Re: [BackupPC-users] Origin and/or implications of "aborted by signal=ALRM"?

[Craig B. <cba...@us...>]

Kris Boutilier writes:

> I have one particular host that has a very large and dense volume that
> BackupPC is trying to rsync (~300gb/1.2million files). The host is connected
> over 100mb ethernet and is running rsync 2.6.3 on Cygwin/Windows 2000. The
> backup starts successfully and the first of two rsync 'shares' (<1Gb) backs
> up fine, but the second quits every time after about 10h 30m with 'Got fatal
> error during xfer (aborted by signal=ALRM)' showing in the BackupPC log for
> the host.
> 
> The rsyncd process isn't running out of memory, nor is the backuppc host.
> There is plenty of disk space, processor capacity and bandwidth. I have
> raised "$Conf{ClientTimeout}" as high as 720000 and reloaded the server
> configuration, with no change.
> 
> Any suggestions on the causes of an Alarm signal, which side it's being
> generated/received on or what process might be generating it? It looks like
> a timeout watchdog kicking in, but the ClientTimeout should be plenty high
> enough...

Yes, it should be.  The only source of an ALRM signal should be
$Conf{ClientTimeout}.  It is run on the server side.  Are you sure
there isn't a per-PC config.pl setting for $Conf{ClientTimeout}
that is a smaller value?

Craig