| openstackgerrit | Matthew Oliver proposed openstack/swift master: Add a grafana/statsd server to docker test tool https://review.openstack.org/577733 | 00:25 |
|---|---|---|
| mattoliverau | ^ notmyname rebased | 00:25 |
| notmyname | woot | 00:25 |
| mattoliverau | just docker-compose up and then browse to localhost:9000 to see grafana, setup the data source and you should see swift metrics (see the readme for default username and passwords (root/root and datasource/datasource)). | 00:27 |
| mattoliverau | though you might need to do a 'docker-compose build' to rebuild the images seeing as it adds the statsd config options to the swift server confs, because I'd assume you'd test on your existing test env :) | 00:28 |
| notmyname | naturally ;-) | 00:37 |
| *** two_tired has joined #openstack-swift | 01:26 | |
| kota_ | good morning | 01:28 |
| kota_ | thanks notmyname to change the dir name generic | 01:29 |
| mattoliverau | I say that because it took me a while to figure it out (/me is still new to this docker thing) :p | 01:29 |
| mattoliverau | kota_: morning | 01:29 |
| kota_ | mattoliverau: o/ | 01:29 |
| *** itlinux has joined #openstack-swift | 02:05 | |
| *** mrjk has quit IRC | 02:11 | |
| *** psachin` has joined #openstack-swift | 02:42 | |
| *** bkopilov has quit IRC | 02:43 | |
| *** bkopilov has joined #openstack-swift | 02:57 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift master: Add a grafana/statsd server to docker test tool https://review.openstack.org/577733 | 02:57 |
| notmyname | kota_: it was a great suggestion, thanks | 03:16 |
| *** two_tired has quit IRC | 03:24 | |
| *** mrjk has joined #openstack-swift | 03:35 | |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: Replace MIME with PUT+POST for EC and Encryption https://review.openstack.org/427911 | 05:03 |
| *** ccamacho has quit IRC | 05:06 | |
| *** vpc has joined #openstack-swift | 05:14 | |
| vpc | hi | 05:14 |
| *** lifeless has joined #openstack-swift | 05:16 | |
| *** cshastri has joined #openstack-swift | 05:22 | |
| *** cshastri has quit IRC | 05:31 | |
| *** cshastri has joined #openstack-swift | 05:47 | |
| openstackgerrit | Merged openstack/swift master: reorganized existing ansible playbooks https://review.openstack.org/577466 | 05:47 |
| *** cbartz has joined #openstack-swift | 05:57 | |
| *** openstackgerrit has quit IRC | 06:04 | |
| *** ccamacho has joined #openstack-swift | 06:09 | |
| *** vpc has quit IRC | 06:15 | |
| *** cshastri has quit IRC | 06:25 | |
| *** openstackgerrit has joined #openstack-swift | 06:38 | |
| openstackgerrit | Nguyen Hai proposed openstack/swift master: add lower-constraints job https://review.openstack.org/556255 | 06:38 |
| *** cshastri has joined #openstack-swift | 06:38 | |
| openstackgerrit | Nguyen Hai proposed openstack/swift master: add lower-constraints job https://review.openstack.org/556255 | 06:38 |
| *** psachin` has quit IRC | 06:45 | |
| *** psachin` has joined #openstack-swift | 06:47 | |
| *** gkadam has joined #openstack-swift | 06:58 | |
| *** neonpastor has quit IRC | 07:00 | |
| *** neonpastor has joined #openstack-swift | 07:01 | |
| *** rcernin has quit IRC | 07:08 | |
| *** tesseract has joined #openstack-swift | 07:14 | |
| *** pcaruana has joined #openstack-swift | 07:20 | |
| acoles | good morning | 07:40 |
| *** geaaru has joined #openstack-swift | 07:52 | |
| *** cbartz has quit IRC | 08:06 | |
| *** mikecmpbll has joined #openstack-swift | 08:10 | |
| *** cbartz has joined #openstack-swift | 08:25 | |
| *** psachin` is now known as psachin | 08:39 | |
| *** gkadam_ has joined #openstack-swift | 08:44 | |
| *** gkadam has quit IRC | 08:45 | |
| *** gkadam_ has quit IRC | 08:45 | |
| *** gkadam_ has joined #openstack-swift | 08:45 | |
| *** gkadam__ has joined #openstack-swift | 08:48 | |
| *** gkadam_ has quit IRC | 08:51 | |
| *** hseipp has joined #openstack-swift | 08:52 | |
| *** mvk has quit IRC | 08:57 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift master: Add ceph-s3 test non-voting job https://review.openstack.org/564700 | 09:00 |
