opendevreview | Hironori Shiina proposed openstack/cinder master: Prevent temporary volume from being deleted accidentally https://review.opendev.org/c/openstack/cinder/+/826949 | 01:25 |
---|---|---|
opendevreview | OpenStack Proposal Bot proposed openstack/cinder master: Imported Translations from Zanata https://review.opendev.org/c/openstack/cinder/+/837443 | 02:16 |
opendevreview | Raghavendra Tilay proposed openstack/cinder master: HPE 3PAR: In multi host env, fix multi-detach operation https://review.opendev.org/c/openstack/cinder/+/824911 | 05:55 |
opendevreview | Raghavendra Tilay proposed openstack/cinder master: HPE 3PAR: In multi host env, fix multi-detach operation https://review.opendev.org/c/openstack/cinder/+/824911 | 09:20 |
opendevreview | Mounika Sreeram proposed openstack/cinder master: [SVf] Delete/Extend issue in reverse replication https://review.opendev.org/c/openstack/cinder/+/831776 | 10:38 |
opendevreview | Harsh Ailani proposed openstack/cinder master: [SVf]:Fix multiple lsvdisk calls for GMCV create volume operation https://review.opendev.org/c/openstack/cinder/+/829110 | 10:54 |
*** bhagyashris_ is now known as bhagyashris | 11:55 | |
opendevreview | sean mooney proposed openstack/os-brick stable/wallaby: Revert "Use file locks in connectors" https://review.opendev.org/c/openstack/os-brick/+/838871 | 13:57 |
sean-k-mooney | rosmaita: fyi https://bugs.launchpad.net/os-brick/+bug/1969794 https://review.opendev.org/c/openstack/os-brick/+/838871 | 14:41 |
rosmaita | sean-k-mooney: tu | 14:41 |
rosmaita | i mean, ty | 14:41 |
sean-k-mooney | you might change your mind when you read it | 14:41 |
sean-k-mooney | the revert is more to demonstrate the issue | 14:42 |
sean-k-mooney | i think the fix should be to set a default | 14:42 |
sean-k-mooney | we could also perhaps modify nova's unit test but that wont adress the impact to real deploymetns | 14:42 |
rosmaita | i'm not sure there's a sensible default for that, and the installation instructions have always said you need to create a lock_path | 14:42 |
sean-k-mooney | well the code did not require it until that backport | 14:43 |
sean-k-mooney | it was only needed if you used external locks | 14:43 |
rosmaita | yes, but we did use them in some places prior to that patch | 14:43 |
rosmaita | after that patch, we use them exclusively | 14:44 |
sean-k-mooney | yep so the quetion is is there any case where you could use os-brick where they were not used at all before | 14:44 |
sean-k-mooney | for example say ceph | 14:44 |
sean-k-mooney | when not host monthing | 14:44 |
sean-k-mooney | well actully no for any of the connectors where they are excluivly used | 14:45 |
sean-k-mooney | was it possibel to run without it | 14:45 |
rosmaita | not sure, because i think for both nova and cinder, installation docs say to set it | 14:45 |
sean-k-mooney | if so then this is a breaking change in lib behaivor | 14:45 |
sean-k-mooney | i dont think nova has docs for this | 14:46 |
rosmaita | sean-k-mooney: it's in here: https://docs.openstack.org/nova/latest/install/controller-install-ubuntu.html#install-and-configure-components | 14:47 |
rosmaita | "In the [oslo_concurrency] section, configure the lock path:" | 14:47 |
sean-k-mooney | that on the contoler | 14:47 |
sean-k-mooney | what about the compute node | 14:47 |
sean-k-mooney | and that is for novas external locks | 14:48 |
sean-k-mooney | not os-brick | 14:48 |
rosmaita | it's in here, too | 14:48 |
rosmaita | https://docs.openstack.org/nova/latest/install/compute-install-ubuntu.html#install-and-configure-components | 14:48 |
rosmaita | os-brick doesn't have a config file, it's configured in the service configs | 14:49 |
rosmaita | (at least i'm pretty sure) | 14:49 |
sean-k-mooney | ya i guess that is true | 14:49 |
sean-k-mooney | well it has option which are loaded form the set of config parsed by the parent process | 14:49 |
rosmaita | right | 14:50 |
sean-k-mooney | you can simplete start nova-compute with --config /etc/nova/nova.conf --config /etc/nova/os-brick.conf | 14:50 |
sean-k-mooney | if you want too | 14:50 |
sean-k-mooney | os-brick is just looking at the global config object | 14:50 |
rosmaita | sean-k-mooney: (i am in 2 meetings right now) | 14:53 |
sean-k-mooney | rosmaita: its not urgent we just have had reports of unit test failing on and off for a few months | 14:53 |
sean-k-mooney | and i finally tracked down why | 14:54 |
sean-k-mooney | so we need to just decided how to fix that | 14:54 |
sean-k-mooney | we can fix it in nova or find a why to make it work by default in os-brick | 14:54 |
sean-k-mooney | just wanted to let ye know | 14:54 |
rosmaita | sean-k-mooney: we discussed this on the release patch: https://review.opendev.org/c/openstack/releases/+/829590 | 14:57 |
rosmaita | i think in real deployments, it will already be set | 14:57 |
rosmaita | but thanks for filing the bug, we can reconsider whether we are handling this correctly | 14:58 |
sean-k-mooney | rosmaita: we will proably just mock this out with a fixture in our base test case | 15:16 |
sean-k-mooney | our unit test shoudl not be creating locks on the file system anyway | 15:16 |
rosmaita | sean-k-mooney: that sounds good ... we'll discuss the bug at next week's bug squad meeting, just to be sure | 15:16 |
opendevreview | Eric Harney proposed openstack/cinder master: WIP: RBD: flatten child volumes on delete https://review.opendev.org/c/openstack/cinder/+/835384 | 17:56 |
opendevreview | Merged openstack/cinder master: Imported Translations from Zanata https://review.opendev.org/c/openstack/cinder/+/837443 | 18:34 |
sean-k-mooney | rosmaita: https://review.opendev.org/c/openstack/nova/+/838942 should fix it on the nova side so we can just backport that to all affected branches | 19:17 |
opendevreview | Eric Harney proposed openstack/cinder master: WIP: RBD: flatten child volumes on delete https://review.opendev.org/c/openstack/cinder/+/835384 | 19:41 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!