| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-03-06 | |||
| 09:59:28 | alex_xu | johnthetubaguy: is that old behaviour? as my understand, the API always return 500 with invalid regex, so it should be never works in the before | |
| 10:00:03 | alex_xu | let me check the bug report, whether I missed something | |
| 10:00:35 | johnthetubaguy | alex_xu: what I mean is, the current behaviour is better than the old 500 | |
| 10:00:46 | johnthetubaguy | alex_xu: its just odd from a user perspective | |
| 10:00:52 | alex_xu | johnthetubaguy: ah, yea | |
| 10:01:34 | johnthetubaguy | alex_xu: they have a list of servers, they are using filtering by name to show some server details, but some of the queries return 400 (or 500) instead of the server | |
| 10:02:29 | johnthetubaguy | defaulting to exact match when it was an invalid regex seemed quite nice, although I guess its hard to debug your regex that way | |
| 10:03:15 | alex_xu | johnthetubaguy: yea, and that sounds like API behaviour change, doesn't sounds like we can change API without microversion | |
| 10:04:52 | johnthetubaguy | if it were 500 -> results like when this was first proposed, it wouldn't be, which is the odd thing I guess | |
| 10:05:50 | johnthetubaguy | if it were name_regex= I wouldn't worry about this either I guess | |
| 10:06:29 | bauzas | lyarwood: AFAIR, nova unittests can run behind a mysql driver backend instead of sqlite | |
| 10:06:50 | bauzas | given the Database fixture does use the default oslo.db values for creating the DB | |
| 10:07:07 | mdbooth_ | bauzas: I did some work on this a couple of years ago and there were a few wrinkles but mostly it just worked | |
| 10:07:15 | mdbooth_ | For unit tests, that is | |
| 10:07:16 | bauzas | for functional tests, we need to make sure it does use the same path | |
| 10:07:38 | lyarwood | bauzas: yeah I think you need to actually configure mysql/mariadb to use the default creds for it to work right? | |
| 10:07:56 | bauzas | yeah | |
| 10:07:57 | mdbooth_ | IIRC, the biggest wrinkle at the time was that sqlite (used by default) allowed millisecond accuracy in timestamps, whereas mysql only does 1 second | |
| 10:08:15 | mdbooth_ | This broke a bunch of unit tests which wrote non-rounded seconds into the db and expected to read the same value back. | |
| 10:08:42 | bauzas | here, I suppose lyarwood wanna test some very specific ones | |
| 10:08:51 | bauzas | not running the whole job | |
| 10:09:09 | lyarwood | bauzas: well downstream we hit a func test failure caused by the sqlite version we are shipping in RHEL | |
| 10:09:15 | alex_xu | johnthetubaguy: yes, agree that is odd | |
| 10:09:24 | mdbooth | Although running the whole job isn't a bad idea. There are a bunch of very interesting semantic differences. | |
| 10:09:26 | lyarwood | bauzas: as we don't support sqlite we need to switch these jobs over to mariadb | |
| 10:09:28 | alex_xu | johnthetubaguy: just check the history, we fix the 500 at https://review.openstack.org/#/c/344945/ | |
| 10:09:49 | bauzas | lyarwood: yeah I heard that from migi | |
| 10:09:58 | johnthetubaguy | alex_xu: yeah, I remember worrying about that bug fix we had pending when approving that one | |
| 10:10:34 | bauzas | lyarwood: so you would run a specific functional tests job backed by mariadb, right? | |
| 10:10:40 | bauzas | not all the UTs ? | |
| 10:11:54 | lyarwood | bauzas: well, all of the func tests that we can using mariadb | |
| 10:12:04 | lyarwood | bauzas: and any UTs that are using the DB as well tbh | |
| 10:12:31 | bauzas | okay, so fixing upstream the wrinkles that mdbooth said would then a good idea | |
| 10:12:53 | bauzas | at least, the first step is to configure the jobs and see what are the results | |
| 10:13:08 | bauzas | I'd only consider functional tests at a first step tho | |
| 10:13:15 | alex_xu | johnthetubaguy: so probably we can introduce "/servers?name=in:VM1,VM2" in the future | |
| 10:13:24 | alex_xu | for the exact match | |
