Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2019-10-01
20:43:19 mriedem 4. stop the patch on
20:43:21 mriedem x
20:43:28 mriedem 5. do another thing that hits the real thing that x was patching
20:43:42 mriedem 5 is blowing up b/c it doesn't appear that stop is stopping the patch
20:44:27 mriedem seems like it should work given https://docs.python.org/3.6/library/unittest.mock.html#patch-methods-start-and-stop
20:46:52 mriedem calling mock.patch.stopall() seems to do what 4 isn't
20:48:51 mnaser welp, it restarted happenign again not long after a service restart
20:50:42 mriedem maybe a bug in the mock library
20:52:31 mriedem anyway, i have a workaround, ignore me
20:58:13 efried mriedem: without seeing the code...
20:58:16 efried can I see the code?
20:58:29 efried Cause what you're describing seems like it ought to work
21:00:28 efried mriedem: a classic blunder is calling stop() on the *mock* (rather than the patcher)
21:00:41 mriedem yeah it wasn't that
21:00:50 mriedem p = mock.patch('x')
21:00:52 mriedem p.start()
21:00:55 mriedem ...
21:00:57 mriedem p.stop()
21:01:05 mriedem x() -> kablammo
21:01:22 mriedem err, p = mock.patch('x', side_effect=kablammo)
21:03:04 mriedem threads could be at play somehow, idk
21:03:40 openstackgerrit Matt Riedemann proposed openstack/nova master: Add functional regression test for build part of bug 1781286 https://review.opendev.org/685998
21:03:40 openstack bug 1781286 in OpenStack Compute (nova) "CantStartEngineError in cell conductor during reschedule - get_host_availability_zone up-call" [Medium,In progress] https://launchpad.net/bugs/1781286 - Assigned to Matt Riedemann (mriedem)
21:03:40 openstackgerrit Matt Riedemann proposed openstack/nova master: Add functional regression test for migrate part of bug 1781286 https://review.opendev.org/686017
21:03:40 openstackgerrit Matt Riedemann proposed openstack/nova master: Add Selection.availability_zone field https://review.opendev.org/685807
21:03:40 openstackgerrit Matt Riedemann proposed openstack/nova master: Set Instance AZ from Selection AZ during build reschedule https://review.opendev.org/686047
21:04:07 mriedem if you're a glutton for punishment, what i was doing was at this point https://review.opendev.org/#/c/686047/1/nova/tests/functional/regressions/test_bug_1781286.py@72
21:04:12 efried In [8]: foo()
21:04:12 efried real
21:04:12 efried In [9]: p.start()
21:04:12 efried Out[9]: <MagicMock name='foo' id='139999791540112'>
21:04:12 efried In [10]: foo()
21:04:12 mriedem i changed that to be something like:
21:04:13 efried patched
21:04:13 efried In [11]: p.stop()
21:04:14 efried In [12]: foo()
21:04:14 efried real
21:04:29 mriedem self.agg_mock = mock.patch(...)
21:04:33 mriedem self.agg_mock.start()
21:04:52 mriedem and then later after fake_notifier.wait_for_versioned_notifications where i have mock.patch.stopall(), i had self.agg_mock.stop()
21:04:55 mriedem and that wasn't working
21:05:52 efried because you subsequently called get_by_host?
21:06:05 efried as opposed to build_and_run_instance?
21:06:22 mriedem the test goes on to call self._wait_for_state_change which hits the API which hit the thing i had previously mockd to fail yeah
21:06:51 efried cause if you had called build_and_run_instance again, it would have reinstated your mock.
21:07:08 efried but stop_all would have reverted your mock of build_and_run_instance
21:07:18 mriedem what probably happened is there were 2 mocks
21:07:25 mriedem and i only stopped one
21:07:35 efried mm, because two instances?
21:07:45 mriedem https://github.com/testing-cabal/mock/blob/master/mock/mock.py#L1533
21:07:52 mriedem 2 calls to build_and_run_instance
21:07:54 mriedem 2 hosts
21:07:56 mriedem it's a reschedule test
21:07:59 efried that'd do it :)
21:08:10 mriedem yup, so stopall to the rescue
21:08:22 efried well, except that's going to stop... *all* of your mocks.
