Tuesday, 2020-01-14

*** ianychoi_ has joined #openstack-swift00:09
*** ianychoi has quit IRC00:11
*** tkajinam has quit IRC00:56
*** tkajinam has joined #openstack-swift00:59
*** gyee has quit IRC01:34
*** tkajinam has quit IRC02:36
*** tkajinam has joined #openstack-swift03:02
*** psachin has joined #openstack-swift03:36
*** psachin has quit IRC04:09
*** evrardjp has quit IRC05:33
*** evrardjp has joined #openstack-swift05:34
*** pcaruana has quit IRC06:17
*** ccamacho has quit IRC07:00
*** tesseract has joined #openstack-swift07:41
*** Fengli has joined #openstack-swift07:44
*** rcernin has quit IRC07:52
*** rdejoux has joined #openstack-swift08:18
*** ccamacho has joined #openstack-swift08:20
*** tkajinam has quit IRC08:24
*** pcaruana has joined #openstack-swift08:56
*** ccamacho has quit IRC09:00
*** rpittau|afk is now known as rpittau09:16
*** ccamacho has joined #openstack-swift09:52
*** pcaruana has quit IRC10:23
*** Fengli1 has joined #openstack-swift10:53
*** Fengli has quit IRC10:55
*** Fengli1 is now known as Fengli10:55
*** pcaruana has joined #openstack-swift11:02
*** rpittau is now known as rpittau|bbl11:11
*** manuvakery has joined #openstack-swift11:29
*** Fengli has quit IRC11:46
*** Fengli has joined #openstack-swift11:47
*** diablo_rojo has joined #openstack-swift12:07
*** rpittau|bbl is now known as rpittau13:26
*** tkajinam has joined #openstack-swift13:56
*** manuvakery has quit IRC13:58
*** Fengli has quit IRC13:58
*** Fengli has joined #openstack-swift14:08
*** Fengli1 has joined #openstack-swift14:27
*** Fengli has quit IRC14:29
*** Fengli1 is now known as Fengli14:29
*** Fengli1 has joined #openstack-swift14:54
*** Fengli has quit IRC14:55
*** Fengli1 is now known as Fengli14:55
*** tkajinam has quit IRC15:10
*** ccamacho has quit IRC16:02
*** rdejoux has quit IRC16:11
*** theintern_ has joined #openstack-swift16:12
*** dosaboy has quit IRC16:28
*** gyee has joined #openstack-swift16:42
*** rdejoux has joined #openstack-swift16:58
*** rpittau is now known as rpittau|afk17:01
*** rdejoux has quit IRC17:04
*** rdejoux has joined #openstack-swift17:04
DHEhas anyone seen what I can only describe as hung service processes? they stand out on the process list for only having 0 or 1 second of CPU time despite running for weeks whereas every other process has way more CPU time.17:09
*** camelCaser has quit IRC17:13
*** camelCaser has joined #openstack-swift17:14
*** evrardjp has quit IRC17:33
*** evrardjp has joined #openstack-swift17:34
*** theintern_ has quit IRC17:46
*** dosaboy has joined #openstack-swift17:59
*** theintern_ has joined #openstack-swift18:15
*** theintern_ has quit IRC18:47
*** theintern_ has joined #openstack-swift18:59
*** pcaruana has quit IRC19:16
*** gmann is now known as gmann_afk19:36
*** theintern_ has quit IRC20:05
*** tesseract has quit IRC20:50
openstackgerritCorey Bryant proposed openstack/liberasurecode master: Fix strncpy to prevent truncation of source  https://review.opendev.org/70253020:58
claygtimburke: i think it would be weird to give "orphaned" containers any special treatment in listings - I don't think we really know for sure which way the split goes?21:02
*** Fengli1 has joined #openstack-swift21:02
*** Fengli has quit IRC21:05
*** Fengli1 is now known as Fengli21:05
timburkei don't like that the proxy knows which way it found the split but doesn't tell the user, is all21:05
timburkelike, if i think i delete everything in a container and delete the container... ok, i can chalk up the container still showing up in account listings because updater hasn't gotten to it yet21:05
*** rdejoux has quit IRC21:06
timburkebut if it's still there after a month... are updaters busted or did versioning stuff get out-of-sync? as a user, i don't know which path i should take to fix it21:06
timburke(arguably, as a user, i shouldn't even need to know about such things... but, y'know...)21:06
timburkewhat i *do* care about as a user is how i can make sure i'm not getting billed for this thing anymore21:07
*** rdejoux has joined #openstack-swift21:18
*** rdejoux has quit IRC21:24
openstackgerritClay Gerrard proposed openstack/python-swiftclient master: object versioning features  https://review.opendev.org/69187721:34
*** rcernin has joined #openstack-swift22:08
claygtimburke: maybe i'm just not clear on the corrective steps - a DELETE request to the container isn't going to indicate there's still versions?22:15
claygtimburke: because i feel like "{name: foo, objects: 0, bytes: 100}" *is* saying something about there still being something in there...22:16
timburkehmm.... that's a good point... objects:0 makes it rather odd...22:18
timburketime to update the probe test ;-)22:18
timburkei think i'd be content to say that "objects: 0, bytes: non-0, container 404s means you're in this state"22:19
timburkecorrective steps are: create container again with versioning enabled, list versions, delete them, delete container again22:20
timburke(at least, assuming everything really *was* supposed to be deleted22:20
*** mvkr has quit IRC22:21
claygyou can't just DO the versioned listing?22:22
timburkenope -- 40422:26
patchbotpatch 701643 - swift - squash: flesh out object-versioning probe test a bit - 1 patch set22:27
claygwell, that's less great22:28
timburkeeven if you know the version-ids you missed, you still can't interact with them; you get a 400 until you re-create the container, with versioning enabled22:28
claygi'd be much more inclined to add some kludge to infer that maybe there's a version container out there even if the named container 404s and try to improve that interaction than adding a new key of some kind to the public listing API22:32
clayglike if you GET foo?versions and foo returns 404 - maybe go ahead and GET \x00versions\x00foo JIC22:33
timburkeyeah, that seems reasonable22:33
timburkethat reminds me -- i'm gettimg more and more skeptical of the need to record the locations name...22:34
timburkelike, i guess better safe than sorry...22:34
timburkei think i'd feel a lot better if it was very clearly "informational only" and we didn't have any logic based on it22:35
timburke...which might get us pretty close to having listings Just Work and the like22:35
*** gmann_afk is now known as gmann22:53
*** tkajinam has joined #openstack-swift22:55
timburkerledisez, https://github.com/swiftstack/docker-swift/pull/48 reminds me -- would you have much concern if Swift increased its max_meta_name_length and max_meta_value_length to 2048 each? we've been pushing that default for a while now, seems to clear up some issues when s3 clients want to use library-provided client-side encryption23:07
timburkewe haven't run into any trouble with keeping max_meta_overall_size at 4k, fwiw23:07
*** mvkr has joined #openstack-swift23:27

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!