| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-20 | |||
| 19:24:18 | mriedem | the point of putting discover_hosts into a script in devstack is to avoid needing ugly branch logic in d-g itself | |
| 19:34:45 | sdague | mriedem: I think so... | |
| 19:36:03 | openstackgerrit | Francesco Santoro proposed openstack/nova: os-vif: convert libvirt driver to use os-vif for fast path vhostuser https://review.openstack.org/385061 | |
| 19:36:16 | melwitt | something blowing up on the bottom change | |
| 19:38:23 | openstackgerrit | STEW TY proposed openstack/nova: Transform instance.rescue notifications https://review.openstack.org/385644 | |
| 19:39:42 | melwitt | /opt/stack/new/devstack-gate/devstack-vm-gate.sh: line 288: localrc: Permission denied | |
| 19:40:10 | melwitt | I wonder if those commands have to be after setup_localrc or something? | |
| 19:43:56 | openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/422323 | |
| 19:47:47 | mriedem | maybe but i'm not sure why | |
| 19:50:48 | mriedem | sdague: ^ do we need to run setup_localrc before we can change permissions on the files in the workspace? | |
| 19:51:22 | sdague | is there a full log somewhere? | |
| 19:52:37 | melwitt | sdague: http://logs.openstack.org/17/423417/1/check/gate-grenade-dsvm-neutron-ubuntu-xenial/479e3c9/console.html#_2017-01-20_19_25_44_044049 I don't know how to see more detail than that | |
| 19:55:10 | mriedem | yeah that's all i see | |
| 19:55:17 | melwitt | and this is the change https://review.openstack.org/#/c/423417/ | |
| 19:55:18 | mriedem | it calls setup_localrc and things blow up | |
| 19:55:53 | mriedem | oh i think i know | |
| 19:56:03 | mriedem | because d-g can't write to the files because stack user owns then now? | |
| 19:56:12 | mriedem | so, zuul user can't write to them in d-g | |
| 19:56:19 | melwitt | ohh | |
| 19:57:59 | melwitt | so I guess normally it sets up things first, then changes things over to the stack user | |
| 20:00:17 | sdague | we do definitely drop privs completely for safety reasons, and remove files from sudoers | |
| 20:02:37 | mriedem | ok, so i'll redo this and just move that stuff into a common function | |
| 20:02:41 | mriedem | so i can also call from the grenade branch | |
| 20:06:51 | sdague | yeh, your analysis that you are changing permissions too early is right, sorry it took a minute for me to gather all the context | |
| 20:07:21 | openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP] ignore me - trying experimental jobs https://review.openstack.org/421209 | |
| 20:08:33 | melwitt | thanks sdague | |
| 20:08:43 | mriedem | pushed | |
| 20:08:52 | mriedem | just in time for laura to walk in | |
| 20:08:59 | sdague | the permission changing is done for the stack user right before the stack.sh hand off. It starts as being under the jenkins user. The stack user can itself sudo up until this point - https://github.com/openstack-infra/devstack-gate/blob/5883a8269a741d77c1853150c56d6d90fe01b83e/devstack-vm-gate.sh#L720-L724 | |
| 20:09:16 | sdague | it's sudo privs are taken away before tempest runs, so that tempest doesn't cheat | |
| 20:09:19 | dims | mriedem :) | |
| 20:09:46 | dims | sdague : gate is 39 deep, something got busted? | |
| 20:10:02 | sdague | dims: as far as I can, it's just a ton of paper cuts | |
| 20:10:20 | sdague | nova notification tests are sometimes failing | |
| 20:10:23 | dims | ouch ok. at least we have weekend to catch up on | |
| 20:10:26 | sdague | I saw some db things failing on cinder | |
| 20:10:31 | dims | will watch a bit as well | |
| 20:10:42 | sdague | the keystone os-api-ref one, which I pushed a fix (and email on) | |
| 20:11:29 | dims | ack y i saw that. thanks sdague | |
| 20:14:50 | dims | ugh. just spotted this in the gate queue - http://logs.openstack.org/11/423111/2/gate/gate-tempest-dsvm-neutron-full-ubuntu-xenial/60c1d57/logs/screen-n-cpu.txt.gz?level=ERROR#_2017-01-20_19_36_46_607 | |
| 20:15:00 | dims | libvirtError: internal error: client socket is closed | |
| 20:16:42 | mriedem | dims: that's known | |
| 20:16:50 | mriedem | libvirt crashes intermittently on xenial | |
| 20:17:10 | mriedem | http://status.openstack.org//elastic-recheck/index.html#1643911 | |
| 20:17:18 | mriedem | http://status.openstack.org//elastic-recheck/index.html#1646779 | |
| 20:17:25 | mriedem | http://status.openstack.org//elastic-recheck/index.html#1638982 | |
| 20:18:36 | dims | ack mriedem | |
| 20:21:54 | mriedem | jroll: vdrok: sambetts|afk: can i mark this as complete too? https://blueprints.launchpad.net/nova/+spec/ironic-plug-unplug-vifs-update | |
| 20:22:00 | mriedem | the vifs plug/unplug patch is merged, | |
