ADSM-L

[ADSM-L] moving a library manager to a new server - a story

2008-04-28 10:58:46
Subject: [ADSM-L] moving a library manager to a new server - a story
From: Richard Rhodes <rrhodes AT FIRSTENERGYCORP DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Mon, 28 Apr 2008 10:57:03 -0400
We have 2 library manager TSM instances which resided on
servers with other TSM instances.  We decided to move these tiny
library manager instances to their own servers/lpars to isolate them.
We did this last Saturday.

We moved the library manager instances by backing them up and
doing a restore onto the new server, then moving the DNS alias
which points to these instances to the new IP addresses.  This all
went VERY well, taking all of 15m.

After we had them back up and running on the new servers we
started to see strange tape behavior.
 - library client would request a tape mount to library mgr
 - library mgr would perform the mount
 - library client would spin the tape
 - library client would request the dismount to the library mgr
 - library mgr would dismount the tape
 - library client never received notice that the dismount occurred!!!!

Very quickly our library clients had LOTS of tape mounts that were not
really being used any more. Tape usage quickly ground to a halt!
The clients and mgr were hopelessly out
of syc!!!  We finally found errors in the library mgr log indicating
a validation error when trying to talk to a library client.
STUPID . . . the server-to-server passwords needed syncing!!!!
We needed to simply do a "update server xxx forcesync=yes" for all
our instances.  IBM support told us we should do this on both the
lib mgr and client sides even thought the lib client side was working.

The confusing thing about this was that communication
seemed to be working just fine.  Support told us that the pwd sync
we needed was on the lib mgr since that is what we had restored, but we
really should do the forcesync on both sides just to be sure.
He indicated that all the server-to-server communications
I listed above, except for that last step, are initiated by the
lib client, and thus worked. The last step for the dismount
is initiated by the lib mgr and thus failed.  Our big mistake was
assuming that server-to-server communications was working when
we saw tape mounts and tape usage at the client.

Rick




-----------------------------------------
The information contained in this message is intended only for the
personal and confidential use of the recipient(s) named above. If
the reader of this message is not the intended recipient or an
agent responsible for delivering it to the intended recipient, you
are hereby notified that you have received this document in error
and that any review, dissemination, distribution, or copying of
this message is strictly prohibited. If you have received this
communication in error, please notify us immediately, and delete
the original message.

<Prev in Thread] Current Thread [Next in Thread>