| kota_ | quick update but I'm not sure the patch works fine. I'll be back there tomorrow morning | 09:01 |
| *** mikecmpb_ has joined #openstack-swift | 09:07 | |
| *** mikecmpbll has quit IRC | 09:09 | |
| *** lifeless has quit IRC | 09:16 | |
| *** mvk has joined #openstack-swift | 09:24 | |
| *** mvk has quit IRC | 10:05 | |
| *** mvk has joined #openstack-swift | 10:21 | |
| openstackgerrit | Nguyen Hai proposed openstack/swift master: add lower-constraints job https://review.openstack.org/556255 | 10:49 |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Add keymaster to fetch root secret from KMIP service https://review.openstack.org/578075 | 11:52 |
| openstackgerrit | Alistair Coles proposed openstack/swift master: Add keymaster to fetch root secret from KMIP service https://review.openstack.org/578075 | 11:54 |
| *** kei_yama has quit IRC | 13:12 | |
| *** frickler has quit IRC | 13:26 | |
| *** frickler has joined #openstack-swift | 13:26 | |
| *** klamath has joined #openstack-swift | 13:35 | |
| *** cshastri has quit IRC | 14:10 | |
| *** psachin has quit IRC | 14:28 | |
| *** itlinux has quit IRC | 14:29 | |
| *** ccamacho has quit IRC | 14:35 | |
| *** ccamacho has joined #openstack-swift | 14:35 | |
| *** cbartz has quit IRC | 15:02 | |
| *** ccamacho has quit IRC | 15:15 | |
| *** itlinux has joined #openstack-swift | 15:16 | |
| *** cwright has quit IRC | 15:31 | |
| *** pcaruana has quit IRC | 15:42 | |
| notmyname | good morning | 15:46 |
| *** cwright has joined #openstack-swift | 15:48 | |
| *** gyee has joined #openstack-swift | 15:50 | |
| *** gyee has quit IRC | 16:03 | |
| *** hseipp has quit IRC | 16:27 | |
| timburke | good morning | 16:30 |
| *** silor has joined #openstack-swift | 16:35 | |
| *** tesseract has quit IRC | 16:40 | |
| *** bharath1234 has joined #openstack-swift | 16:42 | |
| *** bharath1234 has quit IRC | 16:43 | |
| *** mikecmpb_ has quit IRC | 16:52 | |
| *** mvk has quit IRC | 17:17 | |
| *** gyee has joined #openstack-swift | 17:20 | |
| *** mikecmpbll has joined #openstack-swift | 17:41 | |
| *** geaaru has quit IRC | 17:41 | |
| clayg | timburke: so I'm trying to reconsider patch 337960 in light of like basically 2-3 people all telling me we'd be better/happier with the change | 18:49 |
| patchbot | https://review.openstack.org/#/c/337960/ - swift - Send correct SLO ETag for container updates | 18:49 |
| timburke | imagine how much happier we would be if it had gone in two years ago ;-P | 18:49 |
| clayg | I think I might be over-valuing the existing api? maybe EVERYONE that's ever noticed this difference has concluded it's wrong, and the change would only effect the smallest possible subset of swift clients who have all basically vouched they care more about getting it changed than any concern worry for old information | 18:50 |
| clayg | timburke: I'm not sure I will be any happier - ya'll might have been happier? why are y'all so unhappy? don't be unhappy - life is grand. | 18:51 |
| timburke | it's just weird -- you do a PUT, get back an ETag. subsequent GETs return the same ETag. check the container listing... get a different, not-terribly-useful ETag ;-/ | 18:54 |
| timburke | it's kinda tolerable with a DLO, since you just accept that container listings are going to be worthless. but with SLOs, the content-length is good! it's just the ETag that's trash | 18:55 |
| clayg | timburke: isn't there some sort of logical falicy that has to do with side-stepping the issue by adding evidence for a tangenial point? | 18:56 |
| clayg | "this api is better" - "I'm not sure we should change an existing api, and besides what about existing data" - "I don't think you understand the word BETTER" ;) | 19:00 |
| clayg | anyway - we probaby both feel like we're beating a dead horse - notmyname you have any ideas about this specific patch? | 19:00 |
| clayg | I think there's lots of unexplored options to track/expose the SLO-etag in the container db's and get it expose to clients that don't include "start writing it in this place which has historically been where we write the value that matches etag with ?multipart-manifest=get" | 19:01 |
