ADSM-L

SBTIO errors, failed Oracle backups and lesson learned

2003-03-11 15:23:50
Subject: SBTIO errors, failed Oracle backups and lesson learned
From: Jeff G Kloek <Jeff.Kloek AT IPAPER DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Tue, 11 Mar 2003 13:26:43 -0600
Our DBA brought the following errors to us stating that his Oracle backups
were failing, and that these errors were showing up in the sbtio.log:

SBT-145558 02/14/2003 00:14:38 error2.cpp(160): sbterror(): Enter
 rc       : 0
 dsmHandle: 1

SBT-145558 02/14/2003 00:14:38 error2.cpp(168): sbterror()
 ctxID    : 145558
 rc       : 0
 dsmHandle: 1

SBT-145558 02/14/2003 00:14:38 error2.cpp(207): sbterror()
 ctxID                 : 145558
 Error Code            : 0
 Error Message, Native : ANS0278S (RC157)  The transaction will be aborted.
 Error Message, Unicode: <NULL>

SBT-145558 02/14/2003 00:14:38 error2.cpp(213): sbterror(): Exit
 rc       : 0
 dsmHandle: 1

Checking the Activity Log for the same time frame, we found the following:
02/14/03   00:14:38      ANR8447E No drives are currently available in
library
                          ADSM3494.
02/14/03   00:14:38      ANR1401W Mount request denied for volume 402545 -
mount failed

02/14/03   00:14:38      ANR0525W Transaction failed for session 197935 for
node
                          ORAPRD02_ORA (TDP Oracle AIX) - storage media
                          inaccessible.

SBT-145558 02/14/2003 00:14:38 error2.cpp(207): sbterror()
 ctxID                 : 145558
 Error Code            : 0
 Error Message, Native : ANS0278S (RC157)  The transaction will be aborted.
 Error Message, Unicode: <NULL>

While we are NOT doing library sharing, we do have 3 different TSM servers
talking to the same 3494 library.
Each has it's own drives physcially connected, and none of the servers
knows anything about the other server's
tape drives. What we had set for each of our 3590tape devclass was
"mountlimit=drives". What we learned is that the tsm call to find an
available tape drive is an actual call to the LIBRARY to see how many
drives are available, and that doesn't bother to check the server's local
count of available tape drives. Our library has 32 drives in it, but the
server with this problem only had 13 drives connected. So, when TSM went to
try mounting another tape, the above errors occurred. The answer is to set
the tape devclass for each server to the exact number of drives available
to that server.
Just thought it'd be worth sharing.

<Prev in Thread] Current Thread [Next in Thread>
  • SBTIO errors, failed Oracle backups and lesson learned, Jeff G Kloek <=