| *** gregwork has quit IRC | 01:33 | |
| *** mahatic has joined #openstack-swift | 02:50 | |
| *** ChanServ sets mode: +v mahatic | 02:50 | |
| *** psachin has joined #openstack-swift | 03:33 | |
| *** evrardjp has quit IRC | 04:35 | |
| *** evrardjp has joined #openstack-swift | 04:36 | |
| *** rpittau|afk is now known as rpittau | 06:33 | |
| openstackgerrit | Andreas Jaeger proposed openstack/swift master: Switch to new grenade job name https://review.opendev.org/725190 | 06:41 |
|---|---|---|
| *** ccamacho has joined #openstack-swift | 07:23 | |
| *** mikecmpbll has joined #openstack-swift | 07:54 | |
| *** dtantsur|afk is now known as dtantsur | 08:43 | |
| *** rpittau is now known as rpittau|bbl | 10:14 | |
| openstackgerrit | Sean McGinnis proposed openstack/swift master: Use local py2 upper-constraints https://review.opendev.org/724422 | 12:04 |
| *** rpittau|bbl is now known as rpittau | 12:22 | |
| *** mikecmpbll has quit IRC | 12:49 | |
| *** mikecmpbll has joined #openstack-swift | 12:53 | |
| *** psachin has quit IRC | 13:14 | |
| seongsoocho | Hi ~ I have a question about how the object-replicator works. (only for update process) The replicator compare the suffix directory's hash value of partition between local and remote then run rsync if different. So.. How to know which node has newest data? Does rsync send a file only newest ? | 13:35 |
| DHE | as i understand it (mere user here) each object is named with its creation timestamp to the microsecond. so if there are duplicates then rsync will give BOTH objects to the node and it can sort it out. the receiver needs to do a pass over the received data anyway | 14:04 |
| seongsoocho | DHE: Thanks. I modified a file in one node, the replicator restored it. So.. I was confused about how rsync and replicator know the correct file. | 14:10 |
| DHE | there's an auditor that md5'd the file, found it corrupted, and removed it for replication to fix | 14:15 |
| DHE | although... | 14:15 |
| DHE | yeah, --ignore-existing is one of the args, so one machine won't overwrite an existing file on another machine, so if one is corrupted it won't push a corrupted update | 14:19 |
| seongsoocho | DHE: oh i see... Actually, the weird things happen to my cluster. | 14:24 |
| seongsoocho | The object replicator try to replicate empty suffix variable.. so the rsync method return false and replicate job record as a failure.. | 14:25 |
| openstackgerrit | Chris MacNaughton (icey) proposed openstack/swift master: Monkey patch original current_thread _active https://review.opendev.org/725288 | 14:31 |
| *** gyee has joined #openstack-swift | 15:25 | |
| *** rpittau is now known as rpittau|afk | 16:08 | |
| openstackgerrit | Merged openstack/swift master: Use separate name for HeaderKeyDict var vs list of response headers https://review.opendev.org/723736 | 16:26 |
| *** evrardjp has quit IRC | 16:35 | |
| *** evrardjp has joined #openstack-swift | 16:36 | |
| *** dtantsur is now known as dtantsur|afk | 16:57 | |
| *** hoonetorg has quit IRC | 17:23 | |
| seongsoocho | clayg: HI, I read your commit about 3 years ago. ( https://github.com/openstack/swift/commit/aa71d7e77b266a4111b3b61ed41a204a28ec24e8 ) . After reading the commit message, it seems that I need to delete all hashes.pkl to upgrade from earlier rocky to latest. Is it correct ?? | 17:52 |
| openstackgerrit | Tim Burke proposed openstack/swift master: probe tests: Work when fronted by a TLS terminator https://review.opendev.org/705579 | 17:54 |
| *** Jeffrey4l has quit IRC | 18:23 | |
| *** Jeffrey4l has joined #openstack-swift | 18:31 | |
| *** hugokuo has joined #openstack-swift | 18:32 | |
| *** timburke has joined #openstack-swift | 18:32 | |
| *** ChanServ sets mode: +v timburke | 18:32 | |
| *** Jeffrey4l has quit IRC | 18:36 | |
| *** Jeffrey4l has joined #openstack-swift | 18:43 | |
| timburke | seongsoocho, my memory of that patch was just that you might need to delete hashes.pkl if you upgraded *and then attempted to roll back* -- upgrades were fine, i think, it was just that the old code would fail in some bad way if it saw the new format | 18:47 |
| seongsoocho | timburke: aha. thanks. Because .. After I upgrade from ocata to train, some partition's hashes.pkl is missmatch with other node.. So I try to find the reason.. ( with crying ) | 18:50 |
| seongsoocho | Is it safe(?) to remove hashes.pkl in production...? | 18:51 |
| *** hoonetorg has joined #openstack-swift | 19:00 | |
| timburke | yeah -- it'll cause a little extra IO, but should recover pretty quickly | 19:01 |
| seongsoocho | great... try to delete a hashes.pkl .... 🙂 | 19:03 |
| seongsoocho | always thank you ! | 19:03 |
| timburke | happy to help :-) | 19:03 |
| *** threestrands has joined #openstack-swift | 20:34 | |
| *** ccamacho has quit IRC | 21:39 | |
| *** rcernin has joined #openstack-swift | 22:22 | |
| zaitcev | Gate cannot even verify a change to docs. | 22:31 |
| mattoliverau | morning | 23:15 |
| openstackgerrit | Tim Burke proposed openstack/swift master: WIP: Batched updates for object-updater https://review.opendev.org/724943 | 23:16 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Add a tool to peek inside async_pendings https://review.opendev.org/725429 | 23:16 |
| mattoliverau | timburke: I assume you've seen it, but there is an email re last call for project to signup for virtual PTG, and I don't see Swift on this list. | 23:16 |
| timburke | yeah, i *do* need to do that... uhg. i feel like i have no time or mental bandwidth for this :-( | 23:17 |
| openstackgerrit | Merged openstack/swift stable/ussuri: Use separate name for HeaderKeyDict var vs list of response headers https://review.opendev.org/724395 | 23:25 |
| openstackgerrit | Tim Burke proposed openstack/swift master: object-updater: Include update path when logging errors https://review.opendev.org/725432 | 23:34 |
| mattoliverau | timburke: need help? I guess we can talk about times during the next meeting. Want me to go see what we need to do to signup? And if it doesn't require giving all sessions and times make it happen (if I am able to). | 23:36 |
| timburke | mattoliverau, that'd be great, thanks! 🤗 | 23:46 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!