| openstackgerrit | Tim Burke proposed openstack/swift master: py3: fix non-ascii metadata handling in account-server https://review.opendev.org/672610 | 00:01 |
|---|---|---|
| *** jistr has quit IRC | 00:15 | |
| *** jistr has joined #openstack-swift | 00:15 | |
| timburke | wrote up https://bugs.launchpad.net/swift/+bug/1837819 for the patch i mentioned in the meeting today | 00:37 |
| openstack | Launchpad bug 1837819 in OpenStack Object Storage (swift) "Overloaded object primaries cause 404s on GET" [Undecided,New] | 00:37 |
| timburke | and https://bugs.launchpad.net/swift/+bug/1837805 for the py3 account meta problem | 00:37 |
| openstack | Launchpad bug 1837805 in OpenStack Object Storage (swift) "py3: account metadata handling is busted" [High,Confirmed] | 00:37 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Ignore 404s from handoffs for objects when calculating quorum https://review.opendev.org/672186 | 00:41 |
| *** gyee has quit IRC | 00:49 | |
| *** tdasilva has quit IRC | 01:20 | |
| *** tdasilva has joined #openstack-swift | 01:21 | |
| *** ChanServ sets mode: +v tdasilva | 01:21 | |
| *** mvkr_ has quit IRC | 01:41 | |
| *** mvkr_ has joined #openstack-swift | 01:54 | |
| zaitcev | Already fixed?! I just went to IKEA and back real quick... | 02:11 |
| *** psachin has joined #openstack-swift | 03:38 | |
| *** rcernin has quit IRC | 03:55 | |
| *** gkadam has joined #openstack-swift | 04:08 | |
| *** gkadam has quit IRC | 04:09 | |
| *** pcaruana has joined #openstack-swift | 04:44 | |
| *** pcaruana has quit IRC | 04:56 | |
| *** rcernin has joined #openstack-swift | 05:33 | |
| *** m75abrams has joined #openstack-swift | 05:50 | |
| *** rcernin has quit IRC | 06:03 | |
| *** joeljwright has quit IRC | 06:06 | |
| *** rcernin has joined #openstack-swift | 06:19 | |
| *** pcaruana has joined #openstack-swift | 06:21 | |
| *** rcernin has quit IRC | 06:21 | |
| *** rcernin has joined #openstack-swift | 06:21 | |
| -openstackstatus- NOTICE: The git service on opendev.org is currently down. | 06:51 | |
| *** ChanServ changes topic to "The git service on opendev.org is currently down." | 06:51 | |
| *** rcernin has quit IRC | 07:06 | |
| *** tesseract has joined #openstack-swift | 07:15 | |
| *** mikecmpbll has joined #openstack-swift | 07:26 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift master: auto-sharding: send shard-ranges via container UPDATE https://review.opendev.org/672650 | 07:33 |
| *** tkajinam has quit IRC | 07:53 | |
| *** tkajinam has joined #openstack-swift | 07:53 | |
| -openstackstatus- NOTICE: Services at opendev.org like our git server and at openstack.org are currently down, looks like an outage in one of our cloud providers. | 08:34 | |
| *** ChanServ changes topic to "Services at opendev.org like our git server and at openstack.org are currently down, looks like an outage in one of our cloud providers." | 08:35 | |
| *** tkajinam has quit IRC | 08:36 | |
| *** ChanServ changes topic to "OpenStack Swift object storage | Logs: http://eavesdrop.openstack.org/irclogs/%23openstack-swift/ | Meetings: https://wiki.openstack.org/wiki/Meetings/Swift | Priority Reviews: https://wiki.openstack.org/wiki/Swift/PriorityReviews | Ideas: https://wiki.openstack.org/wiki/Swift/ideas" | 08:42 | |
| -openstackstatus- NOTICE: The problem in our cloud provider has been fixed, services should be working again | 08:42 | |
| *** dsariel has joined #openstack-swift | 08:45 | |
| *** dsariel has quit IRC | 09:24 | |
| *** e0ne has joined #openstack-swift | 09:32 | |
| *** joeljwright has joined #openstack-swift | 10:48 | |
| *** ChanServ sets mode: +v joeljwright | 10:48 | |
| *** baojg has quit IRC | 11:17 | |
| *** pcaruana has quit IRC | 11:42 | |
| *** pcaruana has joined #openstack-swift | 12:22 | |
| *** mvkr_ has quit IRC | 13:17 | |
| *** mvkr_ has joined #openstack-swift | 13:58 | |
| *** baojg has joined #openstack-swift | 14:12 | |
| *** ccamacho has joined #openstack-swift | 14:32 | |
| *** ccamacho has quit IRC | 14:53 | |
| timburke | good morning | 15:00 |
| timburke | zaitcev, between here and train, py3 bugs (and *particularly* data format bugs!) are automatic high/critical ;-) | 15:01 |
| *** m75abrams has quit IRC | 15:12 | |
| *** e0ne has quit IRC | 15:23 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Ignore 404s from handoffs for objects when calculating quorum https://review.opendev.org/672186 | 15:39 |
| *** gyee has joined #openstack-swift | 15:44 | |
| *** altlogbot_0 has quit IRC | 15:48 | |
| *** tesseract has quit IRC | 15:50 | |
| *** altlogbot_0 has joined #openstack-swift | 15:50 | |
| *** henriqueof has joined #openstack-swift | 15:55 | |
| *** henriqueof has quit IRC | 16:21 | |
| *** mikecmpbll has quit IRC | 16:35 | |
