| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-07-17 | |||
| 20:17:57 | mriedem | yeah | |
| 20:17:58 | mriedem | "command line args" | |
| 20:18:02 | mriedem | "config files" | |
| 20:18:07 | mriedem | like here http://logs.openstack.org/47/484447/1/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/b2fbd0c/logs/screen-n-cpu.txt.gz#_Jul_17_18_47_11_345788 | |
| 20:18:22 | mriedem | and then it dumps the options it's using | |
| 20:19:05 | mriedem | Ctesias: you're using the wrong group | |
| 20:19:09 | mriedem | it's not [keymgr] | |
| 20:19:13 | mriedem | it's [key_manager] | |
| 20:19:21 | mriedem | http://logs.openstack.org/47/484447/1/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/b2fbd0c/logs/screen-n-cpu.txt.gz#_Jul_17_18_47_11_670125 | |
| 20:19:28 | Ctesias | yeah, that was a typo earlier. I've got [key_manager] in the file | |
| 20:21:05 | Ctesias | If I put a breakpoint in nova/keymgr/conf_key_mgr.py I can go to the line after it grabs the nova configuration and see it's the same | |
| 20:21:10 | Ctesias | (Pdb) CONF['key_manager']['api_class'] 'nova.keymgr.conf_key_mgr.ConfKeyManager' | |
| 20:23:35 | mriedem | and the debug log output when you restart nova-compute shows the same when it dumps the options? | |
| 20:23:58 | Ctesias | yes | |
| 20:24:26 | Ctesias | 2017-07-17 16:20:13.483 DEBUG oslo_service.service [req-ece128f3-fb7c-4da3-af58-78c39fca910c None None] key_manager.api_class = nova.keymgr.conf_key_mgr.ConfKeyManager from (pid=14641) log_opt_values /opt/pf9/venv/lib/python2.7/site-packages/oslo_config/cfg.py:2626-07-17 16:20:13.485 DEBUG oslo_service.service [req-ece128f3-fb7c-4da3-af58-78c39fca910c None None] osapi_v21.project_id_regex = None from (pid=14641) | |
| 20:25:19 | mriedem | hmm, api_class comes from castellan | |
| 20:25:24 | mriedem | which version of that library do you have installed? | |
| 20:26:05 | Ctesias | 0.4.0 | |
| 20:26:14 | mriedem | kfarr: ^ have you heard of anything like this? | |
| 20:26:40 | mriedem | Ctesias: which version of nova? master (pike)? or ocata? | |
| 20:26:47 | mriedem | because that version is old for master https://github.com/openstack/requirements/blob/master/upper-constraints.txt#L227 | |
| 20:27:10 | Ctesias | Newton, actually | |
| 20:27:14 | mriedem | ew | |
| 20:27:21 | Ctesias | speed of business, man | |
| 20:28:01 | mriedem | so things were different in newton https://github.com/openstack/nova/blob/stable/newton/nova/conf/key_manager.py | |
| 20:28:53 | mriedem | looks like it should still set those options | |
| 20:30:38 | mriedem | Ctesias: this might be your problem https://github.com/openstack/nova/blob/stable/newton/nova/keymgr/__init__.py#L29 | |
| 20:30:50 | mriedem | not totally sure | |
| 20:31:02 | mriedem | need to probably talk to kfarr | |
| 20:32:12 | Ctesias | I don't see api_class as an option in /nova/conf/key_manager.py | |
| 20:32:34 | mriedem | that's not the options registry, | |
| 20:32:40 | mriedem | that's the default impl | |
| 20:32:59 | mriedem | the options are registered here https://github.com/openstack/nova/blob/stable/newton/nova/conf/key_manager.py | |
| 20:33:20 | mriedem | and then there is some backward compat magic in https://github.com/openstack/nova/blob/stable/newton/nova/keymgr/__init__.py | |
| 20:37:16 | Ctesias | yeah commenting out the lines in __init__.py fixes it | |
| 20:38:36 | Ctesias | Is this behavior fixed upstream from Newton anywhere yet? | |
| 20:41:19 | mriedem | i think that stuff in init was all removed in ocata | |
| 20:41:33 | mriedem | oh i guess not ocata https://github.com/openstack/nova/blob/stable/ocata/nova/keymgr/__init__.py | |
| 20:41:44 | mriedem | or master https://github.com/openstack/nova/blob/master/nova/keymgr/__init__.py | |
| 20:41:44 | mriedem | damn | |
| 20:42:14 | mriedem | so the fact it's setting that b/c of devstack is a total hack | |
| 20:42:21 | mriedem | let me see if that's at least set in devstack | |
| 20:43:27 | mriedem | Ctesias: https://review.openstack.org/#/c/439855/ | |
| 20:44:42 | Ctesias | cool. Thanks | |
| 20:45:03 | mriedem | Ctesias: so did you just remove this line? https://github.com/openstack/nova/blob/stable/newton/nova/keymgr/__init__.py#L29 ? | |
| 20:45:08 | mriedem | or all of the stuff in there | |
| 20:46:58 | Ctesias | I only commented out the lines that re-set the api_class value | |
| 20:47:18 | Ctesias | but if you're curious I can try to get rid of all of it and see what happens | |
| 20:47:34 | mriedem | i'm curious if just commenting out https://github.com/openstack/nova/blob/stable/newton/nova/keymgr/__init__.py#L29 resolves your issue | |
| 20:47:41 | mriedem | because that would be easy for me to change on master and backport through to newton | |
| 20:47:55 | Ctesias | Yes. | |
| 20:47:59 | mriedem | ok | |
| 20:48:28 | Ctesias | If you would be so kind as to post that change once it comes in so I can follow it back and get upgraded to it that would be excellent. | |
