| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-09-06 | |||
| 17:51:45 | dansmith | jaypipes: I've been welding it with fire-proof steel in my garage waiting for the announcement | |
| 17:51:48 | jaypipes | dansmith: in other news, +W'd the migration allocation spec | |
| 17:51:54 | jaypipes | dansmith: well played. | |
| 17:51:57 | cburgess | jaypipes So thats real? I saw something on facebook but you know you can't trust facebook. | |
| 17:51:58 | dansmith | jaypipes: if this isn't an advertisement for not having children, I don't know what is | |
| 17:52:10 | jaypipes | dansmith: :) ++ | |
| 17:52:35 | dansmith | "Just look at the widespread devastation that can result from a single child. JUST SAY NO" | |
| 17:53:02 | mriedem | yeah, because drunk 40 year old yokuls shooting off guns never caused any harm | |
| 17:53:15 | mriedem | YEEEEEEEEEEHHHHHAW! | |
| 17:53:16 | cburgess | mriedem We probably shouldn't have them either. | |
| 17:53:29 | dansmith | heh | |
| 17:53:35 | mikal | Heh | |
| 17:53:51 | jaypipes | mriedem: who you calling drunk? | |
| 17:53:58 | mriedem | yokuls | |
| 17:53:59 | mriedem | duh | |
| 17:54:01 | mikal | Google had multiple fibre outages in Oregon because of yokels using the fibre for target practise | |
| 17:54:06 | dansmith | mriedem: problem is, that yokul would go to jail for this, but the kid will get a slap on the wrist and community service | |
| 17:54:14 | jaypipes | mriedem: I'll let you know I done drunk my dranks responsibally | |
| 17:55:09 | cburgess | mriedem In dansmith's world I think we have to take you out back and shoot you now. | |
| 17:55:13 | mriedem | told the authorities it was some other bad kids that they just missed when we got there | |
| 17:55:13 | jaypipes | "accident". yeah, right, mriedem. | |
| 17:55:26 | dansmith | mriedem: well, this kid wasn't even _that_ smart | |
| 17:55:33 | mriedem | "geez i'm real sorry mister" | |
| 17:55:39 | dansmith | cburgess: retroactive "was a kid once" punishment? harsh | |
| 17:55:54 | cburgess | dansmith Thats how I role. | |
| 17:56:10 | dansmith | just ask my parents | |
| 17:56:13 | cburgess | lol | |
| 17:56:27 | mriedem | the old benjamin button disease | |
| 17:56:37 | openstackgerrit | Merged openstack/nova-specs master: Add migration-allocations spec https://review.openstack.org/498510 | |
| 17:56:53 | dansmith | ran my middle school bedroom like an office with an inbox and credenza, made my mom call me on the internal phone system if she needed something | |
| 17:57:06 | jaypipes | nice | |
| 17:57:15 | jaypipes | I should do that with Julie and the pugs. | |
| 17:57:32 | jaypipes | "excuse me, Fiona. you need to make an appointment." | |
| 17:57:32 | dansmith | and then put nothing but an answering machine on the other end? | |
| 17:57:42 | jaypipes | dansmith: indeed. | |
| 17:57:48 | cburgess | Now I have this image of a small dansmith sitting a little desk -2ing requests from his mother. | |
| 17:57:58 | dansmith | cburgess: not far off | |
| 17:57:58 | jaypipes | lmao | |
| 17:59:25 | dims | cburgess : LOL | |
| 18:00:06 | mikal | "Press 1 to ask about my grades, press 2 to complain about my girlfriend, press 3 to..." | |
| 18:01:38 | cburgess | mikal lol | |
| 18:03:47 | mikal | "Press 4 to ask what that loud fan noise coming from the garage is" | |
| 18:03:54 | openstackgerrit | Merged openstack/nova master: Make xen unit tests work with os-xenapi>=0.3.0 https://review.openstack.org/500968 | |
| 18:03:59 | mikal | Answer, by the way, its a loud fan | |
| 18:04:04 | dansmith | heh | |
| 18:04:14 | dansmith | I did have a server catch fire in my room one time | |
| 18:04:29 | mikal | Haven't we all... | |
| 18:04:30 | dansmith | still have the (remnants of the) motherboard | |
| 18:04:59 | mikal | Anyways, its time for me to brave LA traffic and go to the airport. | |
| 18:05:09 | mikal | I shall say sweet sweet lies about you all at the opendev thing | |
| 18:05:26 | mikal | "Oh yes, they're quite competent. Have you asked them how they feel about remote compute nodes?" | |
| 18:05:45 | mikal | "They love that stuff, want more of it. Please send many patches." | |
| 18:05:56 | cburgess | lol | |
| 18:06:14 | mikal | "Ask for Dan, he'll help" | |
| 18:06:44 | melwitt | you're going to opendev? so am I | |
| 18:08:15 | cburgess | dansmith Wait what did I do? | |
| 18:08:34 | dansmith | cburgess: you encouraged him by laughing at his jokes | |
| 18:08:58 | cburgess | dansmith Oh that... | |
| 18:14:41 | openstackgerrit | Merged openstack/nova master: doc: Remove deprecated call to sphinx.util.compat https://review.openstack.org/498824 | |
| 18:26:55 | mnaser | dansmith / mriedem: https://bugs.launchpad.net/nova/+bug/1715462 | |
