Tuesday, 2021-08-03

*** mabrams is now known as mabrams|afk07:03
*** mabrams|afk is now known as mabrams07:04
*** diablo_rojo__ is now known as diablo_rojo09:02
diablo_rojotimburke_, did you see my message in the scrollback from yesterday?09:03
manuvakery1Hi Guys09:42
manuvakery1aws s3api list-object-versions --bucket manu-test returns  "An error occurred (InternalError) when calling the ListObjectVersions operation (reached max retries: 2): unexpected status code 412" 09:42
manuvakery1After looking into logs I found that the issue is happening only  on a specific container disk.  unmounting that disk  fixed the issue . What could be the root cause 09:42
manuvakery11724969:Aug  3 09:31:18 swift-storage5 container-server: 10.0.0.23 - - [03/Aug/2021:\ 09:31:18 +0000] "GET /disk2/1168/b5d335382b91478e981d6f3c7b623b1/%00versions%00manu-test" 412 29 \ "GET /v1/b5d335382b91478e981d6f3c7b623b1/%00versions%00manu-test?reverse=&format=json&delimiter=&states=listing&prefix=&limit=1001&marker=" "tx99ab930b8c36435c9d361-0061090ce6" "proxy-server 11947" 0.0002 "-" \ 227639 -09:45
opendevreviewMatthew Oliver proposed openstack/swift master: sharder: update shard storage_policy_index if roots changes  https://review.opendev.org/c/openstack/swift/+/80074810:42
mattoliver^ haven't updated the probe test with that yet. But that allows the replicator to handle the shard spi migration if there is more then 1 policy detected, is a shard and before enqueuing any reconciler jobs. But also leave the sharder to do it too, so who ever gets there first (tho maybe I should just remove it from the sharder) :)10:44
mattoliverAnyway, wanted to get it off my computer and stop looking at it for the night. 10:45
claygdoes swift-object-info not open the .meta?15:24
claygok then https://github.com/openstack/swift/blob/master/swift/cli/info.py#L53415:38
claygbut I guess you can ues it on a .meta - so maybe that's ok15:41
manuvakery1Hi 16:30
manuvakery1aws s3api list-object-versions --bucket manu-test returns  "An error occurred (InternalError) when calling the ListObjectVersions operation (reached max retries: 2): unexpected status code 412" 16:30
manuvakery13:12 PM 16:30
manuvakery1After looking into logs I found that the issue is happening only  on a specific container disk.  unmounting that disk  fixed the issue . What could be the root cause 16:30
manuvakery13:15 PM 16:30
manuvakery11724969:Aug  3 09:31:18 swift-storage5 container-server: 10.0.0.23 - - [03/Aug/2021:\ 09:31:18 +0000] "GET /disk2/1168/b5d335382b91478e981d6f3c7b623b1/%00versions%00manu-test" 412 29 \ "GET /v1/b5d335382b91478e981d6f3c7b623b1/%00versions%00manu-test?reverse=&format=json&delimiter=&states=listing&prefix=&limit=1001&marker=" "tx99ab930b8c36435c9d361-0061090ce6" "proxy-server 11947" 0.0002 "-" \ 227639 -16:30
manuvakery1I formatted the disk and mounted again but issue persists16:32
timburke_manuvakery1, any chance the container server is still running an old version of swift? only place it should return a 412 is at https://github.com/openstack/swift/blob/2.28.0/swift/container/server.py#L885-L886, but that path looks valid for swift 2.24.0+18:34
timburke_i just noticed there's a downside to https://github.com/openstack/swift/commit/ade3b28 -- the pre-rsync REPLICATE call no longer creates the partition dir, causing the rsync to fail with something like "rsync: mkdir "/sdb3/objects-1/37" (in object_sdb3) failed: No such file or directory (2)"18:39
timburke_since we made the attempt, though, the replicator will invalidate suffixes after the failure, though, which *will* create the partition dir -- so the next pass should work fine. it's a little added friction, though :-/18:40
opendevreviewTim Burke proposed openstack/swift master: WIP: X-If-Delete-At failing probe test  https://review.opendev.org/c/openstack/swift/+/80340623:49

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!