #openstack-powervm12:37
*** mdrabe has joined #openstack-powervm12:50
*** a1fisher has left #openstack-powervm13:08
*** jpasqualetto has joined #openstack-powervm13:13
*** tblakes has joined #openstack-powervm13:18
*** tblakes has quit IRC13:19
*** tblakes has joined #openstack-powervm13:19
*** smatzek has quit IRC13:32
*** tjakobs has joined #openstack-powervm13:46
*** kriskend has joined #openstack-powervm13:53
*** tblakes has quit IRC13:57
*** kriskend_ has joined #openstack-powervm13:58
*** kriskend_ has quit IRC13:59
*** smatzek has joined #openstack-powervm14:00
*** kriskend has quit IRC14:00
*** esberglu has joined #openstack-powervm14:00
*** a1fisher has joined #openstack-powervm14:00
*** jpasqualetto has quit IRC14:09
*** tblakes has joined #openstack-powervm14:15
esbergluPowerVM CI is currently experiencing issues. It appears to be related to issues with the network.14:20
esbergluCI runs may fail and will require a recheck once the network issues are resolved14:21
*** jpasqualetto has joined #openstack-powervm14:22
*** burgerk has joined #openstack-powervm14:23
efriedboo.  Thanks for the heads up.14:24
*** apearson has joined #openstack-powervm14:24
esbergluCan't get on to the systems to look at what is impacted atm14:25
adreznecI think there's a ticket open for lab network issues14:25
adreznecSo we'll be in rough shape for a while until that's resolved14:25
efriedThe only thing I was really concerned about was
*** tjakobs has quit IRC14:26
efriedThat guy has been in flight for a while.  It's not urgent.  But was ready to get it closed out.14:27
efriedSo no big deal.14:27
*** efried has quit IRC14:39
*** jpasqualetto has quit IRC14:49
*** tjakobs has joined #openstack-powervm14:51
*** jpasqualetto has joined #openstack-powervm15:11
*** jpasqualetto has quit IRC15:28
*** efried has joined #openstack-powervm15:35
*** jpasqualetto has joined #openstack-powervm15:51
*** a1fisher has quit IRC16:22
*** openstackgerrit has joined #openstack-powervm16:22
openstackgerritMatt Rabe proposed openstack/nova-powervm master: Don't let dlt_vopt exceptions bubble up in CreateAndConnectCfgDrive reversion
esbergluCI is up and running again16:32
esbergluefried: Kicked off another run on the change you linked above16:33
*** k0da has quit IRC17:03
*** a1fisher has joined #openstack-powervm17:36
*** antonym has joined #openstack-powervm17:41
*** a1fisher has quit IRC17:47
*** jpasqualetto has quit IRC17:47
*** jpasqualetto has joined #openstack-powervm17:59
svenkatefried: I seek your attention for an issue around rescheduling VM and allocation of network resouces around it. When a VM reschedule fails, network allocation is not released and new network resource is allocated for second attempt. If both attempts fail, both resources are released. If second attempt succeeds, first allocation stays in ‘unbound’ state until it is cleaned up. I found a fix for this.  In nova/compute/manager - ComputeManager18:15
svenkatclass - _do_build_and_run_instance method, there is a call for self.driver.deallocate_networks_on_reschedule(instance). If this method returns True for specific driver, network is cleanedup if reschedule fails. (when _build_and_run_instance fails and throws RescheduledException). Since nova_powervm driver does not override deallocate_networks_on_reschedule and default implementation in nova/virt/driver returns False, for Novalink based VM18:15
svenkatreschedules, network resources are not cleaned up.  I will put together a patchset for nova_powervm with an override that returns True. ( i see it is done already in community ironic nova driver).18:15
svenkatlet me know if there are concerns and need discussions here.18:19
*** a1fisher has joined #openstack-powervm19:02
*** jayasankar_ has joined #openstack-powervm19:19
jayasankar_efried: I ran the current tempest whitelist with Intree driver change set and ended up with 29 failures out of Total tests:     80519:22
jayasankar_do we have to fix all those failures on the changeset 409402 ? or have to run them or the first changeset ?19:23
