Hi TSM'ers,
We are trying to build a solution where Veritas BackupExec would be
using TSM as a data repository, but are facing something very strange :
although we respected all recommendations while binding applications
together, it seems Veritas backups are filling up our library extremely
fast.
We have a 3575 l18 library on a new test TSM (4.1.0) sytem, only used by
Veritas and an additional NT system with normal TSM BA client, and after
2 days of utilization, already have 6 primary pool tapes filled up with
veritas backups (Veritas only backups 1 small Exchange server !).
Doing a "q con" on the tapes used by veritas shows something like
(snippet of it) :
BACKUPEXEC Bkup LHR2K010 \{A92E0AEF-C7B5-47B1-879E-03D6AB71AD8- F}\
12013
BACKUPEXEC Bkup LHR2K010 \{A92E0AEF-C7B5-47B1-879E-03D6AB71AD8- F}\
12014
BACKUPEXEC Bkup LHR2K010 \{A92E0AEF-C7B5-47B1-879E-03D6AB71AD8- F}\
12015
BACKUPEXEC Bkup LHR2K010 \{A92E0AEF-C7B5-47B1-879E-03D6AB71AD8- F}\
12016
BACKUPEXEC Bkup LHR2K010 \{A92E0AEF-C7B5-47B1-879E-03D6AB71AD8- F}\
12017
BACKUPEXEC Bkup LHR2K010 \{A92E0AEF-C7B5-47B1-879E-03D6AB71AD8- F}\
12018
doing "show version BACKUPEXEC LHR2K010" shows nothing, and although
those tapes are having thousands of files written on them, result of
expiration process shows :
12/16/01 07:01:26 ANR0812I Inventory file expiration process 96
completed: examined 18 objects, deleting 0 backup objects, 0 archive
objects, 0 DB backup volumes, and 0 recovery plan files. 0 errors were
encountered.
How is it possible expiration only went thru 18 objects, as we have
thousands of them on tape ? I am missing something ? Shouldn't we
normally have as many objects examined, as objects in TSM DB ?
Now something worse, if I do a "select * from backups", I have (snippet
of it) :
NODE_NAME: BACKUPEXEC
FILESPACE_NAME: LHR2K010
STATE: ACTIVE_VERSION
TYPE: FILE HL_NAME: \{064A8A12-D91B-4936-BBC8-A0367F94B943}\
LL_NAME: 1000
OBJECT_ID: 51147
BACKUP_DATE: 2001-11-16 01:02:18.000000
DEACTIVATE_DATE:
OWNER: LHR2K010
CLASS_NAME: MC_VT
of course thousands of entries like this one ....
I am really confused, on one hand, lots of tapes filled, and lots of
entries in the "backups" table, and on the other one, expiration that
examines very few records, and a TSM db that is near to be empty ...
Is it due to the fact Veritas uses a specific "backupexec_pi" storage
pool for storing positional information, to locate it's data ?
If yes, should I avoid doing offsite copies from this data, because it's
never going to be expired ?
No idea at all, but if anybody has been building such an application,
all comments, explanations and setting descriptions will be REALLY
WELCOME !!!!!!!!!
Thanks in advance .
Arnaud
|