| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-08-20 | |||
| 14:38:03 | dansmith | but we need to decide how much we want to implement in the dark | |
| 14:38:21 | dansmith | if we go with your idea for strategy, I guess we could just implement one or two and then expand to more | |
| 14:38:45 | mriedem | if we did something like that, | |
| 14:38:52 | mriedem | i'd do distributed and fixed | |
| 14:38:58 | mriedem | where fixed is configurable batch size | |
| 14:39:00 | dansmith | I guess I can also see someone wanting "weighted" where we count the number of instance mappings in each cell and pull more from the heavier cells | |
| 14:39:13 | dansmith | yeah | |
| 14:39:49 | mriedem | definitely hard to say, especially without running each through some perf testing like Kevin_Zheng and yikun are doing | |
| 14:39:54 | dansmith | yup | |
| 14:39:59 | efried | batch, please | |
| 14:40:38 | mriedem | dansmith: have you chatted with tssurya about this at all? | |
| 14:40:44 | dansmith | mriedem: I would think distributed and fixed are obvious things we need so we should be able to go with those regardless and then test out some of the other options | |
| 14:40:48 | dansmith | mriedem: nope | |
| 14:40:48 | mriedem | since cern maps projects to cells, i wonder if this doesn't really affect them | |
| 14:40:55 | dansmith | yeah I expect it doesn't | |
| 14:41:18 | dansmith | there may be some projects in multiple cells, but it doesn't really affect you until you get to lots of cells per project | |
| 14:41:22 | mriedem | they might notice a slight difference (<=1s) if a project has instances in >1 cell | |
| 14:41:24 | dansmith | and they might cut down max_limit too | |
| 14:41:28 | dansmith | yep | |
| 14:41:53 | mriedem | yeah let's do distributed and fixed, and default to distributed | |
| 14:42:30 | dansmith | ack | |
| 14:42:58 | dansmith | I'm pleasantly surprised that this worked out as easy as it did, because I was really dreading having to implement this | |
| 14:44:52 | mriedem | i'm pleasantly surprised you already did it.... | |
| 14:44:57 | mriedem | b/c i didn't want to | |
| 14:48:14 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge extended availability zone response into server controller https://review.openstack.org/502859 | |
| 14:48:15 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge config drive extension response into server controller https://review.openstack.org/584223 | |
| 14:48:16 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge extended server attributes extension response https://review.openstack.org/584590 | |
| 14:48:17 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge keypair extension response into server view builder https://review.openstack.org/584748 | |
| 14:48:18 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge server usage extension response into server view builder https://review.openstack.org/585262 | |
| 14:48:19 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge security groups extension response into server view builder https://review.openstack.org/585475 | |
| 14:48:20 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Merge extended_status extension response into server view builder https://review.openstack.org/592092 | |
| 14:57:11 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Remove ChanceScheduler https://review.openstack.org/576384 | |
| 15:13:03 | tssurya | dansmith, mriedem: yea it wouldn't affect us much | |
| 15:13:36 | mriedem | vdrok: is this also broken for rocky? https://review.openstack.org/#/c/593628/ | |
| 15:13:41 | mriedem | in that we might need an rc3? | |
| 15:14:25 | mriedem | tssurya: it should in the --all-tenants case | |
| 15:14:40 | mriedem | but maybe you don't care about performance of that | |
| 15:14:58 | tssurya | mriedem: yea we usually use some other filters like project or name before that | |
| 15:15:50 | dansmith | tssurya: cool, I figured as much | |
| 15:16:41 | vdrok | mriedem: yup I think so :( | |
| 15:16:56 | mriedem | http://logstash.openstack.org/#dashboard/file/logstash.json?query=message%3A%5C%22Invalid%20inventory%20for%20'CUSTOM_BAREMETAL'%20on%20resource%20provider%5C%22%20AND%20message%3A%5C%22The%20reserved%20value%20is%20greater%20than%20or%20equal%20to%20total%5C%22&from=7d | |
| 15:17:05 | mriedem | melwitt: likely have another bug fix for rc2 | |
| 15:19:22 | dansmith | mriedem: she was looking at that on friday, IIRC | |
| 15:19:42 | mriedem | ok rc2 hasn't merged yet, so i -1ed the release request | |
| 15:22:38 | cdent | yay! to ChanceScheduler going by by | |
| 15:24:13 | stewie925 | sorry guys, I am looking for an openstack channel to bring up an openstack-gerrit issue? | |
| 15:27:46 | jroll | stewie925: #openstack-infra would be best | |
| 15:29:32 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Use placement microversion 1.26 in update_from_provider_tree https://review.openstack.org/593628 | |
| 15:33:11 | mriedem | dansmith: if you want i can report a perf bug for the batch size instance list stuff | |
| 15:33:20 | dansmith | mriedem: I want | |
| 15:33:34 | mriedem | you want it you got it! | |
| 15:33:43 | mriedem | the hottest band in the land, KISS! | |
| 15:40:00 | openstack | Launchpad bug 1787977 in OpenStack Compute (nova) "Inefficient multi-cell instance list" [Medium,Triaged] - Assigned to Dan Smith (danms) | |
