| *** psachin has joined #openstack-swift | 03:36 | |
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #openstack-swift | 04:33 | |
| *** psachin has quit IRC | 04:58 | |
| *** rcernin has quit IRC | 06:02 | |
| *** rcernin has joined #openstack-swift | 06:11 | |
| *** m75abrams has joined #openstack-swift | 06:25 | |
| *** ccamacho has joined #openstack-swift | 07:13 | |
| *** rcernin has quit IRC | 08:06 | |
| *** baojg has quit IRC | 09:21 | |
| *** baojg has joined #openstack-swift | 09:26 | |
| *** rcernin has joined #openstack-swift | 09:54 | |
| *** ccamacho has quit IRC | 09:58 | |
| *** rcernin has quit IRC | 10:03 | |
| *** rcernin has joined #openstack-swift | 10:08 | |
| *** rcernin has quit IRC | 10:40 | |
| *** baojg has quit IRC | 11:14 | |
| *** baojg has joined #openstack-swift | 11:15 | |
| *** ccamacho has joined #openstack-swift | 11:34 | |
| *** tkajinam has quit IRC | 11:37 | |
| *** adriant has quit IRC | 11:58 | |
| *** adriant has joined #openstack-swift | 11:59 | |
| *** jv_ has joined #openstack-swift | 12:57 | |
| *** jv_ has joined #openstack-swift | 12:57 | |
| *** jv_ has quit IRC | 12:59 | |
| *** jv_ has joined #openstack-swift | 12:59 | |
| *** baojg has quit IRC | 13:26 | |
| *** irclogbot_0 has quit IRC | 13:27 | |
| *** charz has quit IRC | 13:27 | |
| *** irclogbot_0 has joined #openstack-swift | 13:28 | |
| *** charz has joined #openstack-swift | 13:28 | |
| *** noonedeadpunk has quit IRC | 13:31 | |
| *** noonedeadpunk has joined #openstack-swift | 13:33 | |
| *** manuvakery has joined #openstack-swift | 14:10 | |
| *** m75abrams has quit IRC | 15:06 | |
| *** Fengli has quit IRC | 15:26 | |
| *** gyee has joined #openstack-swift | 15:41 | |
| *** manuvakery has quit IRC | 16:20 | |
| *** m75abrams has joined #openstack-swift | 18:08 | |
| *** zaitcev has joined #openstack-swift | 18:16 | |
| *** ChanServ sets mode: +v zaitcev | 18:16 | |
| *** ccamacho has quit IRC | 20:52 | |
| clayg | mattoliverau: timburke: I'm merging p 738113 - it's super hard to think about shrinking with the way deleted sharded dbs start thinking they're roots | 21:22 |
|---|---|---|
| patchbot | https://review.opendev.org/#/c/738113/ - swift - Identify deleted shards as shards - 3 patch sets | 21:22 |
| timburke | so true! | 21:23 |
| openstackgerrit | Tim Burke proposed openstack/swift master: sharding: Add some more shard-related output to swift-container-info https://review.opendev.org/721377 | 21:30 |
| timburke | ^^^ also super-useful when investigating all this | 21:33 |
| timburke | clayg, got a gut reaction on p 735484? it seems wrong to me that we've got this extra layer of indirection that seems mainly if not entirely for the sake of tests -- but otoh, it's a decent bit of churn for what's maybe mostly aesthetics | 21:43 |
| patchbot | https://review.opendev.org/#/c/735484/ - swift - Get rid of memcache attr on proxy-server app - 4 patch sets | 21:43 |
| clayg | seems like a reasonable cleanup - "one obvious way to do it" 👍 | 21:45 |
| clayg | zaitcev: y u no merge!? | 21:45 |
| zaitcev | clayg, just waiting for the gate to check it | 21:46 |
| clayg | noice | 21:50 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Address a sharder/replicator race https://review.opendev.org/738114 | 21:57 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Have shrinking and sharded shards save all ranges from root https://review.opendev.org/738149 | 21:57 |
| clayg | timburke: so the order we return over lapping shard ranges isn't entirely stable - we order by (upper, state, lower) - i was experimenting with some overlaps that happened to have the same values for those fields | 22:00 |
| *** Fengli has joined #openstack-swift | 22:01 | |
| timburke | ... which is definitely something i've seen... | 22:03 |
| timburke | still, i'd expect an individual replica of the root to be pretty consistent -- but the order between replicas probably isn't :-( | 22:04 |
| clayg | in my case the order was stable until another row was reclaimed from the shard_range table - then it flipped | 22:05 |
| clayg | ... that's what had me looking for the order by clause | 22:05 |
| timburke | so i definitely managed to lose some rows for a while in my probe test -- but i couldn't get it to happen reliably: https://review.opendev.org/#/c/738149/3/test/probe/test_sharder.py@2694 | 22:07 |
| patchbot | patch 738149 - swift - Have shrinking and sharded shards save all ranges ... - 3 patch sets | 22:07 |
| timburke | that assertion would fail something like 30% of the time as i recall | 22:08 |
| timburke | so i'm feeling pretty good about my new probe test when auto_shard is off... not so much when it's on: http://paste.openstack.org/show/795900/ | 22:22 |
| timburke | trouble comes down to https://github.com/openstack/swift/blob/2.25.0/swift/container/sharder.py#L194-L199 -- i think it needs to only do that when the donor wasn't already shrinking... and even then, i'm not sure it'll be super-robust | 22:24 |
| timburke | i guess another option might be to ensure better alignment when we pick winners and losers... | 22:25 |
| openstackgerrit | Merged openstack/swift master: Identify deleted shards as shards https://review.opendev.org/738113 | 22:31 |
| *** tkajinam has joined #openstack-swift | 22:54 | |
| *** rcernin has joined #openstack-swift | 23:05 | |
| *** rcernin has quit IRC | 23:05 | |
| *** rcernin has joined #openstack-swift | 23:08 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Have shrinking and sharded shards save all ranges from root https://review.opendev.org/738149 | 23:19 |
| *** hoonetorg has quit IRC | 23:24 | |
| *** adriant has quit IRC | 23:32 | |
| *** adriant has joined #openstack-swift | 23:33 | |
| *** charz has quit IRC | 23:35 | |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: Add a reminder to configure firewall at storage nodes https://review.opendev.org/740828 | 23:36 |
| *** hoonetorg has joined #openstack-swift | 23:39 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!