| openstackgerrit | Tim Burke proposed openstack/swift master: watchers: Add a policy-stat watcher https://review.opendev.org/c/openstack/swift/+/766640 | 00:03 |
|---|---|---|
| openstackgerrit | Tim Burke proposed openstack/swift master: Dark data probe test follow-up https://review.opendev.org/c/openstack/swift/+/766549 | 00:12 |
| openstackgerrit | Tim Burke proposed openstack/swift master: watchers: Only pass data_file_path via see_object https://review.opendev.org/c/openstack/swift/+/766612 | 00:12 |
| openstackgerrit | Tim Burke proposed openstack/swift master: watchers: Add a policy-stat watcher https://review.opendev.org/c/openstack/swift/+/766640 | 00:12 |
| zaitcev | oooh, I was just going to tinker with that file path | 00:24 |
| zaitcev | timburke: David and I started with making a quarantine signal. However, we didn't throw an exception, but set a flag somewhere. | 00:26 |
| zaitcev | timburke: what do you think about adding a new exception, like "DiskFileQuarantineRequested"? | 00:27 |
| zaitcev | actually I should be adding that as a comment | 00:28 |
| timburke | that'd be fine. or somehow setting a flag on the watcher -- exception seemed like a reasonable tool to reach for (to me), but it's definitely not the only way to do it | 00:29 |
| zaitcev | Oh | 00:34 |
| zaitcev | I see, we didn't go far enough. | 00:35 |
| zaitcev | The log of this says we abandoned flag because we didn't add another invocation of df._quarantine() to the auditor. https://review.opendev.org/c/openstack/swift/+/756163 | 00:35 |
| *** baojg has quit IRC | 00:46 | |
| *** baojg has joined #openstack-swift | 00:46 | |
| *** tdasilva_ has quit IRC | 01:44 | |
| *** lxkong has quit IRC | 01:48 | |
| *** baojg has quit IRC | 01:49 | |
| *** dsariel has quit IRC | 01:49 | |
| *** baojg has joined #openstack-swift | 01:50 | |
| *** baojg has quit IRC | 01:57 | |
| *** baojg has joined #openstack-swift | 01:58 | |
| -openstackstatus- NOTICE: The Gerrit service on review.opendev.org is being restarted quickly to enable support for Git protocol v2, downtime should be less than 5 minutes | 02:55 | |
| *** lifeless has quit IRC | 03:31 | |
| *** lifeless has joined #openstack-swift | 03:33 | |
| *** rcernin has quit IRC | 03:39 | |
| *** rcernin has joined #openstack-swift | 03:40 | |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: Dark data probe test follow-up https://review.opendev.org/c/openstack/swift/+/766549 | 05:16 |
| *** timburke has quit IRC | 05:31 | |
| *** timburke has joined #openstack-swift | 05:31 | |
| *** ChanServ sets mode: +v timburke | 05:31 | |
| zaitcev | timburke: https://review.opendev.org/c/openstack/swift/+/766612 looks good except for a couple of small things: watcher_name and removal of whole directory where object was, not just the data file alone. | 05:32 |
| *** evrardjp_ has quit IRC | 05:33 | |
| *** evrardjp has joined #openstack-swift | 05:33 | |
| *** lxkong has joined #openstack-swift | 06:16 | |
| *** adeberg5 has joined #openstack-swift | 06:28 | |
| *** adeberg has quit IRC | 06:30 | |
| *** adeberg5 is now known as adeberg | 06:30 | |
| *** rpittau|afk is now known as rpittau | 07:57 | |
| *** rcernin has quit IRC | 08:07 | |
| *** rcernin has joined #openstack-swift | 08:14 | |
| *** evrardjp has quit IRC | 08:15 | |
| *** evrardjp has joined #openstack-swift | 08:16 | |
| *** StormsOfFury has quit IRC | 08:35 | |
| *** TViernion has joined #openstack-swift | 08:36 | |
| *** rcernin has quit IRC | 09:27 | |
| *** baojg has quit IRC | 10:08 | |
| *** baojg has joined #openstack-swift | 10:08 | |
| *** StevenK has quit IRC | 11:35 | |
| *** StevenK has joined #openstack-swift | 11:44 | |
| *** baojg has quit IRC | 12:17 | |
| *** baojg has joined #openstack-swift | 12:18 | |
| *** lxkong has quit IRC | 12:56 | |
| openstackgerrit | Merged openstack/swift master: Give probe tests a second chance to pass https://review.opendev.org/c/openstack/swift/+/766104 | 12:58 |
