Multiple sybase dumps to DLT8000

[Robert Ungless]

Hi

We are presently experiencing a problem where dumping multiple sybase databases to the same DLT8000 tape will randomly request a tape change as though the tape is full.

We are dumping 16 databases totalling under 15GB of data, using a simple script to invoke the sybase dump command. The first dump command is issued 'with init' and the last 'with unload'. Each command contains a capacity of 40GB i.e. the native capacity of a DLT8000.

An example dump command would look like:

dump database referencedb to '/dev/rmt/0mn' capacity=41943040, file='referencedb'


If one of the dump commands reports the problem running sp_volchanged to abort that dump will allow the remaining dumps to complete successfully, so the tape is clearly not full.

This problem is occurring on two different machines (an N4000 and an L2000) over 3 different DLT8000 drives, and a range of tapes , and will occur even on brand new media.

The same problem is present on two sybase dataservers both 11.9.2.4 although the problem is also present on all earlier versions of 11.9.2 on these systems.

The Sybase backup server never logs a true tape error, just requests the tape be changed.

There are no SCSI/hardware errors being logged in syslog.

From the O/S commands such as tar, fbackup and mt perform normally.

The latest stape jumbo patch has been applied.

Sybase tech support have so far offered little by way of help.

Has anyone else seen this or have any further suggestions ?

Many thanks.

Robert