Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-11-06
15:57:06 bauzas I maybe missed the point then
15:57:08 lyarwood so we raise
15:57:18 lyarwood L139
15:57:21 bauzas oh, that's just for logging purpose ?
15:57:24 lyarwood right
15:57:24 dansmith bauzas: the point is not to "try again", the point is to generate the import exception just for logging
15:57:35 bauzas hah, ok
15:57:46 bauzas hence the F401
15:57:55 bauzas because the local import wasn't used
15:58:00 sean-k-mooney its something danpb suggested wehn we were trying to figure out why imports didnt work
15:58:23 bauzas lyarwood: but fwiw, imports aren't global
15:58:25 sean-k-mooney we think it was due to memory issues in the custoemr env since the had a bunc of OOM issue at the time
15:58:32 sean-k-mooney but we dont have any logs to say
15:58:46 sean-k-mooney bauzas: import happen in your current scope
15:58:50 openstackgerrit Lee Yarwood proposed openstack/nova master: rbd: Only log import failures when the RbdDriver is used https://review.opendev.org/761762
15:59:00 bauzas sean-k-mooney: right, that's what I said
15:59:09 sean-k-mooney so if you are at global scope they are if you are in function scope tehy are just in that scope
15:59:26 sean-k-mooney although that is only half though
15:59:29 sean-k-mooney *true
15:59:40 bauzas either way
15:59:56 sean-k-mooney the modul is actully imported gloablly including inialising all its global state it just made available in your local scope
16:00:14 sean-k-mooney if you import it somewere else that modules globalse state is still there
16:00:27 bauzas ok, the docstring helps
16:01:08 bauzas dansmith: honestly, I won't bother for tests here
16:02:00 dansmith lyarwood: couple comments on your comments
16:02:42 dansmith bauzas: well, I understand but a typo in a log message in an exception handler really defeats the point of the helper.. but I'd rather get it fixed than complicate it for testability
16:03:28 bauzas dansmith: fwiw I agree with your comments
16:04:03 bauzas I guess we need both rbd and rados modules to be present in order to work
16:04:18 bauzas so only one exception catch should be enough
16:04:51 dansmith ...like in the actual import.. I'm not sure why it's split in the original change or this one, but maybe lyarwood has a reason
16:07:09 lyarwood yeah I don't recall why, likely just leaving specific breadcrumbs but that's part of the issue with the first patch so I'll merge them back togther now
16:07:22 bauzas I guess because the log is different
16:07:32 dansmith sure, but the exception trace will tell you which one
16:07:40 bauzas yup, I was able to tell it
16:13:16 openstackgerrit Lee Yarwood proposed openstack/nova master: rbd: Only log import failures when the RbdDriver is used https://review.opendev.org/761762
16:14:21 dansmith gibi: ^
16:14:29 gibi looking..
16:14:41 dansmith lyarwood: wait, one more thing
16:15:00 lyarwood ^_^
16:16:23 bauzas dansmith: oh shit, I +Wd
16:16:34 dansmith bauzas: I was just joking
16:16:38 bauzas hah
16:16:53 dansmith but I figured gibi should look too
16:16:58 bauzas anyway, time to call it a wrap
16:17:20 bauzas gibi: please look at https://review.opendev.org/761762 which is going to the gate for the next hours
16:17:28 bauzas anyway, \o
16:18:31 bauzas man, I just found we powerplayed /o\
16:18:53 bauzas apologies
16:19:33 gibi no worries. It looks good to me too
16:19:35 openstackgerrit Sylvain Bauza proposed openstack/nova master: rbd: Only log import failures when the RbdDriver is used https://review.opendev.org/761762
16:20:09 gibi then I will add the Closes-Bug tag the the commit message ^^
16:20:34 gibi or not :D
16:20:39 gibi whatever
16:21:08 gibi bauzas: thanks for the commit message update
16:46:56 stephenfin sean-k-mooney: comment on https://review.opendev.org/#/c/751642/8/nova/virt/libvirt/driver.py@1588
16:53:45 sean-k-mooney i was asked to log it and assert the excption is caught
16:53:53 sean-k-mooney if i remove the logging i cant do either
16:54:14 sean-k-mooney im useing log.debug so i dont think that prints the stack trace the way log.excpetion does
16:55:58 sean-k-mooney stephenfin: im expecting it to log the excption message not the traceback
16:57:06 stephenfin sean-k-mooney: aha, right you are. You need to pass a flag for that to happen
17:21:58 openstackgerrit Merged openstack/os-vif stable/victoria: Refactor code of linux_net to more cleaner and increase performace https://review.opendev.org/756406
17:21:59 openstackgerrit Merged openstack/os-vif stable/victoria: Fix - os-vif fails to get the correct UpLink Representor https://review.opendev.org/759255
17:44:38 stephenfin gmann, johnthetubaguy: Are either of you going to have time to review https://review.opendev.org/#/q/topic:secure-rbac+(status:open+OR+status:merged)+project:openstack/placement with me. Would definitely benefit from your input
17:47:03 stephenfin lyarwood: I can respin https://review.opendev.org/#/c/761762/ real quick if you want?