*** k0da has joined #openstack-powervm19:25
efriedsvenkat Good deal, thank you.  If there is not already a launchpad bug, please open one, and subscribe me & thorst19:30
efriedjayasankar_ We have to investigate those failures.  I'd like you to work with esberglu, as we don't expect to have any failures, because you're using the same config and whitelist as the CI environment.19:31
svenkatefried: sure. I will do.19:31
efriedI'm guessing you're seeing failures that the CI env avoids by configuring default networks just so.19:31
esberglujayansankar_: I'm looking at your results right now19:38
esbergluInvalid input for field/attribute imageRef. Value: -04-01_2g_id. u'-04-01_2g_id' is not valid under any of the given schemas19:39
esbergluLooks like you don't have the correct image_ref set in tempest.conf19:39
esbergluDid you set that yourself? Or did you try to use the script to set that?19:40
jayasankar_I used os_ci_tempest in which I gave the correct image location19:41
esbergluAs I've said before, the script is tailored specifically for the CI environment19:45
esbergluSo I'm suspecting that it is probably not working correctly out-of-the box19:45
esbergluI'm looking at the script right now to see where it might be going wrong19:45
*** smatzek has quit IRC19:47
esbergluYep. If you look in your tempest.conf file19:47
esberglu /opt/stack/powervm-ci/tempest19:47
esberglu /opt/stack/powervm-ci/tempest/tempest.conf19:47
esbergluYou will see that image_ref, image_ref_alt, flavor_ref, flavor_ref_alt19:48
esbergluare not set19:48
adreznecjayasankar_ is there a reason you're trying to use os_ci_tempest instead of just following the standard tempest config process? That script is super-specific to our exact CI environment, which means it won't work for anything else without manual work or effort from esberglu and/or efried to make it more generic19:48
adreznecWhich, to be frank, doesn't seem worth the effort19:49
efriedadreznec Not sure I agree that it's super specific to our CI env.19:49
esbergluadreznec: Yes I agree, I think there may have been some confusion between me and jayasankar_19:49
efriedAny confusion would have been my doing.  I definitely suggested that jayasankar_ use as a jumping off point.19:50
efriedIf those values aren't getting set by, it's going to be because of an issue with os_ci_tempest.conf19:51
esberglujayasankar_: The tempest.conf file found in the powervm-ci repo should have nearly everything set for you19:51
esbergluExcept for a few sections19:51
adreznecefried: Super-specific might be a bit of an exaggeration, but the conf it generates isn't going to be generically compatible for non-CI tempest environments19:51
esbergluLook for the comments about vars being set dynamically19:51
adreznecWhich means that jayasankar_ is going to have to do manual config work either way19:51
efriedadreznec in what way(s)?19:52
esbergluThose are the ones you will have to set yourself19:52
efriedRegardless of whether we're CIing or not, jayasankar_'s env is going to have to have similar tweaks in terms of default networks etc. in order to function properly.19:52
efriedBut other than that, is *specifically* designed to set up all the right flavors, images, etc. to make tempest run without a lot of manual work.19:53
efriedIt was written to help the CI out, but there's nothing CI-specific about it.19:53
efried(name notwithstanding)19:53
esberglujayasankar_: Did you set the IMGFILE variable in os_ci_tempest_intree.conf?20:00
adreznecefried: Sorry, had to switch locations for a meeting. I thought there was stuff in there specific to using SSP, configuring the specific private network range, etc, which are things that jayasankar_ certainly could have his environment configured to use, but wouldn't necessarily be true since the goal of this was to test out things outside our standard CI config.20:01
efriedadreznec Testing things outside our standard CI config is a fair point.  But yeah, jayasankar_ has SSP etc., so there's really nothing I would expect not to work.20:02
adreznecNot saying he couldn't start with it, but being able to manually change the tempest config is going to be a requirement to drive the different test configs here20:02
