[Dovecot] suggestion on crash analysis
Jeff A. Earickson
jaearick at colby.edu
Thu Dec 22 21:02:31 EET 2005
On Thu, 22 Dec 2005, Timo Sirainen wrote:
> It's good that people report backtraces from crashes here, but too often
> I just can't figure out how to reproduce them and can't see anything
> wrong with the code either. It would really help if people tried
> reproducing the bugs themselves and told me how it can be done..
One problem in reproducing an assert/crash is that the person's
mailbox has changed by the time I find the dovecot core file and
run gdb against it. It would be really cool if i_assert would save
a copy of the INBOX (and any other open files) into the user's home
directory at the time it generated the core. That way maybe I could
have a data file to work with.
Another idea... I know there is the rawlog feature, but I don't
want to turn it on for everybody, since asserts are very rare now.
Any chance some kind of memory buffer for recent user actions could
be kept in memory, then dumped to a file in the person's homedir
if an assert occurs?
This would answer the two unknown questions: 1) What was the user
doing?, 2) What was the state of the file he was doing it to?
I am open to suggestions for better procedures at my end too.
More information about the dovecot