I am having problems backing up a client in different building (on a
different subnet). The strange thing is this worked for over a year and
suddenly started failing. It does not appear to be an acl or firewall
issues, because sometimes it will work. I set up a separate config just
for this host so I could test, and find that I can kick of amdump, it
appears to start on the client (I get the debug files), but at some
point it dies and I get an amandad debug with the following that the end:
amandad: time 126.413: dgram_recv: timeout after 10 seconds
amandad: time 126.413: waiting for ack: timeout, retrying
amandad: time 136.413: dgram_recv: timeout after 10 seconds
amandad: time 136.413: waiting for ack: timeout, retrying
amandad: time 146.413: dgram_recv: timeout after 10 seconds
amandad: time 146.413: waiting for ack: timeout, retrying
amandad: time 156.413: dgram_recv: timeout after 10 seconds
amandad: time 156.413: waiting for ack: timeout, retrying
amandad: time 166.413: dgram_recv: timeout after 10 seconds
amandad: time 166.413: waiting for ack: timeout, giving up!
amandad: time 166.413: pid 6926 finish time Wed Sep 22 10:04:29 2004
If I stop the processes on the server, and run amcleanup, and restart
amdump, with _no_ changes to anything, the dump will complete normally,
most of the time. Other times it will fail with the same type of output
in the debug file and I'll have to repeat.
The client and server are running linux, (debian testing) amanda 2.4.4.p3.
Needless to say, the intermittancy of this making troubleshooting
difficult.
The clients on the same subnet as the server back up normally.
--
George Kelbley System Support Group
Computer Science Department University of New Mexico
505-277-6502 Fax: 505-277-6927
|