efriedadreznec Definitely.20:03
adreznecAs long as jayasankar_ understands what the os_ci_tempest scripting is actually doing to set up the environment then you guys can definitely help him to get it working20:04
adreznecJust don't want to spend time spinning on something that is only a short-term solution20:05
esberglujayasankar_: Have you looked through that script at all?20:05
efriedadreznec Yeah, okay.  jayasankar_ the point of you using is to get your environment set up so that you can run tempest.  It'll set up flavors, images, etc., and then set up your tempest.conf.  After that, you should probably just run tempest manually.  (You can look at how does that as an example.)20:06
esberglujayasankar_: has an option to just do the prep work and not actually run tempest20:09
esbergluSo to do what efried said above you can use that20:09
esberglujayasankar_: Questions about any of that?20:11
jayasankar_both are one and the same right ? triggering the scripts manually using the same command over triggering the script?20:11
jayasankar_setting the environment or (using the os_ci_tempest script) is temporary for intree at the moment ?20:12
esbergluNot sure I understand your question fully. But no it is not temporary, it is going to continue to be used for both in and out-of-tree CI20:14
jayasankar_ah I understood that the current white list was developed for the first change set, still it is supposed to pass on any of the latest change set right ?20:16
esbergluYou aren't failing because of problems with the driver right now. You are failing because your configuration is not correct for some reason20:17
esbergluIt thinks that the imageRef is "-04-01_2g_id"20:17
esbergluwhich is incorrect20:17
efriedIt may be a matter of special characters being interpreted by the shell.20:18
*** mdrabe_ has joined #openstack-powervm20:18
esbergluWhat do you have IMGFILE set to in os_ci_tempest_intree.conf20:18
efriedI know wtf is happening.20:20 constructs names of stuff using underscores.20:21
efriedthere's evals all over the place.20:22
*** burgerk has quit IRC20:27
*** apearson has quit IRC20:28
*** tblakes has quit IRC20:29
*** mdrabe has quit IRC20:29
esbergluefried: We can just rename the image to get around that right20:32
esberglujayasankar_: ^^ remove the underscores from the image name and update the conf accordingly20:33
*** apearson has joined #openstack-powervm20:33
efriedI'd remove the hyphens too.  Just call it Ubuntu2g.img or something.20:36
jayasankar_Instead of Ubuntu_16-04-01_2g, Ubuntu-16-04-01-2g20:36
efriedI'd remove the hyphens too.  Just call it Ubuntu2g.img or something.20:37
jayasankar_ah okay20:37
efriedI'm not sure if I'm delimiting anything with hyphens in there.  This should be safe, though.20:37
jayasankar_updated the conf, triggering the again..20:43
*** tblakes has joined #openstack-powervm20:43
*** a1fisher has quit IRC20:45
*** apearson has quit IRC20:52
*** kjw3 has joined #openstack-powervm20:52
jayasankar_This time also Failed tests:29, but got the error  [Errno 17] File exists: '/opt/stack/tempest/.testrepository' in between20:57
jayasankar_am looking at the failed cases.. should we delete .testrepository every time before we trigger ?20:58
*** apearson has joined #openstack-powervm20:59
openstackgerritMerged openstack/nova-powervm master: Don't let dlt_vopt exceptions bubble up in CreateAndConnectCfgDrive reversion
*** svenkat has quit IRC21:01
efriedmdrabe_ ocata backport?  ^^21:14
*** mdrabe_ is now known as mdrabe21:14
mdrabeyes please is that okay?21:14
efriedFine by me.21:14
*** tblakes has quit IRC21:22
*** kylek3h has quit IRC21:58
*** jpasqualetto has quit IRC22:01
*** esberglu has quit IRC22:05
*** esberglu has joined #openstack-powervm22:06
*** esberglu has quit IRC22:10
*** mdrabe has quit IRC22:17
*** efried has quit IRC22:24
*** apearson has quit IRC22:25
*** tjakobs has quit IRC22:25
*** edmondsw has quit IRC22:26
*** esberglu has joined #openstack-powervm22:56
*** esberglu has quit IRC23:01
*** k0da has quit IRC23:45

Generated by 2.14.0 by Marius Gedminas - find it at!