| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-10-10 | |||
| 16:11:29 | mriedem | 1. seems we could workaround the instance uuid lock causing MessagingTimeout by using the long_rpc_timeout on that rpc call, | |
| 16:11:46 | mriedem | 2. your cleanup in the api is best effort - might not catch all cases but it's worth a try, yeah? | |
| 16:12:02 | lyarwood | mriedem: yup sorry just working my way towards this | |
| 16:12:21 | lyarwood | mriedem: 1. nice idea, yeah that would be useful, 2. yes it's just best effort | |
| 16:12:29 | mriedem | 3. long-term we should create the bdm in the api, pass it down to compute, and then delete in api if we hit a failure | |
| 16:12:45 | mriedem | i.e. implement vishy's todo from like 2012 | |
| 16:12:54 | lyarwood | mriedem: yeah just read that comment, yup that would be much much cleaner | |
| 16:12:58 | mriedem | but 3 is not backportable | |
| 16:13:02 | lyarwood | mriedem: both something to do on top of this | |
| 16:13:09 | mriedem | and with that i'm away! | |
| 16:13:10 | lyarwood | mriedem: as I'd like to backport this best effort approach | |
| 16:13:16 | lyarwood | enjoy | |
| 16:13:19 | mriedem_lunch | <3 | |
| 16:16:49 | mloza | dansmith: I forgot about that. I updated the cell mappings. Thanks | |
| 16:39:18 | openstackgerrit | Stephen Finucane proposed openstack/nova master: WIP: Stop testing Python 2 https://review.opendev.org/687954 | |
| 17:06:26 | openstackgerrit | sean mooney proposed openstack/nova master: [WIP] block rebuild when numa toplogy changed https://review.opendev.org/687957 | |
| 17:54:32 | openstackgerrit | Lee Yarwood proposed openstack/nova master: Helper to start computes with different HostInfos https://review.opendev.org/686832 | |
| 17:54:34 | openstackgerrit | Lee Yarwood proposed openstack/nova master: Add regression test for bug #1843639 https://review.opendev.org/682621 | |
| 17:54:34 | openstack | bug 1843639 in OpenStack Compute (nova) "libvirt: post_live_migration failures to disconnect volumes result in the rollback of live migrations" [Medium,In progress] https://launchpad.net/bugs/1843639 - Assigned to Lee Yarwood (lyarwood) | |
| 17:54:34 | openstackgerrit | Lee Yarwood proposed openstack/nova master: libvirt: Ignore volume exceptions during post_live_migration https://review.opendev.org/682622 | |
| 17:56:10 | lyarwood | artom: ^ my bad, I didn't mean to rebase your change there, sorry. | |
| 18:00:19 | efried | mriedem, dansmith, et al. I need to take the rest of the day off, and probably tomorrow too. Can someone run the meeting this afternoon? | |
| 18:00:23 | mriedem | efried: a question in https://review.opendev.org/#/c/687416/5 | |
| 18:00:37 | mriedem | efried: i have to leave early during the meeting (parent/teacher conferences) | |
| 18:00:44 | mriedem | maybe melwitt can run it? | |
| 18:01:04 | efried | The p.m. ones have been really sparse lately anyway, maybe just cancel and pick up next week. | |
| 18:01:21 | dansmith | efried: I'mma be gone today by the time of the meeting | |
| 18:01:26 | mriedem | there is quite a few things in open discussion, | |
| 18:01:35 | mriedem | but if we don't have a quorum then i'm not sure it's worth having... | |
| 18:01:40 | efried | right, exactly | |
| 18:01:45 | efried | I wanted to be present to discuss those things. | |
| 18:01:47 | dansmith | efried: fwiw, the early meeting is pretty much not doable for people in PDT | |
| 18:01:53 | efried | mm | |
| 18:02:07 | mriedem | ML or bust i guess | |
| 18:02:21 | efried | yeah, I'll cancel via ML. | |
| 18:06:42 | mriedem | dansmith: https://review.opendev.org/#/c/687139/ is failing a unit test | |
| 18:06:45 | mriedem | https://storage.gra1.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_543/687139/7/check/openstack-tox-py36/5432a3a/testr_results.html.gz | |
| 18:06:55 | dansmith | not possible | |
| 18:07:16 | dansmith | ah g'dang | |
| 18:09:05 | openstackgerrit | Dan Smith proposed openstack/nova master: Add cache_images() to conductor https://review.opendev.org/687139 | |
| 18:09:06 | openstackgerrit | Dan Smith proposed openstack/nova master: Add image caching API for aggregates https://review.opendev.org/687140 | |
| 18:09:07 | openstackgerrit | Dan Smith proposed openstack/nova master: WIP: Add image precaching docs for aggregates https://review.opendev.org/687348 | |
| 18:16:22 | efried | mriedem: responded about _use_project_id | |
| 18:16:37 | efried | and with that, I'm outta here. You know how to reach me if you need. | |
| 18:21:47 | mriedem | fried-signal? | |
| 18:30:53 | colby_ | Hey All. Im looking to isolate a group of projects to a set of hypervisors. We are running Rocky. So looking at tenant isolation with placement it sound like it will put all the tenants setup onto those host aggregates. Will this also prevent other tenants from being placed on the aggregate? | |
| 18:32:53 | dansmith | colby_: https://docs.openstack.org/nova/latest/configuration/config.html#scheduler.placement_aggregate_required_for_tenants | |
| 18:33:43 | dansmith | colby_: you want isolated aggregates I guess, which was done in train | |
