Comment 3 for bug 750359

Hi Alvin,
  To my reading of those logs something is running out of memory an then things are getting more and more upset (with softlockup detection in each cpu) - this shouldn't happen in principal, something should kill whatever is eating up all the memory.

  The other question is what is running out of memory; it might be useful to start your test and then every few minutes gather the output of

   free
   cat /proc/slabinfo
   ps -eafuw

and then attach the last couple of these before it dies to this report.

One thing I've seen in the past is that rsync leaks ~96bytes/file while the process is running (It's not really a leak - they keep it around to detect if the same file comes around again). If you are backing up something with zillions of small files it can mean your rsync processes get to be huge - I've hit that problem with backing up large source file trees with literally millions of files.

Dave