[prev in list] [next in list] [prev in thread] [next in thread] 

List:       amanda-hackers
Subject:    amcheck ACK timeout on clients with a lot of ZFS file systems
From:       Darin Perusich <Darin.Perusich () cognigencorp ! com>
Date:       2009-04-20 20:49:41
Message-ID: 49ECDFE5.3040201 () cognigencorp ! com
[Download RAW message or body]

I'm performing amcheck's against a client host that has a lot of local
file systems and when the disklist has more then 74 file systems the
selfcheck request fails with timeout waiting for ACK. These are ZFS file
systems and I've defined a amzfs_sendrecv application-tool and
user-zfs-sendrecv dumptype as described in the amzfs-sendrecv man page.
If I change the DLE entries to use GNU-TAR the amcheck completes
successfully. I've tried increasing ctimeout=120 seconds in amanda.conf
or on the command line but the value doesn't appear to be being honored,
at least when there are 75+ file systems being checked.

The client doesn't generating any logs other then the amandad debug file
and the server only creates the amcheck debug log. I can send these logs
to the list or developers.

Both the client and server are running Amanda 2.6.1p1, the server is
running Solaris 9, and the client is Solaris 10 x86, I haven't testing
on sparc yet but will shortly.

The client server will have upwards of 250+ ZFS file systems once it
goes into production.

-- 
Darin Perusich
Unix Systems Administrator
Cognigen Corporation
395 Youngs Rd.
Williamsville, NY 14221
Phone: 716-633-3463
Email: darinper@cognigencorp.com
[prev in list] [next in list] [prev in thread] [next in thread]