"John R. Jackson" schrieb:
>
> >One of my clients does not backup. amcheck detects no error. What do I
> >check for?
>
> First, see if it ever gets done by looking at /tmp/amanda/sendsize*debug.
> Also, watch for the dump process to go away.
After a few hours, all the processes are have gone away. See sendsize
log below.
> What OS is the client?
FreeBSD 4.2-Stable
> Do you have an up to date version of dump (note:
> I don't necessarily know what constitutes "up to date", but that's been
> a problem with some free Unix versions in the past)?
I guess so. It has worked before for a few weeks.
> Normally dump is pretty quick at getting estimates (as compared to
> GNU tar). But if, for some reason, it is working OK on your system
> but slowly, you can increase etimeout in amanda.conf to compensate.
> See amanda(8).
sendsize: debug 1 pid 45023 ruid 1003 euid 1003 start time Thu Jul 19
16:08:41 2
001
/usr/local/libexec/amanda/sendsize: version 2.4.2p2
[snip]
sendsize: pid 45023 finish time Thu Jul 19 20:27:07 2001
Comparing this to the total number of filesystems, sendsize is
outrageously slow. Any idea why? That's four filesystems, only the last
(big) one is this slow.
# df
Filesystem 1K-blocks Used Avail Capacity Mounted on
/dev/da0s1a 254063 35400 198338 15% /
/dev/da0s1f 3062628 1515733 1301885 54% /usr
/dev/da0s1e 254063 5968 227770 3% /var
/dev/ad0s1c 38788812 26828985 8856723 75% /u
# ls ls -l /tmp/amanda/killpgrp.20010719161*
/tmp/amanda/killpgrp.20010719162119.debug
/tmp/amanda/killpgrp.20010719162123.debug
/tmp/amanda/killpgrp.20010719174350.debug
/tmp/amanda/killpgrp.20010719190143.debug
Anyhow, may this be related in any way with the error message in
# cat /tmp/amanda/killpgrp.20010719162119.debug
killpgrp: debug 1 pid 45070 ruid 1003 euid 0 start time Thu Jul 19
16:21:19 2001
/usr/local/libexec/amanda/killpgrp: version 2.4.2p2
error [must be invoked by operator]
killpgrp: pid 45070 finish time Thu Jul 19 16:21:19 2001
Just having reinstalled amanda client, the amanda processes on both
clients are still run by user "backup" instead of "operator". Both
machines have similiar amounts of data to backup. One of both clients
succeds, the other does not.
Help greatly appreciated.
-Christoph Sold