>I am working on the amanda backup project now.  ...

Welcome!

>When enter "su amanda -c amdump daily", my dump for the localhost
>/scsi9-data failed... I check the logs and debugs, it said "FAIL
>planner localhost /scsi9-data 0 [Request to localhost timed out.]

As Mark Schoonover said, please don't use "localhost" for any of your
client names.

Now on to the real problem.  This message means it took too long for
your client to respond to planner with the estimate.  You didn't post
your sendsize*debug file, but we can see part of what happened from
the corresponding amandad*debug:

>amandad: debug 1 pid 11580 ruid 33 euid 33 start time Fri Mar 22 09:07:01 2002
>...
>got packet:
>--------
>Amanda 2.4 REQ HANDLE 000-C03A0608 SEQ 1016806021
>SECURITY USER amanda
>SERVICE sendsize
>OPTIONS maxdumps=1;hostname=localhost;
>GNUTAR /scsi9-data 0 1970:1:1:0:0:0 -1 exclude-list=/usr/lib/amanda/exclude.gtar
>...
>amandad: pid 11580 finish time Fri Mar 22 10:29:09 2002

Note the start and end time.  It took about one and a half hours to do
this estimate.  The default value for etimeout in amanda.conf allows
five minutes per disk per client.  You missed that by a lot :-).

I don't know why GNU tar took so long to do the estimate.  That's not
an Amanda problem.  However, you might check what version of GNU tar
you're using.  It needs to be either 1.12 plus the patches on the
www.amanda.org web page, or else 1.13.19 or later (from alpha.gnu.org).
If --version reports just "1.13", that's a *very*, *very* bad version.

Or it could be that the particular data on the file system is just really
bad for tar to deal with.  For instance, if there are a bajillion files
it can take tar a long time to process each one.  If that's the case you
may just have to increase etimeout a lot in amanda.conf.

>Zhen Liu

John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]

Reply via email to