| 10:13:36 | bauzas | alex_xu: I had a patch up for the in: tag a while ago | |
| 10:13:49 | alex_xu | bauzas: yea, I learned from that | |
| 10:14:07 | johnthetubaguy | alex_xu: oh... the in operator, thats a good way around that one | |
| 10:14:16 | bauzas | alex_xu: disclaimar, I haven't read the above discussion, so no context here yet | |
| 10:14:25 | bauzas | disclaimer* even | |
| 10:14:59 | johnthetubaguy | alex_xu: so I -2ed that patch, I can catch up with sneti on why I did that later on | |
| 10:15:38 | alex_xu | johnthetubaguy: ok, if sneti happy with that, we better to have a patch to tell the user how to match exact name in the current API behaviour | |
| 10:15:57 | alex_xu | in api-ref | |
| 10:16:22 | johnthetubaguy | alex_xu: yeah, I can't remember what we say in there right now | |
| 10:16:53 | johnthetubaguy | "Filters the response by a server name, as a string. You can use regular expressions in the query. For example, the ?name=bob regular expression returns both bob and bobb. If you must match on only bob, you can use a regular expression that matches the syntax of the underlying database server that is implemented for Compute, such as MySQL or PostgreSQL." | |
| 10:16:57 | johnthetubaguy | thats probably OK | |
| 10:17:55 | alex_xu | oops, that depend on the underlying database, that is bad :( | |
| 10:20:10 | alex_xu | and we validate the regex with python first...https://github.com/openstack/nova/blob/master/nova/api/validation/validators.py#L40 | |
| 10:35:59 | openstackgerrit | sahid proposed openstack/nova master: numa: add numa constraints for emulator threads policy https://review.openstack.org/385023 | |
| 10:36:01 | openstackgerrit | sahid proposed openstack/nova master: libvirt: return a CPU overhead if isolate emulator threads requested https://review.openstack.org/385364 | |
| 10:36:02 | openstackgerrit | sahid proposed openstack/nova master: numa: fit instance topology with cpus reserved onto cores https://review.openstack.org/400875 | |
| 10:36:04 | openstackgerrit | sahid proposed openstack/nova master: virt: update overhead to take into account vCPUs https://review.openstack.org/385024 | |
| 10:36:07 | openstackgerrit | sahid proposed openstack/nova master: libvirt: make emulator threads to run on the reserved pCPU https://review.openstack.org/400878 | |
| 10:36:08 | openstackgerrit | sahid proposed openstack/nova master: numa: update numa usage incording the cpu reserved https://review.openstack.org/400877 | |
| 10:36:08 | openstackgerrit | sahid proposed openstack/nova master: numa: take into account cpus reserved https://review.openstack.org/400876 | |
| 10:36:48 | openstackgerrit | Sivasathurappan Radhakrishnan proposed openstack/nova master: Port binding based on events during live migration https://review.openstack.org/434870 | |
| 10:49:53 | johnthetubaguy | alex_xu: been meaning to reach out to you about things a bit related to what is written down here: https://review.openstack.org/#/c/440917/9/doc/contributor-guide/source/non-native-english-speakers.rst | |
| 10:50:26 | johnthetubaguy | alex_xu: looking at creating some place where folks can share good ideas on how to work with upstream when you are a non-native-english-speaker | |
| 10:50:41 | johnthetubaguy | alex_xu: would love your help with that, if you are keen? | |
| 10:58:59 | openstack | Launchpad bug 1670319 in OpenStack Compute (nova) "Reschedule of failed instance doesn't happening when scheduler placed two instances to the same ironic node" [Undecided,New] | |
| 10:58:59 | vsaienk01 | jroll, dtantsur, mriedem fyi found issue when reschedule won't happening for failed instance https://bugs.launchpad.net/nova/+bug/1670319 it leads to sporadic ironic-multinode job failures | |
| 11:03:03 | bhagyashri_s | johnthetubaguy: Hi, I have put my thoughts on patch: https://review.openstack.org/#/c/409644/ could you please take some insight and suggest your opinion about it? | |
