ADSM (NETWORK BACKUP SERVER) INFORMATION
ADSM is the new Network Backup Server, which has a machine name of flotsam.acsu, and a service name of adsm.acsu, and runs the ADSM Software from IBM. This machine is the E3000 located on the machine room floor behind the IBM 3480s, directly to the right of the RF3490 tape unit The drive referred to as "MANUAL_DLT" is the desktop DLT drive sitting on top of the E3000. The disk space for flotsam/adsm is partly on top of the machine, and the rest is inside the racal rack to the machine’s immediate right. To the right of the rack is the main ADSM tape library, which contain four 3570 tape drives, and 120 tape cartridges. The library, an IBM 3575 tape library, is used for the primary tape storage of client data, and also is used to create the duplicate tape copies that are sent to the Vault in NSM. We already have about 85 tapes that cannot reside in the Library, and are currently being stored on the shelf along the back wall of Bonnie’s office (room 102A). The MANUAL_DLT drive is used for restores of the data backed up to the Q7’s before we traded them in, and for the ADSM Database backups. Mounts for this drive, as well as checkin requests for the 3575, will come up on the "adsm" window on uniops, which is started via the "adsm" pulldown menu.
DATABASE BACKUP SCHEDULE
Incremental 4:30pm Monday – Friday
Incremental 4pm Saturday
Full 2pm Sunday
DLT cartridges are used in the MANUAL-DLT tape drive for the ADSM database backups.
DSMDB1 – DSMDB7 used for Monday – Sunday respectively
DSMDB8, DSMDB9 scratch tapes-triggered database backups - red label
DSMDBA, DSMDBB scratch tapes-triggered database backups - red label
DSMDBC, DSMDBD scratch tapes-triggered database backups - red label
DSMDBE scratch tapes-triggered database backups - red label
A0001 – A0024 restores only from old Q7 – being phased out
Scratch tapes for the 3570 are stored in the black cabinets in room 116A. The library tapes for the 3570 are stored in Bonnie’s office (room 102A). The volume serial numbers for the 3570 tapes are the six-digit bar code on the tape excluding the first F.
In the event of a "triggered" database backup, a request will come up on the adsm window to mount a scratch tape in manual_dlt, like this:
ANR8326I 118: Mount DLT volume SCRTCH R/W in drive DLT7000 (/dev/rmt/5mt)
Of library MANUAL_DLT within 60 minutes.
In this case, immediately behind the machine, rack, and 3575 are DLT racks, one of which contain the tapes labeled DSMDB8, DMSDB9, DSMDBA, DSMDBB, DSMDBC, DSMDBD, and DSMDBE in red. They are right next to the other ADSM DLT tapes, A0001 – A0024. Any one of these DSMDB tapes may be selected to satisfy the above scratch mount. Normally, the DSMDBx tape for that day is either left in the MANUAL_DLT drive, or is left on top of it. Should you have to use one of the DSMDB8-DSMDBE tapes, be sure to leave it with the normal daily tape, so that it can go to the vault the next morning. It will then be returned to the rack along with that day’s daily tape, available for reuse. If the scratch tape mount is not satisfied quickly, and the database backup completed before the recovery log fills, the server will crash as happened recently. In this particular case, the log filled in 56 minutes, and even if the tape had been mounted, it may have not completed the backup before the log filled. Those of you who are familiar with logspill on the MVS system can equate this process to that. In any case, Steve Roder will be setting the trigger value to 80%, down from the 90% it was at, so that the backup can complete before the log fills. However, since the trigger will go off sooner, I expect it to go off more often, and probably in the 4-8am time range, so please try to keep an eye on the adsm window. Or even better, once the normally scheduled backup is done, and the tape is unloaded (the green light is lit on the drive to Operate the handle), premount a scratch tape. If a backup gets triggered, it will use the tape in the drive provided it is a scratch tape.
LIBRARY CHECKIN REQUESTS
As for library checkin request, here is what the requests’ look like:
ANR8308I 089: 3570 volume 1BAF7D is required for use in library 3575;
CHECKIN LIBVOLUME required within 60 minutes.
If the tape is not checked in within the 60 minutes, the request times out, the session that caused the checkin request fails, and the tape gets marked as "missing" by ADSM. Steve then needs to modify the tape to tell ADSM that the tape is not missing, and the transaction that needed the tapes has to be rerun. In most cases, these will happen during the weekdays, either from a restore request, or for server operations that USG (Unix Support Group) is running, and are not a problem. In order to checkin a tape, there has to be an available cell. Most likely you will not know if there is one or not, and doing a "query libvol"on the adsm window and counting the tapes (119) means full, as one cell has a cleaning tape in it) is not feasible. Therefore, try the checkin right away like this:
checkin libvol 3575 nnnnnn stat=private checklabel=yes
nnnnnn = volume label
In this example the nnnnnn would have been replaced with 1baf7d
Checklabel=yes requires that the tape be mounted in a drive. If no drive is available, this will fail, so then use the checklabel=no option. If you are checking in a scratch tape, then status would be scratch, not private, as in stat=scratch. If there is no room in the 3575 library unit, the checkin will fail and tell you that. You can then issue:
Checkin libvol 3575 nnnnnn stat=private checklabel=yes swap=yes
Which will cause ADSM to select a tape to eject (checkout). Once the tape that ADSM selects is out, reissue the same command, and you can insert the tape. As each phase of the process completes, ADSM will need a reply. The reply number is in the message, and is responded to via "reply nnn", where nnn is the number in the message, as in:
ANR8323I 086: Insert 3570 volume 1AAB89 R/W into entry/exit port of library 3575 within 60 minutes(s); issue ‘REPLY’ along with the request ID when ready.
.. and the reply would be "reply 86".
If checklabel=yes is specified, you might see:
ANR8447E No drives are currently available in library 3575.
ANR8426E CHECKIN LIBVOLUME for volume 1AAB89 in library 3575 failed.
…so reissue the command with checklabel=no.
Note that you can always specify "SWAP=YES", and a tape will ONLY be checked out if necessary to accomplish the checkin request.
EXAMPLE OF CHECKIN
Checkin libvol 3575 16CD1F stat=private checklabel=yes swap=yes
ANR8323I 127: Insert 3570 volume 16CD1F R/W into entry/exit port of library 3575
Within 60 minute(s); issue ‘REPLY’ along with the request ID when ready.
ANR8335I 127: Verifying label of 3570 volume 16CD1F in drive 3570B1A.2 (/dev/rmt/3st).
ANR8328I 127: 3570 volume 16CD1F mounted in drive 3570B1A.2 (/dev/rmt/3st).
ANR8427I CHECKIN LIBVOLUME for volume 16CD1F in library 3575 completed successfully.
EXAMPLE OF CHECKOUT
Checkout libvol 3575 16CD1F checklabel=yes remove=yes
ANR8434I CHECKOUT LIBVOLUME: Operation for volume 16CD1F in library 3575 started as process 260.
ANS5104I Process number 260 started.
Adsm> ANR8336I Verifying label of 3570 volume 16CD1F in drive 3570B1A.2 (/dev/rmt/3st).
ANR8322I 128: Remove 3570 volume 16CD1F from entry/exit port of library 3575;
Issue ‘REPLY’ along with the request ID when ready.
ANR8499I Command accepted.
Adsm> ANR8438I CHECKOUT LIBVOLUME for volume 16CD1F in library 3575 completed successfully.