Very good question, Of course IBM would like to sell us some more
software BUT!
Now I still have much room for improvement and really don't have
things finished up myself but where I would like to be is:
I've got 5 AIX ADSM servers and 2 MVS (others deal with the MVS so
they can fight their own battle, OH GOD THOUGH... I back up my AIX
ADSM servers to MVS) I'm trying new system configs on the Chicago &
Houston ADSM boxes but the Tulsa boxes are pretty close to the
same...RS/6000, 3494 w/ 3590's (4), & 4 drawers of 4.5GB SSA drives.
To ensure continued backups I have some vacant licenses on each
different server to immediately point clients to, at least very
critical clients. That covers continued backups/archives... NOW if
they need a restore/retrieve... If the environments were identical I
could 1) pray for 40 hour days 2) fake out another environment by
putting the broken server's DB on it (I might get burned with ADSM &
associated licenses when it wakes up on a processor with a different
serial numbers and the such...) 3) Library inventories will be all
our of whack BUT if I'm just trying to cover an extream emergency to
restore a single client I would pull his set of tapes from the one
ATL, check them in the ATL in this temp environment, export out the
data, drop that ADSM environment, restart the REAL ADSM for tha
machine, import that client's data... 4) Now things are sort of normal
and that adsm server could continue standard processing while working
the restore of the client from the broken adsm environment. 5) get a
ski mask and a gun to obtain a replacement machine for the "just
melted the insides out" box 6) go through the motions and get back
the broke environment via reattach hardware/or have all new... bring
back the database, sequential vol hist, da, da, da all that good stuff
from the manuals on how to recover from blah...7) point the clients
back to that server 8) finish up by exporting backups made to other
ADSM servers and import them into this one...
Oh, has anyone tried having a client on one server, pointing that
client to another environment for a couple days, pointing the client
back, doing an export on "the other" server and then importing that
data into the origional one? ? ? ? I would like to know how it did
with expiration processing... such as if you import a file with a
version date that is older than all the copies on/in the current
environment would it keep that as new (since it was the most current
one brought into the server) or would it expire it since it was the
oldest? In other words has anyone noticed if expiration/versions
depend on the date of the file on the client or the date when it
showed up on the server..... Sorry if I've missed something in my
readings and this is a stupid question...
later
Dwight
______________________________ Reply Separator _________________________________
Subject: Database Backup and Server Export
Author: ADSM-L at unix,mime/DD.RFC-822=ADSM-L AT VM.MARIST DOT EDU
Date: 2/19/97 9:15 AM
Hi there,
Currently, I am working on designing the recovery procedure of ADSM server,
in case:
1. The database is corrupted.
2. The ADSM server is down and is unaccessable substantial period of
time.
and would like to get some experienced suggestions from the ADSM community.
We are running ADSM server version 2.1.0.10 on AIX 3.2.5 server with STK
9710 DLT Tape library connected to a SUN Solaris machine, running ACSLS
software, with drives mounted to the ADSM server. The database and the disk
backup pools are defined on a sepertely connected hard drives, almost 30GB
of capacity, with a seperate volume group, vgadsm.
For the first case, i.e, if the database is corrupted, what are the good
practices? I have gone through the Admin Guide and, Chapter 14, discusses
in detail some options for recovery. The Point-in-time recovery along with
volume and device configuration backups seems an obvious choice, but any
feedback would be greatly appreciated.
For the second case, i.e. if the server is down and cannot be repaired,
suppose I have a backup server, what are the steps in recovering to the
other server. I would install the server software on the backup server, I
will hook up the hard drives which are connected to the source server and
are still working, recreate the volume group vgadsm and other filesystems,
and then how would I proceed with the recovery? Should I use the database
backups or should I use export server tapes, discussed in Chapter 13 of the
Admin Guide?
Any suggestion regarding both these questions would be greatly appreciated.
Thanks,
Naveed.
|