I just discovered that our backups were suffering failures due to not being able to parse log files. I suspect that this originated with and out of disk space condition that caused a log file to be truncated. This led to subsequent runs failing with:
** (process:20892): CRITICAL **: 04:20:30.170: '' at /usr/lib64/perl5/vendor_perl/Amanda/DB/Catalog.pm line 839.
This would then cascade because the log file what process was producing would be truncated when it died.
I've filed #188 to at least make the original message helpful in finding where the problem is. Some additional questions:
- Do we have to die here? Can we somehow recover from the error?
- If we have to die - can we avoid truncating the log file? Perhaps by making some more file handles autoflush?
I just discovered that our backups were suffering failures due to not being able to parse log files. I suspect that this originated with and out of disk space condition that caused a log file to be truncated. This led to subsequent runs failing with:
This would then cascade because the log file what process was producing would be truncated when it died.
I've filed #188 to at least make the original message helpful in finding where the problem is. Some additional questions: