Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-02-17
14:47:24 artom Has anyone with gmail solved the CI bot spam problem?
14:47:34 artom Gmail is retarded in that it applies filters to entire "conversations"
14:48:04 artom So while it filters out all the CIs spamming me, it also filters out the useful human reviews, and Jenkins, since they're in the same conversation
14:50:26 sdague artom: I found gmail filtering completely unuseful for gerrit, one of the reasons I moved over to fastmail
14:51:25 artom sdague, very agree on point 1, will explore point 2, thanks :)
14:52:37 superdan I have found gmail useful for not very much
14:54:57 artom superdan, what do you use for the gerrit spam?
14:55:09 superdan artom: sieve :)
14:55:51 cdent sieve is too new and fancy, I use procmail :)
14:56:07 superdan procmail > gmail
14:56:19 openstackgerrit Sean Dague proposed openstack/nova master: Verify project id for flavor access calls https://review.openstack.org/435432
14:56:19 superdan well, sieve > sliced bread > procmail > gmail
14:56:20 superdan to be complete
14:56:20 openstackgerrit Sean Dague proposed openstack/nova master: Verify project_id when quotas are checked https://review.openstack.org/435010
14:56:33 artom superdan, oh, you're one of those
14:56:38 artom With their own email server
14:56:47 artom Do you also grow your own wheat for bread? ;)
14:57:02 superdan artom: I would if the store stopped selling good bread
14:57:31 artom But the store is giving the bread away
14:57:45 artom It just asks that you store all your personal paperwork in their vault
14:57:50 superdan it's giving away bean bread
14:58:03 superdan which is not really bread
14:58:11 superdan it doesn't have slices, although it has lines to look like slices
14:58:48 artom This comparison is starting to break down, and I'm very confused
14:59:08 superdan no real folders in gmail == no slices
14:59:24 superdan anyway, you keep using your store bought bread and I'll keep filtering gerrit spam
14:59:41 cdent artom: that (This comparison...) is what the special guest on the superdan weeky comedy hour always says at the end of the show.
15:00:38 artom cdent, wait, there's a comedy show at the bread store now?
15:00:55 artom How does email fit into that? Is it the heckler?
15:02:00 cdent My brain is starting to break down, and I'm very confused.
15:18:38 des_consolado gah I'm trying to create instances and they just get stuck in the Spawning state and then eventually fail after ages, saying failed to provision instance
15:18:42 openstackgerrit John Garbutt proposed openstack/nova master: POC: improved policy functional tests https://review.openstack.org/435484
15:18:43 openstackgerrit John Garbutt proposed openstack/nova master: POC: add context.check_scope https://review.openstack.org/435485
15:18:50 des_consolado something or other timed out etc
15:18:59 des_consolado I'm tailing every nova log and nothing interesting is there
15:23:52 artom des_consolado, you're probably better off asking in #openstack, as this channel is for development, but it could be waiting for an external event. I've seen these symptoms before where a network-plugged event from Neutron didn't arrive, and the spawn timed out.
15:25:48 des_consolado oh yeah I just read the topic, sorry
15:26:33 des_consolado the other openstack project channels don't seem to make as much of a point about that
15:27:07 des_consolado I also noticed the developers of given projects tend to have all the answers :P
15:29:11 mriedem des_consolado: if it's a virtual interface exception / timeout thing in the n-cpu logs, then yes it's timing out waiting for the external vif plugged callback event from neutron as artom points out
15:29:30 mriedem which probably means something is failing on the neutron side, check the neutron agent health status and logs
15:30:33 sdague johnthetubaguy: I respun with the drop on the delete call
15:32:45 cdent mriedem: was there a change you pushed up as a result of your kvm-on-the-subnode debugging last night? I'd like to look at it if so.
15:33:18 mriedem yeah, sec
15:33:33 mriedem cdent: https://review.openstack.org/#/c/435154/
15:33:38 cdent thanks
15:33:41 mriedem your googling about nested kvm tipped me off
15:33:55 mriedem then i asked in infra and clarkb confirmed the nested kvm issue, and we found the problem in d-g
15:33:57 mriedem mega bash
15:34:00 mriedem what could go wrong
15:34:56 cdent your change was interesting enough to pull jo*go?
