Summary
Exchange base backup fails to append to available appendable tapes.
Symptoms
The following error message is shown:
10.10.10.10 ssjobhnd Mon Oct 31 16:29:03 2005 SNBJH_3853E *** Unable to get media assignment for job 1130801338 (node mynode, disk EXCH$mynode01:). You may have run out of free tapes. Please check the media pool. ***
The data manager log (ssdatmgr.*) lists a number of tapes that cannot be used with the "Not(volser)" nomenclature, for example:
Time(Oct 31 16:29:03):TID(1965): | tCtx(1130801353): using Media Pattern : MediaJobPattern(poolName=Exchange,statusMask=10("Empty"|"Appendable"), Not(CTK055),Not(CTK033),Not(CTK068),Not(CTK015),Not(CTK065),Not(CTK035), Not(CTK042),Not(CTK007),Not(CTK002),Not(CTK008),Not(CTK058),Not(CTK052), Not(CTK104),Not(CTK201),Not(CTK154),Not(CTK062),Not(CTK134),Not(CTK059), Not(CTK027),Not(CTK170),Not(CTK113))
Resolution
This restriction is due to the possibility of multiple backup jobs that backup the same database.
Consider: More than one backup job backing up different database instances. For example, Job 1 backs up Exchange instance 1 and Job 2 backs up Exchange instance 2.
Then, suppose some sequence of backup events leads to the following ordering of information on tape
Job 1
--------
Base backup -> Tape 0
Incr/Diffr backup -> Tape 1
Job 2
--------
Base backup -> Tape 1
Incr/Diffr backup -> Tape 0
Though both databases are restorable separately, you would not be able to restore both databases together in the same backup job since the tape involved in the base backup for one or the other would not be loaded first. It is for this reason that the restriction is in place.
This problem is avoided if a separate media pool is used for each instance of MS SQL Server or each storage group of MS Exchange.