| 18:33:59 | dansmith | but the above will help make sure you have every tenant in an aggregate so you can approximate it | |
| 18:34:38 | mriedem | these are the docs for the thing added in train https://docs.openstack.org/nova/latest/admin/configuration/schedulers.html#isolate-aggregates | |
| 18:34:43 | colby_ | so would I need to put all other tenants on a "general" aggregate that includes all other hosts then? | |
| 18:34:53 | dansmith | yeah | |
| 18:35:34 | colby_ | so then that would need to be modified with every new tenant added to the system? | |
| 18:35:43 | dansmith | yup | |
| 18:36:02 | dansmith | the "require" conf knob helps prevent new tenants from being scheduled until they're mentioned in _some_ aggregate | |
| 18:36:09 | dansmith | but..it's not automatic like isolated aggregates | |
| 18:36:24 | colby_ | ok | |
| 18:37:14 | colby_ | final question when I set the property of filter_tenant_idXXXXX for a tenant how can I remove it if I move them to another aggregate? | |
| 18:37:41 | dansmith | how? | |
| 18:37:55 | colby_ | yea is there a delete property? | |
| 18:38:04 | dansmith | https://docs.openstack.org/nova/latest/admin/aggregates.html | |
| 18:38:12 | dansmith | aggregate unset is what you're looking for presumably | |
| 18:38:58 | colby_ | ah unset thanks | |
| 18:39:11 | dansmith | colby_: please let me also point you at $topic. I wrote the placement agg stuff and reviewed the isolated hosts stuff you were asking about so I was interested but in general this channel is for development | |
| 18:42:01 | colby_ | Thanks. Yes I have not had great luck in getting answers in the general channel usually. I appreciate you help. | |
| 18:48:54 | artom | lyarwood, no worries | |
| 18:51:58 | openstackgerrit | Akihiro Motoki proposed openstack/python-novaclient master: PDF documentation build https://review.opendev.org/687966 | |
| 19:00:49 | colby_ | dansmith: Can I ask one final question? Is there any limits on the suffix to filter_tenant_id? (limited characters number or type?) | |
| 19:01:52 | dansmith | colby_: you're asking me a development question of where the code is that does this query right? | |
| 19:02:08 | colby_ | yea so I can see the checks its doing? | |
| 19:02:23 | dansmith | sure, since that's a development question let me find it | |
| 19:02:26 | dansmith | colby_: https://github.com/openstack/nova/blob/master/nova/scheduler/request_filter.py#L112 | |
| 19:02:34 | colby_ | thanks again | |
| 19:24:44 | mriedem | that doesn't answer the question :) | |
| 19:24:45 | mriedem | https://docs.openstack.org/api-ref/compute/?expanded=create-or-update-aggregate-metadata-detail | |
| 19:24:51 | mriedem | "The maximum size for each metadata key and value pair is 255 bytes." | |
| 19:26:46 | dansmith | mriedem: yeah, true, I was kinda thinking he was asking if he had to use numbers or could use letters | |
| 19:35:18 | mnaser | wild idea: nova-api rejects bfv requests if user doesnt have enough quota instead of it failing at nova-compute | |
| 19:35:34 | mnaser | feels like a logistical nightmare tho | |
| 19:35:41 | mnaser | nova's quota is hard enough as is | |
| 19:50:04 | mriedem | that's not so wild | |
| 19:50:33 | mriedem | pretty sure when the auto-disable compute snafu thing was riling everyone up over bfv over quota failures we talked about moving the volume create to api or conductor to avoid that silliness late in the compute | |
| 19:50:37 | mriedem | the api checks port quota but not volume | |
| 19:51:00 | mriedem | it would also help with some other stuff like az | |
| 19:51:23 | mriedem | i.e. if cross_az_attach=False, creating the volume earlier will tell us if we can't create a volume in a given az | |
| 19:52:00 | mriedem | mnaser: also note that (semi-recently) check cinder quota stuff during volume-backed snapshot | |
| 19:52:34 | mriedem | https://github.com/openstack/nova/blob/master/nova/compute/api.py#L3105 | |
| 19:54:04 | mriedem | doing that during bfv in the api or conductor seems pretty trivial | |
| 19:54:26 | mriedem | we already know (a) how many instances we're going to create and (b) how many volumes we're going to create for those instances | |
| 19:54:52 | mriedem | it'd be racy but so is the port quota check | |
| 20:05:52 | mnaser | mriedem: yeah and honestly i think its better to fail at the api layer than go through X reschedules.. | |
| 20:06:23 | mriedem | fun fact, | |
| 20:06:34 | mriedem | we don't reschedule for prep_block_device failures in the compute | |
| 20:06:38 | mriedem | you're welcome | |
| 20:07:17 | mriedem | https://bugs.launchpad.net/nova/+bug/1488111 | |
| 20:07:17 | openstack | Launchpad bug 1488111 in OpenStack Compute (nova) "Boot from volumes that fail in initialize_connection are not rescheduled" [Wishlist,Confirmed] | |
| 20:08:53 | mriedem | dansmith: https://review.opendev.org/#/c/687140/ is failing a few jobs (functional, api-ref and pep8) - did you want me to wait on re-reviewing that? | |
| 20:09:25 | dansmith | mriedem: ah yeah I didn't run a full slammajamma on that one, so yeah I'll hit that up in the mo'nan | |
| 20:09:42 | mriedem | ok i've wip'ed it | |
| 20:09:47 | mriedem | wip'ed it good | |