| *** ianychoi has joined #openstack-swift | 00:01 | |
| *** gyee has quit IRC | 00:15 | |
| *** NM has joined #openstack-swift | 00:37 | |
| *** NM has quit IRC | 00:53 | |
| *** NM has joined #openstack-swift | 00:57 | |
| *** NM has quit IRC | 01:24 | |
| *** irclogbot_3 has quit IRC | 03:01 | |
| *** irclogbot_0 has joined #openstack-swift | 03:02 | |
| *** rcernin has quit IRC | 03:15 | |
| *** psachin has joined #openstack-swift | 03:20 | |
| *** rcernin has joined #openstack-swift | 03:28 | |
| *** rcernin has quit IRC | 03:57 | |
| *** rcernin has joined #openstack-swift | 04:01 | |
| *** _d34dh0r53_ has joined #openstack-swift | 04:29 | |
| *** tkajinam has quit IRC | 05:00 | |
| *** tkajinam has joined #openstack-swift | 05:07 | |
| *** psachin has quit IRC | 05:49 | |
| *** psachin has joined #openstack-swift | 06:05 | |
| *** pcaruana has joined #openstack-swift | 06:41 | |
| *** e0ne has joined #openstack-swift | 06:45 | |
| *** e0ne has quit IRC | 07:21 | |
| *** rdejoux has joined #openstack-swift | 07:33 | |
| *** e0ne has joined #openstack-swift | 07:36 | |
| *** e0ne has quit IRC | 07:52 | |
| *** ccamacho has joined #openstack-swift | 07:53 | |
| *** e0ne has joined #openstack-swift | 07:59 | |
| *** e0ne has quit IRC | 08:55 | |
| *** e0ne has joined #openstack-swift | 09:00 | |
| *** rcernin has quit IRC | 09:30 | |
| *** e0ne has quit IRC | 09:58 | |
| *** e0ne has joined #openstack-swift | 10:03 | |
| *** e0ne has quit IRC | 10:28 | |
| *** pcaruana has quit IRC | 10:31 | |
| *** mvkr has joined #openstack-swift | 11:23 | |
| *** pcaruana has joined #openstack-swift | 11:36 | |
| *** e0ne has joined #openstack-swift | 11:37 | |
| *** dhellmann has left #openstack-swift | 12:10 | |
| *** psachin has quit IRC | 12:11 | |
| *** NM has joined #openstack-swift | 12:22 | |
| *** e0ne has quit IRC | 12:27 | |
| *** e0ne has joined #openstack-swift | 12:41 | |
| *** timburke has quit IRC | 12:46 | |
| *** timburke has joined #openstack-swift | 12:48 | |
| *** ChanServ sets mode: +v timburke | 12:48 | |
| *** e0ne has quit IRC | 12:55 | |
| *** e0ne has joined #openstack-swift | 12:59 | |
| *** altlogbot_0 has joined #openstack-swift | 13:05 | |
| *** tkajinam_ has joined #openstack-swift | 13:25 | |
| *** tkajinam has quit IRC | 13:27 | |
| *** altlogbot_0 has quit IRC | 13:32 | |
| *** altlogbot_1 has joined #openstack-swift | 13:34 | |
| *** d34dh0r53 has quit IRC | 13:35 | |
| *** altlogbot_1 has quit IRC | 13:38 | |
| *** altlogbot_3 has joined #openstack-swift | 13:40 | |
| *** e0ne has quit IRC | 13:42 | |
| *** altlogbot_3 has quit IRC | 14:00 | |
| *** thurloat has joined #openstack-swift | 14:15 | |
| *** altlogbot_0 has joined #openstack-swift | 14:27 | |
| *** altlogbot_0 has quit IRC | 14:29 | |
| *** altlogbot_1 has joined #openstack-swift | 14:31 | |
| *** e0ne has joined #openstack-swift | 14:31 | |
| *** altlogbot_1 has quit IRC | 14:33 | |
| *** altlogbot_0 has joined #openstack-swift | 14:49 | |
| *** altlogbot_0 has quit IRC | 14:51 | |
| *** altlogbot_2 has joined #openstack-swift | 14:53 | |
| *** altlogbot_2 has quit IRC | 14:55 | |
| *** altlogbot_1 has joined #openstack-swift | 14:59 | |
| *** ccamacho has quit IRC | 15:20 | |
| *** e0ne has quit IRC | 15:24 | |
| *** psachin has joined #openstack-swift | 16:01 | |
| *** gyee has joined #openstack-swift | 16:15 | |
| *** ndk_ has joined #openstack-swift | 16:25 | |
| *** zaitcev_ is now known as zaitcev | 16:34 | |
| timburke | good morning | 16:46 |
|---|---|---|
| zaitcev | guys | 16:54 |
| zaitcev | What if we have 3 nodes hosting replicas for container X, and 1 reports wrong size for X? Like 915520 objects in 1 node, and 862400 on 2 others (consistent). Is there a way to force a replication to happen short of rm(1) on the wrong node? | 16:55 |
| zaitcev | I think replicatior never knows which one is right. It only looks at the timestamps at the local db, and if they are newer, it updates. So the summary amounts never agree.i | 16:58 |
| notmyname | good morning | 17:01 |
| *** e0ne has joined #openstack-swift | 17:03 | |
| *** e0ne has quit IRC | 17:03 | |
| timburke | zaitcev, when you get to https://github.com/openstack/swift/blob/2.21.0/test/unit/obj/test_diskfile.py#L293-L343 -- don't trust that the test is right :-( | 17:05 |
| timburke | on the db question -- what do the sync points look like for the different dbs? do the replicators *think* that they're in sync, or do they (slowly) make progress on them each cycle? | 17:07 |
| zaitcev | timburke: I don't have first-hand access, so I don't know. | 17:07 |
| zaitcev | timburke: I only know the last update timestamp, which is the same on all 3. But not the sync points. | 17:09 |