| 18:26:57 | openstack | Launchpad bug 1715462 in OpenStack Compute (nova) "Instances failing quota recheck end up with no assigned cell" [Undecided,New] | |
| 18:28:00 | sdague | cburgess: what's your biggest concern on privsep? | |
| 18:28:13 | sdague | I apparently didn't see that in the patch (or glossed past it) | |
| 18:28:14 | mnaser | i guess a test to make sure that a cell is assigned if the quota recheck fails would be #1 and then the fix to show that its working would be the correct path? | |
| 18:28:35 | dansmith | mnaser: that'd be ideal yeah | |
| 18:28:44 | mnaser | ok, ill try to work on a test first | |
| 18:28:46 | cburgess | sdague I haven't reviewed the patch this was a review mikal and I did verbally in my dinning room this morning. | |
| 18:29:37 | sdague | cburgess: were mimosas involved? | |
| 18:29:58 | cburgess | sdague I have a few concerns but the big one is around how the privsep daemons get started. I would prefer that if a nova component detects that there is no daemon it just started/restarted it. Having to restart the entire compute/conductor/whatever process is faily heavy weight just to get the privsep daemon running again. | |
| 18:30:06 | cburgess | sdague Oh I wish.. that would have made it better | |
| 18:30:29 | sdague | cburgess: yeah, the watchdogging seems reasonable | |
| 18:30:30 | mnaser | dansmith mind if i bother you a bit as i figure out the best way to go about this with small questions? https://github.com/openstack/nova/blob/master/nova/tests/unit/conductor/test_conductor.py#L1722-L1760 -- would it be beneficial to just add an assert there instead of writing a whole new test? | |
| 18:30:54 | openstackgerrit | Merged openstack/nova master: trivial: Remove "vif" script https://review.openstack.org/491443 | |
| 18:31:19 | dansmith | mnaser: since that test is specifically aimed at checking that late quota check I think that's probably fine | |
| 18:31:33 | mnaser | alright cool, i'll give it a shot, lets hope it fails :> | |
| 18:31:36 | openstackgerrit | Merged openstack/nova master: trivial: Remove files from 'tools' https://review.openstack.org/491444 | |
| 18:31:39 | cburgess | sdague Cool so we agree and mikal did in theory when I went to the gym this morning but now he claims there might be issues and promises performance art in Denver. | |
| 18:36:59 | mnaser | woo, i have a failing test case | |
| 18:39:08 | mnaser | can i get some advice on how would be the ideal way of where to move the cell_mapping saving part? should all of it be moved earlier or should i refactor it into a private fucntion such as _update_instance_mapping(instance, cell) and then call that in the exception handling part of the quota recheck? | |
| 18:39:28 | openstackgerrit | Erik Berg proposed openstack/nova master: Fix binary name. https://review.openstack.org/501359 | |
| 18:39:40 | mnaser | that way we don't risk introducing other weird problems that might come up from shuffling the order of things | |
| 18:42:12 | dansmith | mnaser: this is a sticky place we have to be super careful | |
| 18:42:14 | dansmith | give me a few to read | |
| 18:43:02 | mnaser | dansmith: no problem! that's why i thought by refactoring it into a function and calling it in exception handling, we touch the *least* amount of codepath possible (but i'm sure folks know the codebase far more than me :-) | |
| 18:46:29 | dansmith | mnaser: so I think the right thing to do is really to map the instance right after we create it above | |
| 18:46:51 | dansmith | mnaser: so that if we end up with any instance created (which will show up in a list) it'll have a corresponding map show that show will work | |
| 18:46:58 | dansmith | s/show that/so that/ | |
| 18:47:50 | mnaser | dansmith so right after the with(..) block spanning 988-990? | |
| 18:48:03 | dansmith | mnaser: year | |
| 18:48:10 | dansmith | heh, that was either yeah or yar | |
| 18:48:43 | mnaser | dansmith in my research of the code, i found the _populate_instance_mapping function which seemed pretty robust at setting the instance mapping | |
| 18:48:51 | mnaser | would it beneficial to use that instead? | |
| 18:49:25 | mnaser | we have the host in there so we can pass it (but i'll take what you think is best overall) | |
| 18:50:25 | dansmith | mnaser: yeah if that works should be okay | |
| 18:51:06 | mnaser | dansmith cool, i'll get on this and see if it affects any other tests as well | |
| 18:51:45 | dansmith | cool | |
| 18:55:08 | mnaser | yay, that specific test is passing now, i'll just rerun all the conductor tests because that code shuffle might have affected other tests | |
| 18:55:50 | dansmith | yeah, entirely possible because that extra thing isn't mocked out now | |
| 18:59:38 | mnaser | only one test failing after that.. a bit less painless than i expected :> | |
| 18:59:56 | dansmith | run functional tests too? | |