RSync hangs up on file transfer

[jw schultz]

On Fri, Aug 29, 2003 at 02:56:47PM +0200, Andre Alexander Bell wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hello!
> 
> I'm using RSync to mirror from a server (gandalf) to a backupserver (dwalin).
> RSync does just fit my needs for what I want it to do and I would like to 
> thank your for your work on RSync. Nevertheless I do have a problem which 
> actually comes up from time to time.
> The syncing of each partition is done by rsync calls from cron.daily.
> This is one of these calls:
> 
> rsync -a -v -P --delete --numeric-ids --rsh="ssh" root at gandalf:/images/ \
> 	/backup/images &> /var/log/rsync/backup_gandalf_images
> 
> The other calls are similar and differ just in their pathes. Actually this 
> last call sometimes hangs up. This means RSync does not terminate with error 
> nor does it without error. The RSync process is still 'running' but there is 
> no file transport anymore. Within the logfile it shows up at one last line 
> like this:
> 
> cytometer/C12883-98/fisher/C12883-98_FEU_0073_0000_063x_MASK.tif
>      1247804 100%    1.29MB/s    0:00:00
> 
> The file and directory within which this happens differs from time to time.
> Depending on the number of files that need to be transfered (about 300k files 
> on that partition and about 90k files to be synced) I can kill, restart, kill 
> and restart again and again until RSync is finished, but this is not nice :(.
> While RSync did hang up I got this output from free:
> 
> dwalin (Debian GNU/Linux, RSync 2.5.6 pv 26):
>              total       used       free     shared    buffers     cached
> Mem:        127496     123364       4132          0      16772      44104
> - -/+ buffers/cache:      62488      65008
> Swap:       497972     107600     390372
> 
> gandalf (SuSE GNU/Linux, RSync 2.5.6 pv 26):
>              total       used       free     shared    buffers     cached
> Mem:        256292     251052       5240          0      58360     113760
> - -/+ buffers/cache:      78932     177360
> Swap:       995988      44560     951428
> 
> So both systems have still left swapspace.
> I've read through the RSync man-pages and FAQ on the net. There it has been 
> suggested to play around with some options like blocking and non-blocking-io 
> and bwlimit. This didn't change anything.
> I furthermore did leave RSync alone a whole weekend, but no further files have 
> been processed.
> Finally I did an strace run of RSync with:
> 
> strace -f -ff -o strace.out rsync -a -v -P --delete --numeric-ids \
> 	--rsh="ssh" root at gandalf:/images/ /backup/images/ \
> 	&> /var/log/rsync/backup_gandalf_images
> 
> This outputs are sized about 300MB all together and do contain the systemcalls 
> from start up to my final kill of RSync.
> Within this output the following is repeated again and again at the end:
> 
> select(8, [7], [4], NULL, {60, 0})     = 0 (Timeout)
> select(8, [7], [4], NULL, {60, 0})     = 0 (Timeout)
> 
> I actually do have no idea what causes this Timeout.

The timeout is caused by the fact that select was called
with a timeout argument.  That process is in a loop waiting
for data from the pipe. You need to look at the process on
the other end of the pipe.

> If it yould be of any value I even can offer the strace output-files as 
> download or add any other information you may find usefull here.
> Do you have any suggestion what I can do to find the problem?
> Is there any other internet source I've missed and which I should read to get 
> rid of this problem?
> 
> Thanks in advance
> 
> - -- 
> Andre Bell <andre.bell at gmx.de>
> PGP-Public-Key: http://www.andre-bell.de/public_key.asc
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.2 (GNU/Linux)
> 
> iD8DBQE/T02RnuHMhboRh6QRAlhoAJ4t2wD4RnWCcY4MTYYPY1+ZNyEUVwCeLJ1b
> 7luF+vxxwim0CbFaBl4F9/s=
> =/qyu
> -----END PGP SIGNATURE-----
> 
> --
> To unsubscribe or change options: http://lists.samba.org/mailman/listinfo/rsync
> Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
> 

-- 
________________________________________________________________
	J.W. Schultz            Pegasystems Technologies
	email address:		jw at pegasys.ws

		Remember Cernan and Schmitt