Details
-
Bug
-
Status: Closed (View Workflow)
-
Minor
-
Resolution: Not a Bug
-
10.11.6
-
None
-
Debian bookworm
Description
Nightly backups of mariadb via mariadb-dump hang randomly, and at random spots in the backup.
- full backup is about 5G
- the backup can hang at any point, 10% in, 50%, etc
- the database is mostly static, and mostly idle
(it is a test DB staging before PROD, so it has small change, and is completely idle during backups at night)
What happens:
- cron will start a backup of the DB to a file in /tmp (same issue happens with piped output)
- at a random point, the dump will stall. It just hangs. It will never resume (even days) if left alone.
- while attempting to debug, I noticed that running a second mysqldump immediately kicks off a resume, the same is true of a strace. The dump will then complete successfully
strace just shows the standard upon attach:
113065 restart_syscall(<... resuming interrupted read ...> <unfinished ...>
and everything then just seems normal.
Again, it's not just strace. Doing another mysqldump "kicks" the first into action again.
We're not seeing this on busy servers, as I presume at this point that any sort of DB activity will kick it back into gear. Will verified next time this happens.
(It only happens every few days)