| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-05-14 | |||
| 18:51:53 | sean-k-mooney | i mean we can always fix it when we integrate cyborg rather then now | |
| 18:59:13 | openstackgerrit | Rodrigo Barbieri proposed openstack/nova stable/rocky: Add functional confirm_migration_error test https://review.opendev.org/658834 | |
| 19:04:13 | openstackgerrit | Eric Fried proposed openstack/nova master: [Trivial doc change] Admin can overwrite the locked_reason of an owner https://review.opendev.org/659067 | |
| 19:46:20 | mriedem | dansmith: so i just realized that our 2.33 change for listing hypervisors which added limit and marker support for paging changed the previous behavior where we'd just return everything - we now default to CONF.api.max_limit if you don't specify a limit, and i can't tell if that's good or not - i suppose in terms of api load it's good to limit server-side | |
| 19:46:30 | mriedem | forces the client to page though | |
| 19:46:54 | dansmith | mriedem: you mean before that we would return all of them regardless of the amount? | |
| 19:46:58 | mriedem | right | |
| 19:46:59 | dansmith | like, 50000000000 of them? | |
| 19:47:09 | mriedem | 10K, 14K, whatever cern has yeah | |
| 19:47:19 | dansmith | clamping to max_limit does not seem like a regression to me :) | |
| 19:47:24 | mriedem | heh | |
| 19:47:58 | mriedem | yeah i'm just learning some watcher code where they build their data model for hypervisors and it's super inefficient, and lossy because of this limit thing i realize | |
| 19:48:43 | mriedem | tl;dr they (1) list all hypervisors with details (well, up to 1K), (2) for each hypervisor, get it's servers (3) for each hypervisor, get it's details (again) | |
| 19:48:54 | mriedem | all of that can be done in a single GET with 2.53 | |
| 19:49:08 | dansmith | yoza | |
| 19:49:11 | dansmith | *yowza | |
| 19:49:21 | mriedem | and i *think* this is on ever audit request | |
| 19:49:36 | mriedem | rather than like build on startup, then modify as notifications come in | |
| 19:49:42 | mriedem | *every | |
| 19:52:31 | dansmith | sweet | |
| 19:59:21 | mriedem | oy, and for every compute node, get all of its instances in separate GETs | |
| 20:03:32 | NewBruce | hey sean-k-mooney / mriedem | |
| 20:04:13 | NewBruce | saw the code updates and call for review - will get onto it asap - sorry bout that - got caught up in a couple other bug chasing efforts….. | |
| 20:04:24 | NewBruce | thanks for the work on it though :) | |
| 20:59:32 | mriedem | np | |
| 21:15:26 | mriedem | efried: comments on surya's fix here https://review.opendev.org/#/c/658110/ - if you're in agreement i'll just make the changes quick | |
| 21:18:00 | efried | mriedem: wfm. You want to fix or would you feel better if I did? | |
| 21:18:25 | mriedem | i'll do it, already got the test done | |
| 21:21:33 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Disable limit if affinity(anti)/same(different)host is requested https://review.opendev.org/658110 | |
| 21:30:03 | mriedem | efried: so i addressed my nits on ^ you and alex were +2, want to just fast approve? | |
| 21:32:15 | efried | mriedem: gimme a sec. I'll mark up and then fast approve. | |
| 21:36:29 | efried | mriedem: Comments added. If you are okay with those things, say so and I'll approve. | |
| 21:36:37 | efried | or quick respin to address would be better :) | |
| 21:38:21 | mriedem | oh hmm | |
| 21:38:24 | mriedem | multiple hints aye | |
| 21:38:28 | mriedem | couldn't i just throw a fake hint in there? | |
| 21:39:44 | mriedem | i.e. are you ok with this? | |
| 21:39:44 | mriedem | fake_spec = objects.RequestSpec( | |
| 21:39:44 | mriedem | flavor=flavor, scheduler_hints={hint: [uuids.fake], | |
| 21:39:44 | mriedem | 'some-fake-hint': ['fake-value']}) | |