| 15:40:00 | mriedem | dansmith: https://bugs.launchpad.net/nova/+bug/1787977 | |
| 15:40:03 | dansmith | thanks | |
| 15:52:41 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/rocky: Use placement microversion 1.26 in update_from_provider_tree https://review.openstack.org/593678 | |
| 15:52:43 | mriedem | dansmith: for rc2 ^ | |
| 16:25:09 | openstackgerrit | Merged openstack/nova master: Remove mox in libvirt/test_driver.py (5) https://review.openstack.org/570842 | |
| 16:33:40 | johnsom | Hi nova folks. I'm getting failures with stable/rocky devstack that is blocking some gates. Would someone have a minute to take a look? http://logs.openstack.org/55/593455/1/check/neutron-lbaasv2-dsvm-api/33e93f3/logs/devstacklog.txt.gz#_2018-08-20_16_04_21_951 | |
| 16:34:04 | johnsom | And http://logs.openstack.org/55/593455/1/check/neutron-lbaasv2-dsvm-api/33e93f3/logs/screen-n-cpu.txt.gz#_Aug_20_16_02_14_425974 | |
| 16:35:52 | johnsom | The later is n-cpu exiting with a critical error | |
| 16:46:54 | dansmith | johnsom: it can't talk to conductor | |
| 16:47:52 | melwitt | . | |
| 16:48:05 | johnsom | I have been digging through logs for a while. It seems to be repeatable across gate hosts. | |
| 16:55:34 | johnsom | It looks like compute service list shows two nova-conductors in the up status. Any pointers where else I should look in the logs? There aren't any errors I see in the n-cond log file | |
| 16:59:59 | openstackgerrit | Konstantinos Samaras-Tsakiris proposed openstack/nova-specs master: Placement model for passthrough devices https://review.openstack.org/591037 | |
| 17:06:24 | kosamara | efried sean-k-mooney: ^^ | |
| 17:06:31 | efried | ack | |
| 17:07:08 | sean-k-mooney | kosamara: currently reivew an sriov bond spec but ill take a look at that next | |
| 17:42:16 | openstackgerrit | Dan Smith proposed openstack/nova master: Make instance_list perform per-cell batching https://review.openstack.org/593131 | |
| 17:42:17 | openstackgerrit | Dan Smith proposed openstack/nova master: List instances from all cells explicitly https://review.openstack.org/593717 | |
| 18:50:58 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Fix soft deleting vm fails after "nova resize" vm https://review.openstack.org/546920 | |
| 19:08:01 | melwitt | mriedem: ack on the new patch needed for rc2 | |
| 19:09:33 | openstackgerrit | Merged openstack/nova master: Merge extended availability zone response into server controller https://review.openstack.org/502859 | |
| 19:09:42 | melwitt | problem was reported on friday, when I dug into the logs, I saw that version 1.26 wasn't being sent for the call to allow reserved == total, but didn't realize on the original patch the change missed another call site that needed 1.26 | |
| 19:10:18 | melwitt | glad the fix was found | |
| 19:23:54 | gryf | hi. there is a code in review for some time now. can I get some attention from core reviewers? https://review.openstack.org/#/c/536351 | |
| 19:27:08 | efried | gryf: I think johnthetubaguy ought to have another look at ^ and see if his concerns have been addressed. Other cores will be reluctant to approve the patch with another core's -1 on it. | |
| 19:27:50 | melwitt | dansmith: could you take a look at the proposed fix? ^ it seems to make sense but I'm not that familiar with the particulars of resource reporting considering overcommit | |
| 19:28:22 | dansmith | melwitt: oh yeah I looked at this last week but never said anything | |
| 19:28:35 | dansmith | melwitt: tbh, I'm not sure why we should change that at the moment | |
| 19:28:54 | dansmith | it's just reporting a different value for something we don't look at during scheduling anyway | |
| 19:30:34 | melwitt | they're saying it's being rejected at the check_can_live_migrate_source stage, not related to scheduling | |
| 19:31:02 | melwitt | the disk value is being rejected | |
| 19:32:12 | dansmith | yeah, but neither value is totally right I don't think | |
| 19:33:00 | gryf | current value is totally wrong | |
| 19:33:55 | gryf | instead of remaining space, entire disk size is provided :/ | |
| 19:34:09 | dansmith | okay well, this isn't really related to resource reporting and I'm in the middle of something else at the moment, | |
| 19:34:22 | dansmith | but clearly I'm not familiar enough with that to review it without a deep dive, so.. :) | |
| 19:34:22 | gryf | dansmith, no rush | |
| 19:34:32 | gryf | oh :C | |
| 19:35:49 | gryf | either way, we already have that code applied, since we bump on this | |
| 19:36:15 | gryf | I thought that maybe it is worth to have it fixed on stable | |
| 19:37:57 | mriedem | anyone know if big john is going to be in denver? | |
| 19:44:06 | openstackgerrit | Merged openstack/nova master: Use placement microversion 1.26 in update_from_provider_tree https://review.openstack.org/593628 | |
| 19:51:35 | mriedem | melwitt: if you're ok with https://review.openstack.org/#/c/593678/ for stable/rocky, you could +1 and then we can find another stable maint team core to approve to move rc2 along | |
| 19:52:03 | melwitt | ok | |
| 20:20:50 | sean-k-mooney | mriedem: FYI my summary of the live migration testing i did last week and a half is here http://lists.openstack.org/pipermail/openstack-dev/2018-August/133546.html | |
| 20:21:16 | sean-k-mooney | i didnt mean to add [all] but that does not matter now | |
| 20:22:00 | sean-k-mooney | i have opened 2 bugs in nova and 3 in neutron | |