| timburke | but we've fixed stupid APIs before (admittedly, only after i brow-beat you into it) | 19:01 |
| timburke | you'd have a hard time convincing me that https://github.com/openstack/swift/commit/94bac4a was a *bad* idea | 19:02 |
| clayg | timburke: is it correct that with the quotes you CAN tell if an entry in the litsing is a multipart-manifest=get etag vs a slo-etag? | 19:03 |
| clayg | @timburke but that change had an escape hatch? | 19:03 |
| clayg | besides that's basically what I'm advocating for - track both - let the client choose | 19:03 |
| timburke | i could probably do something like store "<old etag>; slo_etag=<slo etag>" in the container db and have slo split it out as a separate field, like what i'm doing in https://review.openstack.org/#/c/575860/ ... | 19:04 |
| patchbot | patch 575860 - swift - Include '-' in multipart ETags | 19:04 |
| timburke | and then have *both* of them look for pre-existing X-Object-Sysmeta-Container-Update-Override-Etag values... | 19:06 |
| timburke | it's gonna be a little weird, though -- the closest thing to a sane way to do it that i can think of involves s3api sending <md5 of put manifest>; s3_etag=<multipart-upload etag>, then slo just grabbing the params from the header and sending <md5 of on-disk manifest>; slo_etag=<slo etag>; s3_etag=<multipart-upload etag> | 19:09 |
| notmyname | clayg: I need to catch up on the conversation and digest | 19:13 |
| openstackgerrit | John Dickinson proposed openstack/swift master: added docker test target tools https://review.openstack.org/577467 | 19:14 |
| clayg | timburke: well that sounds good tho? | 19:14 |
| timburke | hmm... or maybe most of https://review.openstack.org/#/c/575860/ should just move to slo? but then there'd be this other place where S3-knowledge leaks out... | 19:15 |
| patchbot | patch 575860 - swift - Include '-' in multipart ETags | 19:15 |
| clayg | too much work? maybe the <md5 of put manifest> is useless and the the existing listing records are sunk cost? | 19:15 |
| clayg | we could not call it s3_etag - then it's just "the slo_etag" which is apparently useful to capture in the listings for 1space and s3api - possibly other clients? | 19:16 |
| timburke | <md5 of put manifest> is decidedly useless, and decidedly not the right thing to send to container servers. which makes it weird that it's the most-sensible thing for s3api to send | 19:16 |
| clayg | timur: timburke: torgomatic: I think you three should form a cabal - get kota_ on board and then just have notmyname use the PTL hammer to override me | 19:17 |
| timburke | no, *i want both* -- S3 clients expect that '-' in listing etags, and it seems perfectly reasonable that we should expose the etag that a swift client would receive on GET, too | 19:18 |
| clayg | timburke: i'm probably confused - i thought you want only one - the slo-etag? the one you get when you HEAD w/o multipart-manifest=GET? | 19:21 |
| clayg | timburke: you said "<md5 of put manifest> is decidedly useless" | 19:22 |
| clayg | you mean you want slo-etag and s3-etag (which is slo-etag + number of segments) | 19:23 |
| clayg | I think the swift container listing API could grow as many new keys as we want "etag_segments" & "num_segments" would both be great additions to the existing api | 19:24 |
| clayg | anyway, i was going to add the patch to the meeting for tomorrow - but I won't be there | 19:25 |
| timburke | <md5 of put manifest> has nothing to do with anything stored anywhere, and doesn't deserve to be on-disk in the container server. because slo translates most of the client keys to some other format (see also https://github.com/openstack/swift/commit/e1f1296) | 19:27 |
| timburke | s3-etag is not just slo-etag + number of segments -- there's a difference in whether the concatenated etags are hex-encoded (swift) or not (s3) | 19:27 |
| timburke | what i really want is for `s3cmd sync` to stop doing useless, expensive work every time there's a multipart upload | 19:27 |
| *** silor1 has joined #openstack-swift | 19:34 | |
| *** silor has quit IRC | 19:35 | |
| *** silor1 is now known as silor | 19:35 | |