| 21:40:20 | efried | That would be good too, but I'd like to see >1 of the hints we're conditioning on. | |
| 21:41:32 | efried | mriedem: If it were me, I would nix ddt, put the guts of the test in a _helper, and loop over it with various permutations of hints constructed in a loop from components. | |
| 21:42:03 | efried | mriedem: if you tire of this, I can take a crack at it. | |
| 21:43:55 | mriedem | ddt is nice in that the test names have the ddt values in the test name in case one of them blows up | |
| 21:43:58 | mriedem | anyway i've got a thing | |
| 21:44:14 | mriedem | this didn't come up in the review comments on the patch before this so i figured it was fine - unless that came up in irc | |
| 21:46:29 | mriedem | i'm not going to mention something about nested providers in the commit message b/c this is going back to queens | |
| 21:46:31 | mriedem | where those weren't supported | |
| 21:46:32 | mriedem | is that ok? | |
| 21:46:45 | efried | Yeah, ignore that one, just thought to point it out. | |
| 21:47:37 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Disable limit if affinity(anti)/same(different)host is requested https://review.opendev.org/658110 | |
| 21:49:44 | efried | mriedem: +A | |
| 21:49:52 | mriedem | whew | |
| 21:50:23 | efried | I know, try to do a nice thing... | |
| 22:48:30 | openstackgerrit | Eric Fried proposed openstack/nova master: WIP: TC Vision Reflection https://review.opendev.org/658932 | |
| 22:48:57 | efried | A little more every day | |
| 22:49:35 | efried | maybe by meeting time I'll have enough for folks to start ripping apart | |
| #openstack-nova - 2019-05-15 | |||
| 01:54:19 | openstackgerrit | Boxiang Zhu proposed openstack/nova master: Fix failure to boot instances with qcow2 format images https://review.opendev.org/640271 | |
| 02:05:45 | openstackgerrit | Merged openstack/nova master: Add zvm driver image type capabilities https://review.opendev.org/655732 | |
| 02:26:42 | openstackgerrit | Brin Zhang proposed openstack/nova master: Cap sphinx for py2 to match global requirements https://review.opendev.org/659201 | |
| 02:30:18 | openstackgerrit | Brin Zhang proposed openstack/python-novaclient master: Cap sphinx for py2 to match global requirements https://review.opendev.org/659202 | |
| 02:35:34 | openstackgerrit | Merged openstack/nova master: Disable limit if affinity(anti)/same(different)host is requested https://review.opendev.org/658110 | |
| 02:44:17 | openstackgerrit | Brin Zhang proposed openstack/nova-specs master: Cap sphinx for py2 to match global requirements https://review.opendev.org/659205 | |
| 02:46:51 | openstackgerrit | Tetsuro Nakamura proposed openstack/nova stable/stein: Skip _exclude_nested_providers() if not nested https://review.opendev.org/659206 | |
| 02:51:00 | openstackgerrit | Tetsuro Nakamura proposed openstack/nova stable/rocky: Skip _exclude_nested_providers() if not nested https://review.opendev.org/659207 | |
| 02:58:30 | openstackgerrit | Boxiang Zhu proposed openstack/nova master: Add host and hypervisor_hostname flag to create server https://review.opendev.org/645520 | |
| 03:43:36 | openstackgerrit | Brin Zhang proposed openstack/python-novaclient master: Cap sphinx for py2 to match global requirements https://review.opendev.org/659202 | |
| 07:06:15 | openstackgerrit | Brin Zhang proposed openstack/python-novaclient master: Cap sphinx for py2 to match global requirements https://review.opendev.org/659202 | |
| 08:08:36 | sean-k-mooney | bauzas: have you seen http://logs.openstack.org/97/652197/17/check/grenade-py3/7b3cac4/logs/grenade.sh.txt.gz#_2019-05-15_02_34_39_701 recently | |
| 08:08:52 | sean-k-mooney | im not sure if our grenade job is broken after the change to os-dev | |
| 08:08:59 | sean-k-mooney | *opendev | |