| clayg | tdasilva: so a big problem with trying to move towards "use the quoted etag" instead of "use the manifest etag" is DLO | 16:36 |
| clayg | SLO stores the ETAG on the object - so the only real problem is how to "find" it with the _resolve_etag_is_at function | 16:37 |
| clayg | with DLO we'd have to install a callback to *calculate* the ETag during the _resolve function 😬 | 16:38 |
| clayg | still, it'd be a pretty sweet general mechinism 🤔 | 16:38 |
| tdasilva | clayg: the alternative for DLO is it store the manifest etag which is empty string md5, right?? | 16:40 |
| clayg | sure - but I don't want DLO and SLO to be different | 16:40 |
| clayg | i mean as far as which etag you use for hardlinks | 16:40 |
| clayg | that would be even worse for the user | 16:40 |
| clayg | i think... | 16:41 |
| tdasilva | sorry, i agree, i was just trying to remember what was the "previous" behavior (i.e., using the manifest etag" | 16:41 |
| tdasilva | the manifest etag of a DLO is the empty string md5sum | 16:41 |
| clayg | also a hardlink that only works if the DLO quoted etag is static somehow adds a consistency checksum, which is sort of ... weird? interesting? useful? | 16:42 |
| clayg | tdasilva: yeah, but I feel like "the md5 of the empty string" part is sort of an implementation detail - in either *LO case you just use the etag of the real ondisk object as returned with multipart-manifest=get | 16:43 |
| clayg | it's a simple straight forward story at the least... it's also what I have implemented ... but that's probably also beside the point | 16:44 |
| clayg | I'm just not sure!!! APIs are hard 😢 | 16:44 |
| tdasilva | clayg: was just thinking that hardlinks with a "real" etag for DLOs are pretty useless then, true? | 16:46 |
| tdasilva | s/with/without | 16:46 |
| tdasilva | i'm trying to think of what impact that would have... | 16:49 |
| tdasilva | i mean, would we even version a DLO object? i don't think we version objects on POSTs, do we? I forget... | 16:50 |
| clayg | we do not version objects on POST, no | 16:51 |
| tdasilva | https://github.com/openstack/swift/blob/master/swift/common/middleware/versioned_writes.py#L64,L66 | 16:51 |
| tdasilva | doesn't matter, user could create a new manifest with a PUT, and we would version that... :shrugs | 16:52 |
| clayg | I don't think I have any opinon on the utility of hardlinks to DLOs with our without manifest or quoted etags ... | 16:53 |
| tdasilva | DLO and versioning is super weird, at least with SLO we have a manifest of the segments that make up the object | 16:54 |
| clayg | but if the use-case for DLOs really is that they are dynamic - and for whatever reason you wanted to have a hardlink pointing at one (e.g. versioned writes want have a hardlink pointed at one) then you'd kinda want your hardlink not to break if the listing changes... I think? | 16:54 |
| tdasilva | right, that's what i mean...there are really no guarantees of maintaining versions of a DLO object...i don't think.... | 16:56 |
| zaitcev | timburke: I'll make sure to review patch 672610 today | 17:00 |
| patchbot | https://review.opendev.org/#/c/672610/ - swift - py3: fix non-ascii metadata handling in account-se... - 1 patch set | 17:00 |
| timburke | clayg, tdasilva this make me think of https://bugs.launchpad.net/swift/+bug/1612991 ... | 17:08 |
| openstack | Launchpad bug 1612991 in OpenStack Object Storage (swift) "Post-as-copy does not allow a DLO manifest to be updated" [Critical,Fix released] - Assigned to Janie Richling (jrichli) | 17:08 |
| clayg | didn't we rip out post-as-copy? | 17:09 |
| clayg | oh, well it's already fixed anyway | 17:09 |
| clayg | so but now fast-post can update x-object-manifest or does it bleed through like sysmeta? | 17:10 |
| timburke | in the bug i laid out three paths forward -- one of them would've only allowed updates on PUT like we do for symlinks | 17:10 |
| timburke | we went with allowing updates on POST instead | 17:11 |
| timburke | in part becaues there may *already be* objects like that out in the wild | 17:11 |
| timburke | come to think of it... how do dlo and symlink interact today, on master? can a symlink have x-object-manifest set on it? i suppose it must be allowed... so you can get very different responses depending on whether you ask for obj, obj?symlink=get, or obj?symlink=get&multipart-manifest=get ?? | 17:13 |
| tdasilva | wow, that sounds awful | 17:15 |
| tdasilva | timburke: clayg: fyi...I started to write down a bit of a summary of what I'm learning about s3 versioning and object expiration behavior: https://etherpad.openstack.org/p/swift-s3-versioning-expiration | 17:19 |
| timburke | thanks tdasilva! | 17:20 |