21:08:31 efried o
21:08:38 efried only the ones you started with start()
21:08:40 efried brittle
21:08:42 mriedem that's fine for this test; anyway, thanks for making me think through it - now i can explain why i'm using stopall when artom -1s my change
21:10:26 openstackgerrit Matt Riedemann proposed openstack/nova master: Set Instance AZ from Selection AZ during build reschedule https://review.opendev.org/686047
21:33:03 openstackgerrit Matt Riedemann proposed openstack/nova master: Set Instance AZ from Selection AZ during migrate reschedule https://review.opendev.org/686050
21:36:41 openstackgerrit Matt Riedemann proposed openstack/nova master: Set Instance AZ from Selection AZ during migrate reschedule https://review.opendev.org/686050
21:46:13 openstackgerrit Matt Riedemann proposed openstack/nova master: Update cells v2 up-call caveats doc https://review.opendev.org/686053
21:55:01 openstackgerrit Ghanshyam Mann proposed openstack/nova-specs master: Re-propose policy-defaults-refresh spec for Ussuri https://review.opendev.org/686058
22:08:19 openstackgerrit melanie witt proposed openstack/nova stable/stein: Reduce scope of 'path' query parameter to noVNC consoles https://review.opendev.org/686063
22:14:03 efried melwitt: did you miss the train cherry-pick?
22:14:19 efried or did I?
22:15:06 melwitt efried: no, I missed it. sorry, I forgot the branch was cut
22:15:11 melwitt will fix
22:15:40 efried at least it's clean :P
22:16:57 openstackgerrit melanie witt proposed openstack/nova stable/train: Reduce scope of 'path' query parameter to noVNC consoles https://review.opendev.org/686066
22:17:50 openstackgerrit melanie witt proposed openstack/nova stable/stein: Reduce scope of 'path' query parameter to noVNC consoles https://review.opendev.org/686063
22:22:01 openstackgerrit melanie witt proposed openstack/nova stable/rocky: Reduce scope of 'path' query parameter to noVNC consoles https://review.opendev.org/686067
23:36:06 openstackgerrit Merged openstack/nova stable/stein: Make nova.compute.rpcapi.ComputeAPI.router a singleton https://review.opendev.org/684405
23:36:12 openstackgerrit Merged openstack/nova stable/stein: Func test for migrate reschedule with pinned compute rpc https://review.opendev.org/684406
23:36:17 openstackgerrit Merged openstack/nova stable/stein: Handle legacy request spec dict in ComputeTaskManager._cold_migrate https://review.opendev.org/684407
23:36:24 openstackgerrit Merged openstack/nova master: Reduce scope of 'path' query parameter to noVNC consoles https://review.opendev.org/685194
#openstack-nova - 2019-10-02
01:36:53 openstackgerrit Jing Zhang proposed openstack/nova master: Force small pages accounting for CPU pinned VMs so that memory can be reserved on NUMA 0 for host processes. https://review.opendev.org/686079
02:24:13 SonPham Hi, I want to commit my little project
02:24:54 SonPham It is a small module to snapshot VM and keep running process
02:25:12 SonPham what workflow i need to do?
04:40:15 SonPham Hi, I want to commit my little project
06:47:23 gibi sean-k-mooney: neutron creates child RPs under the compute RP and reports bandwidth inventory on the child RPs only. Also the resource allocation is done by nova as part of the instance (and migration) allocation. So the zun and the neturon case is pretty different
06:48:10 SonPham Hi, I want to commit my little project
07:02:42 gibi SonPham: hi! most of the core developers are active after 12:00 UTC. The contribution process is documented here https://docs.openstack.org/nova/latest/contributor/index.html
07:09:52 gibi SonPham: regarding your feature. How does it different from the existing snapshot feature of OpenStack?
07:10:50 SonPham gibi i'm using libvirt for snapshot
07:11:17 SonPham it can keep running process
07:11:59 SonPham but volume image type must be change from RAW to QCOW2
07:15:00 gibi SonPham: I think nova can also do live snapshot
07:17:26 gibi SonPham: the relevant codepath starts around here https://github.com/openstack/nova/blob/bf37bec80baa527ac013dfaa7480ef2761ed2cb9/nova/virt/libvirt/driver.py#L2267
07:19:16 openstackgerrit OpenStack Proposal Bot proposed openstack/nova stable/train: Imported Translations from Zanata https://review.opendev.org/685534
07:30:17 SonPham gibi i think openstakc snapshot is an image of disk. not keep process
07:31:59 gibi SonPham: yes, it creates a copy of the disk while the libvirt domain is still running
07:33:52 gibi SonPham: you what I mean is that the existing snapshot code does not require to shutdown the domain
07:33:59 gibi s/you//
07:34:45 gibi SonPham: do you mean that in your case the snapshot itself stores the running state of the whole domain?
07:35:12 SonPham gibi my project create snapshot when domain running. but when you revert snapshot. all of process still keep lock like you snapshot
07:37:05 SonPham gibi yep . stores the running state of the whole domain

Earlier   Later