| openstackgerrit | Tim Burke proposed openstack/swift master: Use get_container_info to check existence before container PUT https://review.opendev.org/580333 | 00:01 |
|---|---|---|
| *** openstack has joined #openstack-swift | 13:13 | |
| *** ChanServ sets mode: +o openstack | 13:13 | |
| *** zaitcev has joined #openstack-swift | 13:27 | |
| *** ChanServ sets mode: +v zaitcev | 13:27 | |
| *** rcernin has quit IRC | 13:31 | |
| *** jistr is now known as jistr|call | 13:31 | |
| *** pcaruana has quit IRC | 13:46 | |
| *** pcaruana has joined #openstack-swift | 13:47 | |
| *** zaitcev has quit IRC | 14:06 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: clarify s3api ACL test HEADs https://review.opendev.org/667933 | 14:10 |
| *** baojg has joined #openstack-swift | 14:14 | |
| *** zaitcev has joined #openstack-swift | 14:18 | |
| *** ChanServ sets mode: +v zaitcev | 14:18 | |
| *** zaitcev_ has joined #openstack-swift | 14:25 | |
| *** ChanServ sets mode: +v zaitcev_ | 14:25 | |
| *** zaitcev has quit IRC | 14:29 | |
| *** jistr|call is now known as jistr | 14:31 | |
| *** baojg has quit IRC | 14:59 | |
| *** pcaruana has quit IRC | 15:14 | |
| zaitcev_ | Guys, we can run probe tests on a larger cluster than SAIO, right? | 15:14 |
| *** zaitcev__ has joined #openstack-swift | 15:22 | |
| *** ChanServ sets mode: +v zaitcev__ | 15:22 | |
| *** zaitcev__ is now known as zaitcev | 15:24 | |
| *** zaitcev_ has quit IRC | 15:26 | |
| *** m75abrams has quit IRC | 15:43 | |
| *** tdasilva has joined #openstack-swift | 15:50 | |
| *** ChanServ sets mode: +v tdasilva | 15:50 | |
| *** ccamacho has quit IRC | 15:55 | |
| *** e0ne has quit IRC | 16:02 | |
| * clayg any idea what this is about: | 16:04 | |
| clayg | https://github.com/openstack/swift/blob/master/swift/common/middleware/s3api/acl_handlers.py#L220 | 16:04 |
| clayg | zaitcev: "lager" - i mean as long as it has exactly 4 nodes with 1 drive on each 🤣 | 16:05 |
| clayg | kota_: I'm trying to make acl checking use cached sysmeta when possible and running into a weird problem where my "create_bucket" is returning "bucket already owned by you" instead of 200OK | 16:06 |
| clayg | kota_: looking at the logs I can see the PUT request (which seems to work!) and then immediately I see a POST - it's burried down in some magic in the s3request.get_response - because the bucket controller method with the original PUT never returns? | 16:07 |
| clayg | FWIW the bucket does get created, w/o acl's ... I just don't know why I'm hitting this automatic POST request 😞 | 16:08 |
| *** zaitcev_ has joined #openstack-swift | 16:11 | |
| *** ChanServ sets mode: +v zaitcev_ | 16:11 | |
| *** zaitcev has quit IRC | 16:15 | |
| timburke | clayg, on the s3 acls: there's a PUT (which could fail with BucketAlreadyExists or BucketAlreadyOwnedByYou) followed by a POST that adds the acls. i think the concern there was that if the POST fails without talking to *any* backend servers, you'd get the create but no acls at all | 16:16 |
| timburke | so, exactly the problem you're running into ;-) | 16:17 |
| timburke | clayg, also, you can have two drives on each of the four nodes! otherwise we can't EC :P | 16:18 |
| *** zaitcev__ has joined #openstack-swift | 16:26 | |
| *** ChanServ sets mode: +v zaitcev__ | 16:26 | |
| *** zaitcev_ has quit IRC | 16:30 | |
| *** mikecmpbll has quit IRC | 16:35 | |
| *** zaitcev__ has quit IRC | 16:40 | |
| *** gkadam has quit IRC | 16:51 | |
| *** zaitcev__ has joined #openstack-swift | 16:52 | |
| *** ChanServ sets mode: +v zaitcev__ | 16:52 | |
| clayg | @timburke can you explain the extra PUT/POST request to me again? I'm trying to debug create bucket and I'm seeing a whole extra PUT/POST before MY PUT is even executed https://github.com/openstack/swift/blob/master/swift/common/middleware/s3api/controllers/bucket.py#L288 (and when it does it unsurprisingly sees a 202 and translates it to 409) | 17:10 |
| clayg | i guess I'll start backing out my changes and try to figre out how this EVER worked? | 17:11 |
| timburke | hmm... weird... | 17:12 |
