We're running amanda-2.4.5p1 server on a white box under debian/sarge.
We're backing up from 8 hosts without any problems beyond an occasional
client timeout. Working clients are running debian/sarge, RHAS 2.1,
and AIX 5, and the same 2.4.5p1 release of amanda.
We have a CentOS 4.1 client that is failing consistently with client-side
timeouts. The client starts, dumps 35 MB to the server, then times out.
The partition being dumped is around 1 GB. The time out occurs when
iptables on the client is turned on, and also when it is turned off.
Gtar is 1.15.1.
Timeout messages in /tmp/amanda/sendbackup.* on the client look like
this:
---
sendbackup: debug 1 pid 18885 ruid 518 euid 518: start at Fri Feb 10 02:20:02
2006
[snip]
sendbackup: time 0.001: got all connections
[snip]
sendbackup-gnutar: time 0.062: doing level 0 dump from date: 1970-01-01 0:00:00
GMT
[snip]
sendbackup-gnutar: time 0.064: /usr/local/amanda/2.4.5p1/libexec/runtar: pid
18889
sendbackup: time 958.026: index tee cannot write [Connection timed out]
sendbackup: time 958.026: pid 18887 finish time Fri Feb 10 02:36:00 2006
sendbackup: time 958.026: 125: strange(?): sendbackup: index tee cannot write
[Connection timed out]
sendbackup: time 958.028: error [/usr/local/amanda/2.4.5p1/bin/gtar got signal
13]
sendbackup: time 958.028: pid 18885 finish time Fri Feb 10 02:36:00 2006
---
We reduced the keepalive interval on both client and server but didn't
see a change.
Any suggestions would be welcome.
Thanks,
Roy Heimbach
--
Roy Heimbach <royh AT hpc.unm DOT edu> / 505-277-8348
User Services / HPC@UNM / University of New Mexico
|