| timburke | why did we add the env marker to dnspython in https://github.com/openstack/swift/commit/9a865f8 ? https://github.com/openstack/requirements/commit/7e61e83 seemed to indicate that it should work on both py2 and py3... | 20:07 |
| timburke | i guess designate had some troubles with 1.14.0 on py3? https://github.com/openstack/requirements/commit/1ecbcd8 | 20:08 |
| timburke | i mean, gate passed on https://review.openstack.org/#/c/560111/ -- and that must've come after https://review.openstack.org/#/c/540190/ ... but then why am i seeing `pkg_resources.DistributionNotFound: The 'dnspython>=1.14.0' distribution was not found and is required by swift` when i run tox -e py36? hrm... | 20:12 |
| patchbot | patch 560111 - swift - update requirements to support matching global exc... (MERGED) | 20:12 |
| patchbot | patch 540190 - swift - py3: port common/wsgi.py (MERGED) | 20:12 |
| *** silor has quit IRC | 20:19 | |
| *** mvk has joined #openstack-swift | 20:47 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: port healthcheck https://review.openstack.org/577574 | 21:20 |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: port crossdomain https://review.openstack.org/577577 | 21:20 |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: port domain_remap tests https://review.openstack.org/577578 | 21:20 |
| openstackgerrit | Tim Burke proposed openstack/swift master: swob: Fix up some WSGI string business https://review.openstack.org/578210 | 21:20 |
| torgomatic | timburke: thanks for splitting those off | 21:32 |
| timburke | np -- sorry that the signature stuff is probably going to require more thought | 21:33 |
| timburke | fwiw, i'm still working on raising a TypeError if you try to pass unicode as a swob.Response body | 21:34 |
| timburke | it's ... messy ... which makes me think that it's kinda important that i do it sooner rather than later :-) | 21:36 |
| *** lifeless has joined #openstack-swift | 21:38 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift master: object-replicator: separate workers for primary/handoffs https://review.openstack.org/578213 | 21:41 |
| torgomatic | timburke: yeah, I didn't put much thought into the whole process; I pretty much just ran the tests, fixed things that they failed with, and called it a day | 21:41 |
| torgomatic | I'd have liked to functionally test it, but that's not really possible at the moment | 21:41 |
| timburke | yep :-( | 21:42 |
| timburke | there were a couple different paths toward getting a proxy server that could actually start under py3, but the responses were all garbage so none was even to the point of being helpful for that... | 21:43 |
| *** mikecmpbll has quit IRC | 21:44 | |
| torgomatic | I'm not super convinced that middleware is the best place to start porting stuff, but the modules tend to be somewhat self-contained so it's easy to get their tests working | 21:46 |
| *** rcernin has joined #openstack-swift | 21:47 | |
| torgomatic | and I suppose porting *something* is better than porting *nothing* | 21:47 |
| timburke | and getting middleware to a sane place means that we can start to have some confidence about our inputs/outputs when trying to port the proxy | 21:48 |
| *** itlinux has quit IRC | 21:54 | |
| *** afazekas has quit IRC | 21:56 | |
| *** afazekas has joined #openstack-swift | 21:59 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Have yield_suffixes just take a partition_path https://review.openstack.org/578221 | 22:15 |
| openstackgerrit | Samuel Merritt proposed openstack/swift master: object-replicator: separate workers for primary/handoffs https://review.openstack.org/578213 | 22:19 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Log unexpected errors when trying to clean up empty dirs https://review.openstack.org/578224 | 22:23 |
| *** threestrands has joined #openstack-swift | 22:38 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: swob: Stop auto-encoding unicode bodies https://review.openstack.org/578227 | 22:57 |
| mattoliverau | morning | 23:38 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!