| tdasilva | so the information I gave yesterday in the meeting is thus old, buckets can be versioned and have an expiration policy | 17:20 |
| timburke | and it looks like we might want versioned_writes in the object-expirer pipeline? maybe? | 17:22 |
| timburke | i wonder if "Amazon S3 retains the current version as a noncurrent version by adding a delete marker, which then becomes the current version" applies for null versions (ie, objects written before versioning was enabled), too... | 17:23 |
| tdasilva | yeah, i believe it does....the reason I say this is because even after it "versions" a "null version id object", it doesn't give it a version id, but it still keeps that object | 17:25 |
| timburke | 👍 | 17:35 |
| *** mvkr_ has quit IRC | 17:35 | |
| clayg | yeah seems like whoever is processing bucket policy expiration is going to want to interact with versioning settings - or at a minimum require an understanding of what versioned writes thinks of as "current" to implement a NoncurrentVersionExpiration setting | 17:38 |
| *** henriqueof has joined #openstack-swift | 17:38 | |
| clayg | but I'm not sure that the same thing that handles bucket policy expiration is the object-expirer and the x-delete-at header | 17:41 |
| clayg | i mean... I kinda feel like it's probably *not* the x-delete-at header, and i'm just not sure if the object-expirer or not | 17:42 |
| *** psachin has quit IRC | 17:50 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: Fix title-casing in HeaderKeyDict https://review.opendev.org/672803 | 18:09 |
| timburke | zaitcev, if you're already thinking about py3 and metadata... ^^^ would be good, too | 18:10 |
| zaitcev | timburke: ok | 18:10 |
| timburke | thanks | 18:10 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: Fix up requests so we can send non-RFC-compliant headers on py3 https://review.opendev.org/672808 | 18:38 |
| timburke | ^^^ a step toward properly addressing https://review.opendev.org/#/c/645388/12/test/functional/test_account.py@722 | 18:38 |
| patchbot | patch 645388 - swift - py3: Cover account/container func tests - 12 patch sets | 18:38 |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: Fix title-casing in HeaderKeyDict https://review.opendev.org/672803 | 18:59 |
| *** BjoernT has joined #openstack-swift | 18:59 | |
| clayg | of course, why would I think this would happen in middleware, trolololo https://github.com/openstack/swift/blob/master/swift/container/server.py#L564 | 19:32 |
| timburke | yup :-( | 19:33 |
| timburke | if only we'd had listing_formats before we wrote slo! | 19:34 |
| timburke | or... something. bleh. | 19:34 |
| timburke | i kinda really want to add a callback for listing dict manipulations -- it sucks that we have to re-write that in encryption, s3api, slo, symlinks, ... | 19:36 |
| *** e0ne has joined #openstack-swift | 19:39 | |
| *** dasp has quit IRC | 19:51 | |
| *** dasp has joined #openstack-swift | 19:54 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: Fix title-casing in HeaderKeyDict https://review.opendev.org/672803 | 19:55 |
| timburke | :-( crap like this is why i have such a viscerally-bad reaction to the idea of unpublishing artifacts: http://logs.openstack.org/08/672808/1/check/openstack-tox-lower-constraints/88e4d9e/job-output.txt.gz#_2019-07-25_18_58_34_500811 | 20:01 |
| timburke | not two weeks ago, swiftclient's lower-constraints job was *fine* http://logs.openstack.org/15/670215/3/gate/openstack-tox-lower-constraints/8e55ea2/job-output.txt.gz#_2019-07-12_02_02_02_856543 | 20:01 |
| *** gyee has quit IRC | 20:22 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Allow "harder" symlinks https://review.opendev.org/633094 | 20:23 |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: symlink-backed versioned_writes https://review.opendev.org/633857 | 20:23 |
| *** BjoernT has quit IRC | 20:55 | |
| *** gyee has joined #openstack-swift | 20:59 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: py3: fix non-ascii metadata handling in account-server https://review.opendev.org/672610 | 21:18 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: Fix up requests so we can send non-RFC-compliant headers on py3 https://review.opendev.org/672808 | 21:21 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient master: Drag forward prettytable in lower-constraints https://review.opendev.org/672830 | 21:21 |
| *** pcaruana has quit IRC | 21:28 | |
| *** e0ne has quit IRC | 21:56 | |
| *** henriqueof has quit IRC | 22:01 | |
| *** rcernin has joined #openstack-swift | 22:16 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Give internal clients the option of getting container HEADs from memcache https://review.opendev.org/668064 | 22:24 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!