| zaitcev | timburke, so what is wrong with that test? All I can see is, it divorces the back-end support from the question of what's legal on the wire. So, we have these non-UTF-8 things (including keys), and we do some things when we encounter them. What is the problem? | 17:16 |
| timburke | things like the "# sanity check, while still mocked, that we did persist unicode" comment followed by tests like "self.assertIsInstance(k, str)" | 17:21 |
| timburke | i think i botched something in https://github.com/openstack/swift/commit/36c4297 | 17:21 |
| *** e0ne has joined #openstack-swift | 17:23 | |
| timburke | i remember starting to look at diskfile, getting to that failure, getting lost in the various bits of mocking, and deciding to walk away and come back to it later | 17:23 |
| timburke | that's all | 17:23 |
| zaitcev | botched in what way? It's the same code verbatim for six.PY2 | 17:24 |
| zaitcev | same code! | 17:24 |
| zaitcev | So rest assured, nothing is bothched yet, except maybe the py3 case | 17:24 |
| zaitcev | Ah | 17:24 |
| timburke | six.text_type and str are different in py2 | 17:25 |
| zaitcev | You mean the exact compatibility on the disk, right? | 17:25 |
| zaitcev | But it's okay, as long as nobody got py3 into production we can fix it all. | 17:25 |
| zaitcev | anyway | 17:26 |
| timburke | yeah, for sure. we'll definitely be able to get both py2 and py3 reading (and writing? i'm pretty sure?) the same diskfiles -- i'm not too worried about that | 17:26 |
| zaitcev | I'm going to Dallas real quick in the afternoon, I'll be back and work on it while it rains on Saturday. | 17:26 |
| timburke | just wanted to call out a test that probably needs re-writing | 17:27 |
| zaitcev | I see. | 17:27 |
| zaitcev | I'll try to dig into this properly this time, instead of just pushing things through tox and hoping it works. | 17:28 |
| *** e0ne has quit IRC | 17:29 | |
| *** early has quit IRC | 17:32 | |
| *** early has joined #openstack-swift | 17:35 | |
| clayg | zaitcev: I'm in Dallas(ish)! yes boo rain | 17:36 |
| clayg | timburke: any great ideas about the hardlink to a symlink problem? I think the simplest form of my current thinking is "given how much versioned writes breaks if names aren't unique - why do we need to use hardlinks again?" | 17:55 |
| timburke | clayg, i'm starting to think that we should maybe use normal symlinks if we're versioning a symlink, and hard links if it's a normal object... | 18:00 |
| timburke | fwiw, i wanted hard links as a way to (1) provide some assurances about the backing data not changing and (2) have a handy way to write down etag/size so we can fix up listings for the primary container | 18:00 |
| clayg | oh right, the listing fixup is huge | 18:00 |
| clayg | dammit - i remembered that earlier and then forgot again | 18:01 |
| timburke | idk -- maybe i'm worrying too much about listings | 18:02 |
| timburke | but it *seems* like a thing worth preserving if/when we can | 18:02 |
| *** irclogbot_0 has quit IRC | 18:08 | |
| *** irclogbot_1 has joined #openstack-swift | 18:11 | |
| *** rdejoux has quit IRC | 18:16 | |
| notmyname | if you're interested in some of the data pipelines for self driving cars, you'll likely find this video interesting. https://on-demand.gputechconf.com/gtc/2019/video/_/S9649/ | 18:21 |
| notmyname | ...and there may be some indirect references to swift being a pretty important part of that ;-) | 18:21 |
| *** psachin has quit IRC | 18:46 | |
| *** NM has quit IRC | 19:24 | |
| *** NM has joined #openstack-swift | 19:27 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Make staticweb return URL-encoded Location headers https://review.openstack.org/571907 | 19:38 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Support Unicode in account and user names during func tests https://review.openstack.org/571908 | 19:38 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Use more pathologically-bad names for func tests https://review.openstack.org/651940 | 19:38 |
| *** pcaruana has quit IRC | 20:26 | |
| *** NM has quit IRC | 21:31 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Use more pathologically-bad names for func tests https://review.openstack.org/651940 | 21:41 |
| *** itlinux has joined #openstack-swift | 22:06 | |
| *** NM has joined #openstack-swift | 23:53 | |
| timburke | holy crap. i just realized that the s3api port will put us over 5k py3 unit tests: http://logs.openstack.org/93/651893/3/check/swift-tox-py37/89f000e/job-output.txt.gz#_2019-04-11_22_42_58_968194 | 23:58 |
| timburke | py27 has ~6800 -- that's like 74%! | 23:58 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!