Hello everybody,

We have the following procedure to backup the ADSM Database on our
VM/ESA System.

Our backup window (incremental + Db2/6000) usually end around 5:00 in
the morning and is almost always going to a disk storage pool.

Everyday at 6:00 am we start migration (UPDATE STG stgdsk HIGHMIG=0) to
make sure that every file is copied on stable storage (Cartridges in
this case).

No backup activity is supposed to happen during the day and at noon we
run the following procedure:

Disable Server
Kill all active Node sessions
Halt The server
Backup all Database Volumes And Log Volumes with Vm:Backup Software
Restart the Server

It's faster than ADSM DUMP DB, and the restore is much much much more
faster than a LOADDB ..

We might have the following problem when the database crashes between
the end of the daily migration (in the morning) and the start of the
DB Backup : volumes have been updated !!

In fact we think we can control the situation and come back to a stable
state (the state one day in the past) by looking in the ADSM server log
file (The VM one not the Activity Log .. because the latter one is
stored in the Database) for all volumes written during last migration (
with messages like :ANR5208I Dismounting volume 013966 (updated).) and
pass an AUDIT VOL xxxxx FIX=YES for each of them.

At this point do you see any potential problem ?

The problem I cannot handle is when we reclaim partially empty tapes:
The operation is destructive for the data and to my knowledge there is
no way of replaying back the reclamation ...

Do you have any suggestion/remark except not doing reclamation .

TIA, Michel



--
Michel Gaillard
Michel Gaillard
DSI/DPI
Alcatel CIT Velizy France
E-Mail: Michel.Gaillard AT vz.cit.alcatel DOT fr
Voice : (33.1) 30 67 32 56