| 08:19:03 | sean-k-mooney | ya... 337 hits in the last 12 hours http://logstash.openstack.org/#/dashboard/file/logstash.json?query=message:%5C%22Could%20not%20install%20packages%20due%20to%20an%20EnvironmentError:%20HTTPSConnectionPool(host%3D'git.openstack.org',%20port%3D443)%5C%22 | |
| 08:23:46 | bauzas | sean-k-mooney: nope, not yet | |
| 08:25:22 | bauzas | sean-k-mooney: there is a redirect for https://git.openstack.org/cgit/openstack/requirements/plain/upper-constraints.txt | |
| 08:25:31 | sean-k-mooney | http://logstash.openstack.org/#/dashboard/file/logstash.json?query=message:%5C%22Could%20not%20install%20packages%20due%20to%20an%20EnvironmentError:%20HTTPSConnectionPool(host%3D'git.openstack.org',%20port%3D443)%5C%22 | |
| 08:25:35 | bauzas | sean-k-mooney: so we shouldn't get the exception | |
| 08:26:05 | sean-k-mooney | bauzas: we should not but there have been 337 filures since in the last week | |
| 08:26:13 | sean-k-mooney | so i think its not always working | |
| 08:26:20 | sean-k-mooney | i just mention this on the infra channel | |
| 08:26:36 | bauzas | ok | |
| 08:26:41 | sean-k-mooney | i might create an elastic recheck query and file a bug but im not sure where to file it | |
| 08:27:00 | bauzas | good question | |
| 08:27:19 | bauzas | we used to create bugs in Nova even if they were related to infra | |
| 08:27:28 | bauzas | and we marked them as critical | |
| 08:27:48 | bauzas | we also had a tag for them | |
| 08:28:13 | sean-k-mooney | ok i can do that. its affecting neutron jobs too but the grenade jobs seam to be the onse teh hit it most | |
| 08:28:22 | bauzas | sean-k-mooney: https://bugs.launchpad.net/nova/+bugs?field.tag=gate-failure | |
| 08:29:20 | sean-k-mooney | ah cool ill file one and add the gate-failure tag | |
| 08:48:38 | openstackgerrit | Surya Seetharaman proposed openstack/nova stable/stein: Disable limit if affinity(anti)/same(different)host is requested https://review.opendev.org/659239 | |
| 08:49:04 | sean-k-mooney | bauzas: should i set it to critical by the way or just high. i have hit it twice in a row but i dont think its failing in all builds its just annoying | |
| 08:49:18 | bauzas | sean-k-mooney: on 1:1, sorry | |
| 08:49:28 | sean-k-mooney | no worries | |
| 08:58:55 | openstackgerrit | Alex Xu proposed openstack/nova master: Remove the request_destination check in the scheduler https://review.opendev.org/659243 | |
| 08:59:57 | alex_xu | ^ efried, is it something we can removed? | |
| 09:02:53 | bhagyashris | efried, bauzas: Hi, I just want to discuss about http://logs.openstack.org/81/555081/28/check/openstack-tox-docs/6314f69/html/specs/train/approved/cpu-resources.html#configuration-options regarding the reserved_host_cpu calculation for PCPU resource, | |
| 09:03:02 | bhagyashris | f I have total 16 cpus and 4 are assigned to cpu_dedicated_set and 2 are assign to cpu_shared_set then in that case reserved_cpus for PCPU resource class will be (16- (4+2)) = 10 and for VCPU resource class it will be 0 . | |
| 09:03:09 | bhagyashris | Then in that case if I request instance which requires 1 PCPU then placement returns the 0 allocation_candiacted () (reason is it fails at _capacity_check_clause()) | |
| 09:03:58 | bhagyashris | Please correct me if I am wrong while calculating the reserved host cpu. | |
| 09:04:41 | kashyap | sean-k-mooney: dansmith: Yesterday you said: just like how you don't want Nova to not land instances based on "fan speed, or cpu temperature, or firmware patch levels..." you also don't want to stop launching instances on Compute hosts with vulnerable hypervisors. | |
| 09:04:43 | openstackgerrit | Surya Seetharaman proposed openstack/nova stable/rocky: Disable limit if affinity(anti)/same(different)host is requested https://review.opendev.org/659246 | |