| 20:48:29 | mriedem | i can mess around with pushing that change later then - i can't say i totally understand *why* it would cause an issue since it should just set a default, not override your value | |
| 20:48:35 | mriedem | unless it thinks that your value was the default or something? | |
| 20:48:42 | Ctesias | beats the hell out of me | |
| 20:49:09 | mriedem | i'll just remove it on the grounds that we have the value set in devstack so removing the hack is good enough | |
| 20:50:40 | mriedem | Ctesias: reporting a bug against nova would also help with the backports | |
| 20:50:51 | mriedem | https://bugs.launchpad.net/nova/+filebug | |
| 20:52:08 | Ctesias | will do | |
| 21:13:23 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Update policy descriptions for base. https://review.openstack.org/472622 | |
| 21:13:23 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Consistent policies https://review.openstack.org/479774 | |
| 21:14:39 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Update policy descriptions for base. https://review.openstack.org/472622 | |
| 21:15:21 | mriedem | oomichi: another easy one and the final patch for the policy docs bp https://review.openstack.org/#/c/472622/ | |
| 21:27:48 | mriedem | who here feels like adding back in the ability to load up out of tree scheduler drivers? https://bugs.launchpad.net/nova/+bug/1704788 | |
| 21:27:49 | openstack | Launchpad bug 1704788 in OpenStack Compute (nova) "Hardcoded choices for nova scheduler driver" [Undecided,Confirmed] | |
| 21:30:03 | dansmith | heh | |
| 21:30:24 | mriedem | bauzas: since you were +W on that thing, how do you feel about ^ | |
| 21:30:32 | mriedem | answer tomorrow when you're awake | |
| 21:30:56 | mriedem | because we really probably should have made that (1) a deprecation cycle and (2) ffs at least reno it | |
| 21:33:22 | openstackgerrit | Merged openstack/nova master: Handle Cinder 3.27 style attachments in swap_volume https://review.openstack.org/456971 | |
| 21:34:11 | openstackgerrit | Merged openstack/nova master: [placement] Update allocation-candidates.yaml for gabbi 1.35 https://review.openstack.org/484356 | |
| 21:34:23 | cdent | \o/ | |
| 21:34:34 | mriedem | cdent: btw, just so you know, | |
| 21:34:49 | mriedem | we went hiking, toured a historic village, and went swimming yesterday | |
| 21:35:00 | mriedem | so it wasn't all work | |
| 21:35:04 | cdent | ah well, I am so relieved | |
| 21:36:36 | melwitt | mriedem: replied to your latest comments. the things that seem up in the air are 1) whether to add vm_states.SOFT_DELETED + task_states.RESTORING instances to the count and recheck in restore | |
| 21:36:48 | melwitt | 2) whether to check quota on resize confirm | |
| 21:38:20 | mriedem | ok | |
| 21:38:57 | mriedem | oh dan https://review.openstack.org/#/c/472622/ <3 | |
| 21:40:13 | melwitt | mriedem: I'm not sure what I think other than, if I do both 1) and 2) I guess that eliminates the TODOs I had written bc checking quota at resize confirm sort of serves as a recheck for resize (as I described in the review comment) | |
| 21:40:47 | melwitt | so maybe that one should be guarded by "if CONF.quota.recheck_quota" | |
| 21:44:44 | melwitt | correction, the revert_resize TODO would remain | |
| 21:44:51 | dansmith | mriedem: we might want to get this in before the quota-ectomy so we can backport: https://review.openstack.org/#/c/462521/4/nova/compute/manager.py | |
| 21:45:39 | mriedem | dansmith: that's 3 changes in a series | |
| 21:45:42 | mriedem | the 2nd change has no tests | |
| 21:46:09 | dansmith | ah, yup | |
| 21:49:07 | dansmith | the bottom one doesn't either and it seems like they'd be trivial | |
| 21:58:15 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Remove key_manager.api_class hack https://review.openstack.org/484501 | |
| 22:15:48 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Log compute node uuid when the record is created https://review.openstack.org/484447 | |
| 22:29:41 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Convert HostState.limits['numa_topology'] to primitive https://review.openstack.org/484439 | |
| 22:29:43 | mriedem | dansmith: ^ is that what you were looking for wrt being specific? | |
| 22:30:11 | dansmith | mriedem: well kinda, although I don't see why you need to check for if it's an object | |
| 22:30:24 | dansmith | that's just too much ambiguity unless it's really needed | |
| 22:30:30 | dansmith | if it's set, it's that object, right? | |
| 22:30:43 | mriedem | honestly i'd rather not even do this specific check at all | |
| 22:30:47 | mriedem | and just handle it generically | |
| 22:31:06 | mriedem | i don't like the tight coupling between the filters and the scheduler manager here | |
| 22:32:49 | dansmith | idk, I just think more magic object conversions hide where this is happening | |