| 11:04:21 | johnthetubaguy | bhagyashri_s: I don't think we should do the delete in the API, we try to keep the API doing as little work as possible, and dispatch any long running work to either a conductor worker or a compute worker | |
| 11:05:52 | bhagyashri_s | johnthetubaguy: ok. | |
| 11:08:47 | bhagyashri_s | johnthetubaguy: thank you. | |
| 11:10:26 | johnthetubaguy | bhagyashri_s: no problem, it was an idea certainly worth thinking about, but yeah it would chance how we scale the ystem | |
| 11:11:41 | openstackgerrit | Roman Podoliaka proposed openstack/nova master: nova-status: require placement >= 1.4 https://review.openstack.org/441881 | |
| 11:11:50 | bhagyashri_s | johnthetubaguy: yeah. | |
| 11:17:08 | gibi | johnthetubaguy: hi! as we agreed on the PTG there is a spec proposed to show the scheduler-hints in the response of the GET /servers/ https://review.openstack.org/#/c/440580/ . If you have time could you check it please? | |
| 11:18:04 | johnthetubaguy | gibi: yeah, I hope to get to that | |
| 12:24:31 | alex_xu | johnthetubaguy: haha, that is sweet doc, happy to help that | |
| 12:25:45 | johnthetubaguy | alex_xu: awesome, thanks. The SWG is trying to help share best practices around all parts of the community, and look for new ideas to improve things too | |
| 12:28:04 | alex_xu | johnthetubaguy: got it, i will check if I have some idea can put in that doc | |
| 12:33:04 | johnthetubaguy | alex_xu: attempting to write up things in a central place here: https://review.openstack.org/441923 | |
| 12:33:15 | johnthetubaguy | alex_xu: would love your ideas around all that | |
| 12:34:34 | alex_xu | johnthetubaguy: cool, go tit, I will try to summary some | |
| 12:34:40 | alex_xu | s/tit/it/ | |
| 12:38:41 | satyar | Hi Everyone | |
| 12:39:12 | openstack | Launchpad bug 1668267 in OpenStack Compute (nova) "Nova fails to correctly update the quota when deleting lots of VMs, some of which previously failed due to a quota error" [High,In progress] - Assigned to Pushkar Umaranikar (pushkar-umaranikar) | |
| 12:39:12 | satyar | my quota usages are full because of the bug https://bugs.launchpad.net/nova/+bug/1668267 | |
| 12:39:20 | satyar | any way to free it up? | |
| 12:39:38 | johnthetubaguy | satyar: did you try the nova-manage command to reset the quotas? | |
| 12:39:46 | satyar | yes | |
| 12:39:50 | satyar | didnt worked out | |
| 12:40:25 | johnthetubaguy | thats interesting in its self I guess | |
| 12:41:28 | johnthetubaguy | going for lunch right now, just say that pop up, will be back in 30 mins or so | |
| 12:41:46 | satyar | sure | |
| 12:50:26 | openstackgerrit | Hironori Shiina proposed openstack/nova master: Ironic: Support boot from Cinder volume https://review.openstack.org/215385 | |
| 13:08:32 | satyar | johnthetubaguy: nova-manage account quota refresh workout for me if you know the file which does the update of quota after deletion i can look at the code to fix it | |
| 13:08:45 | satyar | i would say try to fix it ;) | |
| 13:10:05 | cdent | sdague: thanks for making that post clarifying the log translation question. I think _not_ translating would be good (it breaks googling). | |
| 13:14:31 | sdague | cdent: I've been on all sides of this one, and am fine with letting the i18n team set the policy | |
| 13:14:43 | sdague | but it looks like the policy changed, and it wasn't fully communicated | |
| 13:14:47 | cdent | yes | |
| 13:32:37 | johnthetubaguy | satyar: this is the code: https://github.com/openstack/nova/blob/master/nova/compute/api.py#L1776 | |
| 13:33:28 | johnthetubaguy | satyar: this path is quite new, it might be worth some extra logging to see what is happening: https://github.com/openstack/nova/blob/master/nova/compute/api.py#L1730 | |
| 13:36:45 | satyar | Thanks johnthetubaguy | |
| 13:37:52 | johnthetubaguy | satyar: its worth looking in your DB at the reservations table, its possible there are stale ones in there (we should probably get that nova-manage cmd to clear those out too) | |