Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2018-12-11
17:00:36 openstackgerrit Dmitry Tantsur proposed openstack/openstacksdk master: Import code for building ironic-compatible configdrives https://review.openstack.org/624441
17:00:38 dtantsur mordred: dunno if you're going to hate this ^^^ :)
17:05:18 Shrews dtantsur: why would he hate it? seems very convenient to me
17:05:59 cdent elmiko, edleafe: we've just an EOY API-SIG report request in the mail, can I leave that in your capable hands?
17:06:01 dtantsur dunno, it's not about directly interacting with the API
17:11:32 edleafe cdent: Yeah, I can do that for the SIG
17:11:43 cdent kewl
17:24:19 elmiko edleafe: thanks =)
17:24:55 edleafe elmiko: I should have something in time for the office hour on Thursday for you to look over.
17:40:22 mordred dtantsur: yeah - I agree with Shrews - I think it's a good idea - kind of like the image signing code
17:40:30 dtantsur cool :)
17:43:52 elmiko edleafe: awesome, thanks again for taking the lead on that
20:23:10 Shrews mordred: i think the latest dogpile.cache release breaks us
20:23:25 Shrews http://logs.openstack.org/41/624441/1/check/nodepool-functional-py35-src/88da860/controller/logs/screen-nodepool-builder.txt.gz#_Dec_11_17_48_29_135083
20:23:45 Shrews dogpile.cache 0.7.0 release about 17 hours ago
20:27:53 openstackgerrit David Shrewsbury proposed openstack/openstacksdk master: Avoid dogpile.cache 0.7.0 https://review.openstack.org/624485
20:35:13 Shrews also, dtantsur|afk since that is affecting your configdrive change
20:48:18 Shrews mordred: we're doing a custom version of cache_on_arguments() so i wonder if the fault is ours?
20:48:39 Shrews mordred: educated guess at the commit that broke us: https://github.com/sqlalchemy/dogpile.cache/commit/7e8d9b0a2eb552adb6734c8570c63eb6ea2e3a7a
20:51:48 Shrews for the life of me, i cannot remember why we have this headache-inducing piece of code
21:02:38 Shrews ah ha: https://review.openstack.org/171456
21:03:40 Shrews might need to involve SpamapS
23:17:51 mordred Shrews: aWESOME
23:17:57 mordred kmalloc: ^^
#openstack-sdks - 2018-12-12
00:36:14 kmalloc Nice
00:37:50 kmalloc Heh
04:38:19 openstackgerrit Merged openstack/python-openstackclient master: Replace assertEqual(True/False, expr) with assertTrue/assertFalse https://review.openstack.org/617862
07:01:24 openstackgerrit Merged openstack/os-api-ref master: Fixing duplicate label issue. https://review.openstack.org/624400
13:07:30 openstackgerrit Merged openstack/python-openstackclient master: Supports router gateway IP QoS https://review.openstack.org/622130
13:42:06 openstackgerrit Monty Taylor proposed openstack/openstacksdk master: Avoid dogpile.cache 0.7.0 https://review.openstack.org/624485
13:42:18 mordred Shrews: ^^ added requirements patch and a depends-on
14:36:50 mordred kmalloc: whence you awaken, if you have any spare cycles today ...
14:37:18 mordred kmalloc: something is broken with ksa discovery caching (I think)
14:38:45 mordred kmalloc: you can see it by using openstacksdk like "import openstack ; openstack.enable_logging(http_debug=True) ; conn=openstack.connect(cloud='some cloud') ; print(conn.compute.servers()) ; print(conn.compute.servers())" and note that multiple discover http calls get made
14:57:57 kmalloc Will look as soon as I coffee
14:58:18 kmalloc I have cycles for this for sure
15:24:48 Shrews mordred: well, the thing is, i'm not convinced that we aren't doing something weird/unexpected that's causing the failure
15:29:03 mordred Shrews: oh - totally. I'm not sure dogpile itself is broken - but for now the combo of sdk and dogpile is - so we should go ahead and block it
15:29:09 mordred until we can figure it out
15:29:36 mordred Shrews, kmalloc: the ksa discovery cache issue is separate from the dogpile cache issue fwiw
15:29:53 mordred the discovery cache issue isn't resulting in anything breaking - which is why I think we haven't noticed it
15:30:02 mordred it's just resulting in extra discovery calls
15:30:06 Shrews oh, there is a new release today of dogpile.cache...
15:30:12 mordred yay!
