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

List:       amanda-users
Subject:    Selfcheck timing out using >2GBpatched amanda 2.2.6
From:       Chela Kunasz <chela () jila02 ! Colorado ! EDU>
Date:       1996-06-30 17:50:32
[Download RAW message or body]

Hello Amanda Users:

I have been using amanda (2.2.6) for over a year with good success.
My server is an Ultrix-based DECstation, I'm using a DLT2700 stacker
(and also a DLT2000) and have Digital UNIX (OSF/1), Ultrix, and SunOS
clients.  Everything has been working very well.

I need to be able to backup filesystems >2GB.  So I acquired the patches
users). I installed these patches in an alternate directory which was a
copy of source for my current amanda binaries.  I had some modifications
in my current source, standard fixes for Ultrix and extensions of time
needed for selfchecks.  Simple things which produced working code.
None of these small mods had to do (as far as I could see) with the
patches for handling >2GB filesystems.  I changed some entries in the
amanda.conf files and made a new simple disklist.  I deleted the old
databases and logs from the new directory and made appropriate directories
to hold these, corresponding to my new entries in my new amanda.conf.

Then I did a make clean, make install, and created necessary softlinks
for my server in the new directory which should support the modified
amanda.  Then I did a make clean (still on the server) and did a make
client and make install-client, having suitably modified my options.h
and config.h for the server-as-client.

I made a script which would toggle between the two amandas on my
server.  This script mainly changes the entry in /etc/inetd.conf
and does a kill -HUP PID# and switches the links I have in /usr/local/bin
for things like amdump, amrestore, etc.  I started AMANDA up with no
tape in the drive, just to see if it would start.  All fine and dandy.
I deleted the database and logs from that check, but left the directories
ready for a new restart.

Then I installed the patched amanda client software, from scratch, on a
Digital UNIX client.  This went fine.  I created a script to toggle
between the new and old amanda on this computer as well.  This script
is also simple; it does the switch to a new /etc/inetd.conf entry,
a kill -HUP PID# and switches links in /usr/local/bin.  Having switched
both server and client to the new setup, I did an amcheck config
and the server checked out okay as a client, the tape and holding
disk were fine, but amcheck said the selfcheck timed out for the
DEC UNIX computer.  I tried killing the inetd on the client and
restarting it. That did not help.  Then I switched back, using scripts
such as those described above, to switch back to my currently running
amanda, and everything was fine.  For the last two days, the old
amanda is working fine.  Since no changes were made to .rhosts, or
/etc/services or the filesystems to be backed up on my server and
the test client, I am very confused why the new amanda does not work,
i.e. why the selfcheck times out.  I did an rsh ...path/selfcheck client,
and checked in its /tmp directory and found that there was an entry
made in the client's /tmp/selfcheck-debug file.  All filesystems to
be backed up are in /etc/fstab.  When I run the new amcheck, it is
clear that the new amanda.config is being used, because the hosts
which are checked and the tape label is as expected for amanda.conf
in my new directory.  I'm puzzled.

Has anyone any clues on what I am doing wrong or have left out?
I'm especially interested to hear from anyone who has this working
on an Ultrix server.  The server actually seems fine and ready to
back up its own filesystems with the new amanda system.

THANKS for any suggestions!
-Chela Kunasz
 University of Colorado
 JILA-Campus Boxx 440
 Boulder, CO 80303-0440
 Chela.Kunasz@jila.colorado.edu

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