| openstackgerrit | Alistair Coles proposed openstack/swift master: swift-manage-shard-ranges: add repair and analyze commands https://review.opendev.org/c/openstack/swift/+/765624 | 15:01 |
| *** rpittau is now known as rpittau|afk | 15:15 | |
| openstackgerrit | Alistair Coles proposed openstack/swift master: swift-manage-shard-ranges: add repair and analyze commands https://review.opendev.org/c/openstack/swift/+/765624 | 15:17 |
| acoles | can anyone help me understand/fix this horrible hack I needed to catch an exception whose type isn't matching as expected? https://review.opendev.org/c/openstack/swift/+/765624/5/test/probe/test_sharder.py#381 | 15:22 |
| *** rcernin has joined #openstack-swift | 15:24 | |
| *** rcernin has quit IRC | 15:28 | |
| zaitcev | Impressive. | 17:05 |
| *** gyee has joined #openstack-swift | 17:33 | |
| timburke | acoles, something something eventlet monkey-patching? though you'd think https://github.com/eventlet/eventlet/commit/0425e4c should rule that out... what version of eventlet is the 3.6 env using? | 18:38 |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: Let developers/operators add watchers to object audit (simplified) https://review.opendev.org/c/openstack/swift/+/706653 | 18:59 |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: Dark data probe test follow-up https://review.opendev.org/c/openstack/swift/+/766549 | 18:59 |
| openstackgerrit | Pete Zaitcev proposed openstack/swift master: watchers: Only pass data_file_path via see_object https://review.opendev.org/c/openstack/swift/+/766612 | 18:59 |
| zaitcev | timburke: I added fixes that I wanted, but kept these separate for you to review easier. I'm going to squash them all in. | 19:02 |
| timburke | sounds like a plan! | 19:05 |
| timburke | zaitcev, have any thoughts on https://review.opendev.org/c/openstack/swift/+/706653/40/swift/obj/auditor.py#202 ? i'm realizing that with a lot of the watchers i want to write, i want them emitting structured data to recon (as opposed to finding some way to stuff output into logging, then having to do a bunch of extra log-parsing) | 19:23 |
| timburke | but i don't think that'll work well with multiple audit workers (currently)... | 19:24 |
| timburke | part of me wonders if the watcher API could/should include a static method like `aggregate(list_of_per_worker_watchers)` or something... but when run as a daemon, those workers just keep looping; there's no single point where you've got a bunch of completed watchers in hand... | 19:27 |
| zaitcev | timburke: I saw that comment and I hoped that watches that report data can dump it into some kind of socket in concert with some listener process. Maybe directly to Prometheus. | 19:42 |
| zaitcev | For me, auditor is basically find /srv -name '*.data' -print0| xargs -0 mywatcher | 19:43 |
| timburke | hmm... good point. but we still need a way to disambiguate between watchers that are runing concurrently on the same node, so we know whether new stats should sum with or replace old ones. i think we're going to need that device_key (maybe renamed as worker_id or something?) getting sent in to end()... | 19:57 |
| timburke | interesting: http://lists.openstack.org/pipermail/openstack-discuss/2020-December/019379.html | 20:08 |
| timburke | seems like polishing up sharding/shrinking and removing that "experimental" label would fit really well :-) | 20:09 |
| *** lxkong has joined #openstack-swift | 20:24 | |
| openstackgerrit | Ade Lee proposed openstack/swift master: replace md5 with swift utils version https://review.opendev.org/c/openstack/swift/+/751966 | 21:46 |
| *** lxkong has quit IRC | 22:32 | |
| openstackgerrit | Merged openstack/swift master: s3api: Clone ACLs when creating +segments container https://review.opendev.org/c/openstack/swift/+/763106 | 22:43 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!