| timburke | so *i think* the idea is that we don't want to overwrite any existing acls, so we do the PUT *first* and expect that it may fail because the bucket already exists (owned by us or someone else), then we do the POST iff we got a 201 | 17:14 |
| timburke | idk why there's be two PUTs tho | 17:14 |
| * clayg might be related to this, i'm not sure what's going on here: | 17:16 | |
| clayg | https://github.com/openstack/swift/blob/master/swift/common/middleware/s3api/s3request.py#L1554 | 17:16 |
| clayg | well, we're definately doing the PUT first - and it doesn't fail - it creates the container! then it does it POST - then when it does the original PUT to create the bucket it returns 409 | 17:17 |
| clayg | ok, that early return is definately related - i had pulled it out because I didn't want that interface to return a resp since it didn't always have to make a request anymore... | 17:23 |
| clayg | but apparently in this PUT/POST case it was returning that resp w/o making the original ... or something | 17:24 |
| clayg | anyway, unsurprisingly I broke it - i just don't understand that rationalization of what the existing code thought it was doing... | 17:24 |
| clayg | ok, i see what I did - i really just want to change the interface of _handle_acl - i can fix handle_acl | 17:38 |
| *** tdasilva has quit IRC | 17:53 | |
| *** pcaruana has joined #openstack-swift | 17:54 | |
| openstackgerrit | Tim Burke proposed openstack/swift feature/losf: Merge remote-tracking branch 'gerrit/master' into review/tim_burke/py3 https://review.opendev.org/668007 | 17:55 |
| *** mvkr has quit IRC | 18:03 | |
| *** rdejoux has quit IRC | 18:09 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Give ECAppIter greenthreads a chance to wrap up https://review.opendev.org/665773 | 18:20 |
| *** zaitcev__ is now known as zaitcev | 18:39 | |
| *** zaitcev has quit IRC | 19:10 | |
| *** zaitcev has joined #openstack-swift | 19:23 | |
| *** ChanServ sets mode: +v zaitcev | 19:23 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Put more information in the str() for InternalError https://review.opendev.org/668031 | 19:32 |
| *** pcaruana has quit IRC | 21:05 | |
| *** tesseract has quit IRC | 21:19 | |
| *** rcernin has joined #openstack-swift | 21:24 | |
| *** zaitcev has quit IRC | 21:30 | |
| *** zaitcev has joined #openstack-swift | 21:42 | |
| *** ChanServ sets mode: +v zaitcev | 21:42 | |
| *** zaitcev_ has joined #openstack-swift | 21:46 | |
| *** ChanServ sets mode: +v zaitcev_ | 21:46 | |
| *** zaitcev has quit IRC | 21:47 | |
| *** mvkr has joined #openstack-swift | 22:01 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP: Give operators the option of serving container HEADs from memcache https://review.opendev.org/668064 | 22:10 |
| timburke | ^^^ has some interesting improvements for s3api, which *needs* to issue HEADs and the like to give the auth system a chance to translate from access_key_id to account | 22:11 |
| timburke | downsides include things like func tests not passing because the versioned_writes tests rely on up-to-date object-counts in HEADs | 22:13 |
| *** zaitcev__ has joined #openstack-swift | 22:29 | |
| *** ChanServ sets mode: +v zaitcev__ | 22:29 | |
| *** zaitcev_ has quit IRC | 22:33 | |
| *** zaitcev__ is now known as zaitcev | 22:38 | |
| *** zaitcev has quit IRC | 23:18 | |
| *** zaitcev has joined #openstack-swift | 23:18 | |
| *** ChanServ sets mode: +v zaitcev | 23:18 | |
| mattoliverau | morning | 23:19 |
| *** joeljwright has quit IRC | 23:20 | |
| *** joeljwright has joined #openstack-swift | 23:26 | |
| *** ChanServ sets mode: +v joeljwright | 23:26 | |
| *** tonyb has quit IRC | 23:32 | |
| *** tonyb has joined #openstack-swift | 23:32 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP: Give operators the option of serving container HEADs from memcache https://review.opendev.org/668064 | 23:42 |
| timburke | mattoliverau, o/ | 23:44 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: Isolate docs requirements https://review.opendev.org/667477 | 23:48 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!