Tape Device Reference Count

[The Brit]

Can someone enlighten me as to what the "Reference Count" on a Tape device refers to?

I have a backup running from a BL860c (8.3-1H1) to a HP Ultrium 4-SCSI drive in an MLT G3 tape library.

I noticed that the Reference Count was 4. I thought this was the number of allocated devices/processes, however if that's the case, the number seems to be a bit high.

(I am bringing this up because a long running script has suddenly started running very slow)
I do not share the tape library or drives, and there are no contending processes on my blade.

I do however share the SAN infrastructure.

I was wondering if the reference count might indicate some "non-existing process" allocation.

Dave.

[Volker Halle · Accepted solution]

Dave,

a device reference count represents of number of channel assigned to or allocations of a device from the local OpenVMS system, which prevent the deletion of that device. Only if the reference count of a device reaches ZERO, that device could (theoretically) be deleted, because noone else is using it.

In case of backup to a tape, 4 seems to be a 'normal' reference count:

- the allocation of the tape to the process
- the mount
- 2 channels assigned from backup

You would only have a problem, if the device is NOT allocated and NOT mounted and the reference count would not be zero at that time. This has happened from time to time in the past. There have been various attempts to solve problems like this. Some code error path could cause the reference count to not be correctly decremented (this would be a software problem). Then you would have to reboot or use some priviledged program to just zero the reference count.

Volker.

[The Brit]

Thank you Volker,

Your response was exactly what I was hoping for, i.e. educational.

It looks like our problem is with the tapedrive.

Dave.

[Hoff]

Tapes and hard disks are comparatively slow devices.

The usual triggers for slower backups are input or bus (SAN or SCSI) contention, fragmentation, lack of caching, quotas, media errors (or tape head contamination), hardware errors (host, bus, disk, tape, termination), and the glacial speed of hard disks as compared with more modern storage. There's also the case where the volume of data has markedly increased; that's always worth a check. Also lack of host memory. SAN devices are problematic because you can't see the contention.