Corrupted transaction log and redownloading
Hi,
today we had to hard-reboot one of our dovecot proxy backends and
after boot up, lot's of accounts were showing this error and appeard
empty:
Error: Corrupted transaction log file ...
After deleting the indexes (not uid file, problem wasn't in deleting
indexes, i'm doing this often), e-mails showed up ok but e-mail
clinets started to download them again.
Why this happened and how to prevent it (other than 'not
hard-rebooting')? I know that reboot could cause corruption of uid
files in some cases but this time it looks like this was caused by
corrupted indexes - how index corruption can cause redownloading and
why?
Thanks for any info.
azur
On 25 February 2019 18:49 azurit--- via dovecot <dovecot@dovecot.org> wrote:
Hi,
today we had to hard-reboot one of our dovecot proxy backends and
after boot up, lot's of accounts were showing this error and appeard
empty: Error: Corrupted transaction log file ...After deleting the indexes (not uid file, problem wasn't in deleting
indexes, i'm doing this often), e-mails showed up ok but e-mail
clinets started to download them again.Why this happened and how to prevent it (other than 'not
hard-rebooting')? I know that reboot could cause corruption of uid
files in some cases but this time it looks like this was caused by
corrupted indexes - how index corruption can cause redownloading and
why?Thanks for any info.
azur
dovecot should fix corrupted transaction log file, but it usually requires another log in from the user, and sometimes
doveadm force-resync -u victim "*"
# or if you have INBOX/ namespace prefix doveadm force-resync -u victim "INBOX/*"
Aki
participants (2)
-
Aki Tuomi
-
azurit@pobox.sk