| 20:22:05 | mriedem | the interface attach/detach thing isn't | |
| 20:22:21 | jroll | mriedem: yes, that doesn't mention the latter, that should be a separate blueprint imo | |
| 20:22:24 | mriedem | which is https://review.openstack.org/#/c/419975/ | |
| 20:22:26 | mriedem | ok | |
| 20:22:27 | mriedem | cool | |
| 20:23:01 | mriedem | nice both of those bps are merged then (portgroups and vifs update) | |
| 20:28:35 | bauzas | oh man, the gate... | |
| 20:37:26 | mriedem | ew we unconditionally run osprofiler on all unit test runs | |
| 20:37:30 | mriedem | even if i just want to test a specific path | |
| 20:50:40 | openstackgerrit | Julian proposed openstack/nova: Create cmd/script that prints JSON Schema of Notification Payload Classes https://review.openstack.org/411836 | |
| 20:52:30 | openstackgerrit | Matt Riedemann proposed openstack/nova: os-vif: convert libvirt driver to use os-vif for fast path vhostuser https://review.openstack.org/385061 | |
| 20:52:55 | mriedem | sean-k-mooney: leakypipes: i'm happy with ^ now | |
| 20:53:53 | mriedem | actually, hmm | |
| 20:54:18 | leakypipes | mriedem: need to boot into Windows VM... biab. (yep, Webex calls) | |
| 21:00:58 | bauzas | mriedem: sdague: I'm currently writing a patch for adding the microversion header in the report client | |
| 21:01:12 | bauzas | mriedem: sdague: do you think we should default to the min version or the max one ? | |
| 21:01:18 | bauzas | I think the latter | |
| 21:01:39 | bauzas | but maybe lemme upload the change | |
| 21:01:49 | sdague | bauzas: I thought we always default to min unless specified | |
| 21:02:03 | bauzas | sdague: I first thought about that | |
| 21:02:46 | bauzas | sdague: okay, lemme try to do this | |
| 21:02:56 | bauzas | and upload it so we can disucss | |
| 21:08:26 | mriedem | bauzas: i put up something like that last night | |
| 21:09:03 | openstackgerrit | Matt Riedemann proposed openstack/nova: os-vif: convert libvirt driver to use os-vif for fast path vhostuser https://review.openstack.org/385061 | |
| 21:09:09 | mriedem | bauzas: in here https://review.openstack.org/#/c/407309/ | |
| 21:09:17 | mriedem | that requests 1.1 specifically for the aggregates GET it needs | |
| 21:09:28 | mriedem | which may fail, but then it's just that aggregates GET that fails and not everything else | |
| 21:09:35 | bauzas | mriedem: :( | |
| 21:09:44 | mriedem | the scheduler report client is currently not doing version discovery | |
| 21:09:51 | bauzas | lemme upload it | |
| 21:11:50 | mriedem | sfinucan: +2 on https://review.openstack.org/#/c/385061/ | |
| 21:15:31 | openstackgerrit | Matt Riedemann proposed openstack/nova: Ironic: Add soft reboot support to ironic driver https://review.openstack.org/403745 | |
| 21:17:57 | bauzas | mriedem: KABOOOM | |
| 21:18:08 | bauzas | mriedem: my patch is trampled by the above ^ | |
| 21:18:17 | bauzas | mriedem: so I'll remove the get header | |
| 21:19:04 | openstackgerrit | Eric Fried proposed openstack/nova: Introduce base PowerVM driver (disk drv) https://review.openstack.org/409401 | |
| 21:29:51 | mriedem | https://review.openstack.org/#/c/403745/ also looks good to go | |
| 21:30:30 | slaweq_ | oomichi: hello | |
| 21:30:54 | slaweq_ | oomichi: some time ago You reviewed my small patch https://review.openstack.org/#/c/420176/ | |
| 21:31:15 | slaweq_ | oomichi: I answered to Your comment there, can You maybe have a look on it? | |
| 21:31:19 | slaweq_ | oomichi: thx in advance | |
| 21:32:47 | mriedem | claudiub: you're going to have to sync up with markus_z or someone from s390 on https://review.openstack.org/#/c/420960/ else we'll just have to move that forward without the s390 features | |
| 21:37:05 | mriedem | melwitt: booya http://logs.openstack.org/05/423005/6/check/gate-grenade-dsvm-neutron-ubuntu-xenial/90ab2f2/logs/devstack-gate-discover-hosts.txt.gz | |
| 21:37:59 | mriedem | the stable jobs are still failing, looks like they are still trying to run discover_hosts, | |
| 21:38:05 | mriedem | http://logs.openstack.org/05/423005/6/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial-newton-nv/39d0108/logs/devstack-gate-discover-hosts.txt.gz | |
| 21:38:21 | mriedem | i wonder if the "test -f discover_hosts.sh" is making it exit with 1 | |
| 21:39:00 | mriedem | i think it is | |
| 21:39:00 | mriedem | user@xubuntu:~/git/devstack$ cd ~/git/devstack && test -f discover_hosts1.sh && sudo -H -u user stdbuf -oL -eL cat discover_hosts.sh | |
| 21:39:01 | mriedem | user@xubuntu:~/git/devstack$ echo $? | |
| 21:39:01 | mriedem | 1 | |
| 21:40:41 | mriedem | which means i basically can't avoid branch logic in d-g for this | |