Tape drive goes in hung stage..

[sandip debnath]

Hi All, Good morning,
I am facing a problem with a tape drive in one of the VMS servers having the OS version 8.2.

After deleting the entry of any backup which was submitted and ongoing earlier,we are able to see the status of tape drive as...

Device Device Volume Free Trans
Name Status Label Blocks Count

FINUKF$MKA600: Online
FINUKG$MKA600: Mounted 1WP26
dismount
alloc foreign


... and it takes more than 10 hours to come to normal change. Any sugesstions will be highly appreciated.

Sandip

[Ian Miller.]

Can you supply the result of
SHOW DEVICE/FULL MKA600:
and
MCR SYSGEN SHOW TAPE_MVTIMEOUT

What sort of tape drive is it.

"deleting the entry of any backup which was submitted and ongoing earlier"
Do you mean there was a batch job performing a backup to this tape drive and you did DELETE/ENTRY=x for that job?

[sandip debnath]

Hi ian, yes you are right.. it is same DELETE/ENTRY=X ...

here are two outputs..

SYSGEN> SH TAPE
Parameter Name Current Default Min. Max. Unit Dynamic
-------------- ------- ------- ------- ------- ---- -------
TAPE_ALLOCLASS 0 0 0 255 Pure-numbe
TAPE_MVTIMEOUT 600 600 1 64000 Seconds


and ...

sh dev FINUKG$MKA600:/fu

Magtape FINUKG$MKA600:, device type COMPAQ SDLT320, is online, allocated,
mounted foreign, volume is marked for dismount, record-oriented device,
file-oriented device, served to cluster via TMSCP Server, error logging is
enabled, operations being canceled, device is busy, controller supports
compaction (compaction enabled).

Error count 117 Operations completed 104875704
Owner process "BATCH_5926" Owner UIC [SEMA_SYSTEM,PRIV_BATCH]
Owner process ID 2025BC7A Dev Prot S:RWPL,O:RWPL,G:R,W
Reference count 3 Default buffer size 512

Volume label "1WP26 " Relative volume no. 0
Record size 0 Transaction count 1
Mount status Process Mount count 0
ACP process name ""
Density SDLT320 Format Normal-11

Volume status: no-unload on dismount, odd parity.
even the process ID is showing as ..

sh proc/id=2025BC7A
%SYSTEM-F-SUSPENDED, process is suspended

[Karl Rohwedder]

Never a good idea to kill a process wich has mounted tapes...

Have you tried a resume on the process, e.g.
SET PROCESS/RESUME/ID=...?

You may have a closer look at the process
with

$ ANA/SYS
SDA>SHOW PROC/CHAN[/INDEX=PID]

but I fear it has a DELETE pending.

regards Kalle

[John Abbott_2]

Is this an Integrity or Alpha server ?
Have you installed any other patches ? ($ PROD SHO HIST)

$ pipe anal/image sys$common:[sys$ldr]sys$pkwdriver.exe; | sear sys$input identification:
- Please post the results

Was this a backup using vms backup, copy, or another tool ? block size used ?

System spec (I'm interested in h/w model and cpu speed).

Thanks
John.

[Ian Miller.]

Power cycling the drive may help.

Re
$ ANA/SYS
SDA>SHOW PROC/CHAN/INDEX=PID

I would start without the /CHAN
$ ANA/SYS
SDA>SHOW PROC/INDEX=PID

as SHOW PROC/CHAN can hang your process depending on the state of the target process.

[Jim_McKinney]

> Error count 117

Does this increment with each job? Unless you already have a satisfactory explanation for the 117 you should examine the system error log.

[Volker Halle]

Sandip,

your SHOW DEV/FULL FINUKG$MKA600: reports


operations being canceled, device is busy


'Device is busy' means, that there is a physical IO operation outstanding to the device. It's hung somewhere in OpenVMS or in the tape.

If the IO operation finishes, the process will probably disappear and the device will be freed. If this takes 10 hours, this seems to imply, that there are some really long timeouts involved here. Most likely in the SCSI port or class driver.

Volker.

[Shardha]

Dear Sandip,

If possible, reboot the system. this will slove the problem.

Shardha

[John Abbott_2]

Hi Sandip, was your backup job hung/idle/RWxxx prior to you aborting it ?

If so, I'd appreciate it if you could reply to my post.

Thanks.
John.

[Jim Lahman_1]

I just had a tape timeout error. It turns out that a few of the pins on the SCSI cable were recessed than the other pins. It looks like the Alpha scsi port has some tight scsi pin connectors so when the cable if jiggled or pulled out, the pins look ok at an initial glance.

I replaced the cable with the spare that we have and it fixed the problem.