15:40:54 cdent mriedem: were you a good boy and search for 'localrc' where $localrc_file ought to be in the rest of the file, or only fix the presenting problem?
15:41:41 mriedem cdent: i pinged him
15:42:34 mriedem cdent: i did
15:43:49 cdent 🍪
16:01:47 mriedem yikes http://logs.openstack.org/11/349011/3/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial-ocata/bee08f9/logs/screen-n-api.txt.gz?level=TRACE#_2017-02-14_22_13_03_345
16:01:55 mriedem Passing insecure dynamic vendordata requests because of missing or incorrect service account configuration.
16:01:57 mriedem all over the logs
16:02:38 melwitt I have noticed that too
16:03:01 melwitt haven't made time to look into it yet
16:04:54 mriedem http://git.openstack.org/cgit/openstack/nova/tree/nova/api/metadata/vendordata_dynamic.py#n52
16:04:58 mriedem we should just log that once
16:05:01 mriedem i'm reporting a bug
16:05:10 mriedem if LP will stop timing out
16:07:22 melwitt nice
16:12:54 sdague mriedem: isn't that feature opt in?
16:13:03 sdague so if it's off, shouldn't we just not log
16:13:14 openstack Launchpad bug 1665693 in OpenStack Compute (nova) ""Passing insecure dynamic vendordata requests because of missing or incorrect service account configuration." warnings all over n-api logs" [Medium,Confirmed]
16:13:14 mriedem https://bugs.launchpad.net/nova/+bug/1665693
16:13:21 mriedem sdague: maybe, i haven't dug in
16:14:38 sdague mriedem: I do agree, it's logging a lot
16:19:40 sdague mriedem: if you get bored - https://review.openstack.org/#/c/435010/ is the code for the project_id enforcement
16:20:07 sdague we get some review on it and I could probably have that in land state for next week. Early blueprint completion for the win.
16:25:33 mriedem sfinucan: did i answer your questions here? https://review.openstack.org/#/c/380433/
16:25:46 mriedem sdague: ok
16:30:22 mriedem sdague: when do you get into ATL?
16:30:26 mriedem mtreinish: ^
16:30:40 sdague mriedem: like 9pm on Sunday
16:31:02 mriedem geez
16:31:35 sdague I can send you my tripit info if you like
16:33:23 sdague mriedem: why, when are you getting in?
16:34:49 sdague I think my reasonable options were this flight, or a super early in the morning one on Sunday, so I went with not having to wake up at unusual times
16:36:22 mriedem sdague: comments in https://review.openstack.org/#/c/435010/
16:36:28 mriedem sdague: 2:45 on sunday
16:41:44 mriedem i know why we're getting the billions of vendordata warnings then, it's because it's hit on each init of the InstanceMetadata object
16:41:51 mriedem which is every server that gets created with config drive
16:44:23 mriedem sdague: so can we just pick one of these options to tell if we can safely assume dynamic vendordata is just not configured to work? http://logs.openstack.org/11/349011/3/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial-ocata/bee08f9/logs/screen-n-api.txt.gz#_2017-02-14_21_31_00_838
16:44:33 mriedem like auth_section or auth_type?
16:44:35 mriedem jamielennox: ^
16:45:39 mriedem sounds like auth_section is optional
16:45:41 mriedem " auth_section: the group from which further auth plugin options should be taken. If section is not provided then the auth plugin options will be taken from the same group as provided in the parameters. "
16:45:58 mriedem but auth_type has to be set right?
16:46:04 mriedem lbragstad: ^?
16:48:05 mriedem well, i guess it doesn't matter, we can just log it once
16:48:18 mriedem we still make the requests, they just aren't with the service token
16:55:27 lbragstad mriedem yes - i believe auth_type needs to be set
17:13:05 openstackgerrit Chris Dent proposed openstack/nova master: Update placement_dev with info about new decorator https://review.openstack.org/435539
17:13:58 sdague mriedem: of course, by tightening up the exception handling, I can't use the testing exception :)
17:16:33 cdent Is there a rule of thumb on when lack of response on a bug report is cause for marking invalid?
17:17:50 sdague cdent: if you set it to Incomplete, and remove assignee, it will expire in 60 days (IIRC)

Earlier   Later