On Wed, Apr 19, 2006 at 10:02:36AM -0500, Fran Fabrizio wrote:
> 
> I recently upgraded the server side to 2.4.5 and a subset of my hosts 
> are consistently timing out on estimates each night, even after doubling 
> the time limit for estimates.  The affected clients are running 
> 2.4.4p2.  Is there a known issue here?  Some clients also running 
> 2.4.4p2 are working, so I am not sure that is the problem.
> 
> amcheck reports no errors on any of the hosts.
> 
> I would like to debug this by not having to wait for the nightly backup 
> to fail but by just testing one backup target, tweaking, and testing 
> again.  I could define another config altogether, but then I have to 
> load tapes for it, etc... is there an easy way to just force a run, but 
> of only one target, without adversely affecting things when I go back to 
> the full run?

from amdump manpage:

   SYNOPSIS
      amdump config [ host [ disk ]* ]*


I'd probably label a new tape and write a shell script to
do things like this:

   make sure your desired tape is loaded
   reset the entry for that tape in tapelist file to whatever is "new"
   run the appropriate amdump command for that DLE

I don't know of a one-time to set "no record".  So I'd make up
a new dumptype for testing that is no record and no index.

Be sure to change dumptypes when done testing.


> 
> Further, any other speculation about why estimate timeouts would start 
> happening in a subset of clients but not other clients in a config where 
> the only change is a server upgrade from 2.4.4 to 2.4.5 which 
> corresponded with a move from redhat 9 to fedora core 5 for the server?  
> Thoughts about the best way to debug?

selinux turned on in FC5?  probably not in rh9.

does something in that have to be activated or configured
for amanda to work?

-- 
Jon H. LaBadie                  jon AT jgcomp DOT com
 JG Computing
 4455 Province Line Road        (609) 252-0159
 Princeton, NJ  08540-4322      (609) 683-7220 (fax)