17:51:47 openstackgerrit Stephen Finucane proposed openstack/nova master: rbd: Only log import failures when the RbdDriver is used https://review.opendev.org/761762
18:08:52 lyarwood stephenfin: yeah thanks, slightly confused how this worked before without this?
18:09:26 stephenfin you were only reading the value of the variable
18:09:38 stephenfin the global keyword is necessary when you write to it
18:10:39 stephenfin foo = None
18:10:39 stephenfin def bar():
18:10:40 stephenfin foo = 'test'
18:10:40 stephenfin bar()
18:10:40 stephenfin print(foo)
18:10:50 stephenfin ^
18:12:18 lyarwood right so the error is misleading, it's not the read in the conditional that's at fault, it's the import trying to overwrite it?
18:12:44 stephenfin sort of
18:12:54 stephenfin by trying to write to it, the variable becomes locally scoped
18:13:02 lyarwood yup got it
18:13:38 stephenfin flake8 can't catch these kind of issues, unfortunately
18:14:54 sean-k-mooney lyarwood: before it was locally importing every time and using the local import if it was not visable
18:15:08 sean-k-mooney you can read gloabls without the keyword
18:15:24 sean-k-mooney you just can write to them it creates a local alias of it
18:16:52 lyarwood sean-k-mooney: yeah got it
18:17:27 lyarwood stephenfin: still failing btw
18:17:42 stephenfin darn, same error?
18:17:56 lyarwood stephenfin: different, this looks more like a test issue
18:18:00 lyarwood stephenfin: https://7d2bb7e91c376d06349f-0e5efcd7961a07f1b6cc368c11ff4914.ssl.cf1.rackcdn.com/761762/5/check/openstack-tox-cover/b207b75/testr_results.html
18:19:18 lyarwood stephenfin: I need to drop now, feel free to leave that and I can address it on Monday
18:19:34 stephenfin Yeah, it's not the easy fix we though so let's leave it
18:19:45 stephenfin lyarwood: Have a good one o/
18:20:42 lyarwood and you \o
18:23:30 sean-k-mooney damit... i dont have precommit on my laptop and didnt run pep8
18:28:59 mnaser i've been confused about this for aw hile
18:29:20 mnaser say i have a queens install and enabled_vgpu_types was changed -- do i have to go muck about the placement service?
18:29:47 mnaser i cant find where this is 'defined' and being 'published' to placement
18:32:12 mnaser inventories only shows 3 resorce classes which seem to be disk/vcpu/mem
18:39:53 gmann stephenfin: ack, lance and I are going to have call on Tuesday for testing strategy for these changes, at least we need unit test like we did in nova.
18:40:24 gmann stephenfin: these are in my list for next week.
18:40:59 gmann these are meeting details - http://lists.openstack.org/pipermail/openstack-discuss/2020-November/018512.html
19:40:53 mnaser `virsh nodedev-dumpxml` shows availableInstances = 0

Earlier   Later