| *** priteau_ is now known as priteau | 12:25 | |
| opendevreview | Merged openstack/swift master: proxy-server exception logging shows replication_ip/port https://review.opendev.org/c/openstack/swift/+/860866 | 21:34 |
|---|---|---|
| mcape | Hi! The rocky - yoga update story, part 4. Object layer went without any issues. Hurray! However, update for container servers (just servers first, replicators/sharders were left on rocky) got quite a bunch of 500 errors. | 22:42 |
| mcape | They fall into two categories, and look like https://pastebin.com/23iCcPmP, and https://pastebin.com/dDES2KYM. I rolled back quickly, to try to digest what is going on. It feels like that this errors are coming from same deleted shards i am getting a lot of container-sharder warnings about, | 22:42 |
| mcape | however it is unclear now how to better proceed. | 22:42 |
| mcape | Here's an example of such a shard: https://pastebin.com/7X0AKYPG | 22:42 |
| mcape | I thought it will be probably better to apply patches: "Identify deleted shards as shards" https://review.opendev.org/c/openstack/swift/+/738113 and "Do not delete root_path on ContainerBroker.delete_db" https://review.opendev.org/c/openstack/swift/+/773832, as timburke pointed before, and look how it will work out. However "Identify deleted shards as shards" is not applying cleanly on my files from rocky. | 22:42 |
| mcape | So I guess i should bite the bullet and just do upgrade for all container-level stuff and hope that new code will sort everything out without crashing the clusters. | 22:43 |
| mcape | Any input is much appreciated. | 22:43 |
| mcape | Have there been changes in the structure of the container databases in the new versions that will not allow me to go back in case of problems? | 22:43 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!