15:31:18 Shrews https://github.com/sqlalchemy/dogpile.cache/commit/0ef761550ef623506f3cd050b4120e78193d1780
15:31:46 dtantsur looks like the fix
15:33:41 Shrews dtantsur: i just rechecked your change to verify if that fixes it for us
15:34:19 dtantsur Shrews: I suspect the new version is not in u-c yet, is it?
15:36:18 Shrews hrm, the PR seems to be a bit different from what we experienced https://github.com/sqlalchemy/dogpile.cache/issues/139
15:36:27 Shrews dtantsur: oh, probably not
15:45:45 Shrews mordred: dtantsur: tested locally, that does NOT appear to fix us :(
15:47:13 dtantsur :(
15:49:31 mordred :(
16:00:57 Shrews aaaaand now our nodepool src test is failing b/c it cannot find a python package. what a fun day
16:05:17 Blotis test
16:51:26 openstackgerrit Merged openstack/python-openstackclient master: Use devstack functional base job https://review.openstack.org/526117
16:53:18 Shrews mordred: i'm wondering if we even need our custom version of cache_on_arguments() anymore. for one, we don't use the 'resource' arg. for another, the original need for it (according to https://review.openstack.org/171456) was because of needing to set the invalidate func, but it looks to me that dogpile.cache always does that for us
16:54:02 Shrews mordred: we tried the per-resource thing, but ended up reverting it: https://review.openstack.org/362901
16:54:47 Shrews maybe i'll try a patch that eliminates our version and see what happens
16:59:54 mordred Shrews: yah - I really want to go back and revive that per-resource thing - I was thinking once we have the sdk and shade layers tied together a bit better, we could potentially do the per-resource sharing in the Resource layer
17:00:27 mordred Shrews: but ... I think maybe just making our current version simplier if it works is probably a great idea
17:05:39 openstackgerrit David Shrewsbury proposed openstack/openstacksdk master: WIP: test using standard cache method https://review.openstack.org/624767
17:06:26 Shrews that ^^^ is probably going to fail spectacularly, but should be fun to watch
17:09:24 Shrews oh, yeah. that's all wrong
17:13:27 Shrews that's a whole lot more complex than i initially thought, so nm :/
17:49:06 edleafe elmiko: Here's a first pass at the blurb for the OSF: http://paste.openstack.org/show/737157/
20:03:03 kmalloc mordred: haven't gotten to that sdk issue yet
20:03:29 kmalloc mordred: on my todo, some dr stuff came up and needed quick attention
20:03:47 kmalloc i'll probably have some kind of direction with that and the other dogpile bits by tomorrow.
20:06:47 Shrews kmalloc: oh, you switched irc nicks on me at some point
20:07:16 Shrews things begin to make sense
20:10:10 kmalloc Shrews: yes... like... a long while ago i swapped nics :P
20:10:56 Shrews i may not be quick, but i sure am slow
20:11:07 morgan Shrews: better?
20:11:09 morgan Shrews: :P
20:11:26 Shrews morgan: i think you are having an identity crisis
20:12:23 morgan nah, just /nick was not letting me go to morgan from kmalloc... oddly
20:24:02 morgan anyway...
20:26:50 mriedem does ksa impose any kind of http response timeout?
20:27:47 mriedem hmm https://github.com/openstack/keystoneauth/blob/ccf6cb79033b2083d9177823094f7836eb68ae0d/keystoneauth1/session.py#L248
21:20:51 morgan mriedem: hmm
21:21:04 morgan mriedem: not by default
21:26:48 mriedem i saw the --timeout option in the CLI defaults to 600
21:26:52 mriedem but that's still more than plenty
21:34:39 morgan ah yeah
21:34:51 morgan keystoneauth is very liberal in the "we don't force timeouts"
21:35:00 morgan but consumers of KSA aren't so easy going :P
21:35:04 morgan but 600s is a long timeout
#openstack-sdks - 2018-12-13
05:50:51 openstackgerrit Rajat Dhasmana proposed openstack/python-openstackclient master: Fix: Restore operation returns 'VolumeBackupsRestore' object is not iterable https://review.openstack.org/624860
07:44:08 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624873
07:47:57 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624874
07:53:00 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624875
07:59:33 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624879
08:03:11 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624880
08:04:05 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624881
08:04:13 openstackgerrit Rajat Dhasmana proposed openstack/python-openstackclient master: Fix: Restore operation returns 'VolumeBackupsRestore' object is not iterable https://review.openstack.org/624860
08:08:20 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624883
08:09:57 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624884
08:11:51 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624886
08:12:50 openstackgerrit machunnan proposed openstack/js-openstack-lib master: change http:// to https:// https://review.openstack.org/624887

Earlier   Later