migration queue databased from OpenVMS 7.3-2 to OpenVMS 8.3

[Desmond Or]

I have update my ALpha server from OpenVMS 7.3-2 to OpenVMS 8.3. Before the upgrade, I stopped the queue manager and a lot of batch command files had been submitted and pending on the batch queues. After the upgrade, I started the queue manager and let all batch command files to go. But all command files had finished unsuccessfully with an error "File not found". Does anybody know whether I need to convert the queue database after the VMS upgrade? According to my experience, I don't need to convert the queue database. Is that true? Please advise what cause the problem.

[Robert Gezelter]

Desmond,

I do not recall any special issues. Precisely what was involved? Was the update done in-place?

- Bob Gezelter, http://www.rlgsc.com

[Desmond Or]

Yes, I successfully implement the upgrade and the patches. It looks everything is good except this issue.

[Richard W Hunt]

If the disks on which the batch files resided weren't mounted when the queues were started, that would do it. When you started the queues, had you rebooted into OVMS 8.3 or did you start the queue mangler before doing the reboot?

Don't know about your system, but on my system, my user files aren't on the system disk and I don't mount my user disks during an upgrade. Just an idle thought...

[Desmond Or]

Richard,

Yes, the data disk is different to the system disk. I verify with the system startup file that the data disk has been mounted before queue start.

[Jon Pinkley]

If the data disk was mounted before the queue manager was started, then the most likely cause is that the physical device name of the data disk has changed. Several things can cause this even if the physical device itself has not changed, for example using port allocation class, changing the allocation class of the system, or converting the data disk into a shadow set.

For example:

$ submit/hold/noprint/log=sys$scratch disk$itrcdemo:[demo]showtime/retain=always /que=sigma$oper
Job SHOWTIME (queue SIGMA$OPER, entry 549) holding
$ sho ent/ful 549
Entry Jobname Username Blocks Status
----- ------- -------- ------ ------
549 SHOWTIME JON Holding
On available batch queue SIGMA$OPER
Submitted 17-JAN-2008 20:08:15.03 /KEEP /LOG=DSA1200:[USERS.][JON.SCRATCH].LOG; /NOPRINT /PRIORITY=100 /RETAIN=ALWAYS
File: _$4$LDA3:[DEMO]SHOWTIME.COM;1
$ sho dev disk$itrcdemo

Device Device Error Volume Free Trans Mnt
Name Status Count Label Blocks Count Cnt
$4$LDA3: (SIGMA) Mounted 0 ITRCDEMO 1478 1 1
$ dism disk$itrcdemo
$ ld sho lda3
%LD-I-CONNECTED, Connected $4$LDA3: to DSA3403:[000000]DISK3.DSK;1
$ ld disc lda3
$ ld conn DSA3403:[000000]DISK3.DSK;1 lda4:/shar
$ mou/sys/noassist lda4 itrcdemo
%MOUNT-I-MOUNTED, ITRCDEMO mounted on _$4$LDA4: (SIGMA)
$ sho dev disk$itrcdemo

Device Device Error Volume Free Trans Mnt
Name Status Count Label Blocks Count Cnt
$4$LDA4: (SIGMA) Mounted 0 ITRCDEMO 1478 1 1
$ sho ent/ful 549
Entry Jobname Username Blocks Status
----- ------- -------- ------ ------
549 SHOWTIME JON Holding
On available batch queue SIGMA$OPER
Submitted 17-JAN-2008 20:08:15.03 /KEEP /LOG=DSA1200:[USERS.][JON.SCRATCH].LOG; /NOPRINT /PRIORITY=100 /RETAIN=ALWAYS
File: _$4$LDA3:[DEMO]SHOWTIME.COM;1
$ set ent/rel 549
$ sho ent 549/ful
Entry Jobname Username Blocks Status
----- ------- -------- ------ ------
549 SHOWTIME JON Retained on error
%RMS-F-DEV, error in device name or inappropriate device type for operation
-RMS-F-DEV, error in device name or inappropriate device type for operation
-SYSTEM-W-NOSUCHDEV, no such device available
On available batch queue SIGMA$OPER
Submitted 17-JAN-2008 20:08:15.03 /KEEP /LOG=DSA1200:[USERS.][JON.SCRATCH].LOG; /NOPRINT /PRIORITY=100 /RETAIN=ALWAYS
File: _$4$LDA3:[DEMO]SHOWTIME.COM;1
Completed 17-JAN-2008 20:11:08.35 on queue SIGMA$OPER
$

If this was the problem, use forum search for "DISPLAY_JOBS", a command procedure that can be used to create a command procedure to create a command procedure that will be able to resubmit the jobs. You will have to edit the command procedure to replace the old device name with the new.

[Jon Pinkley]

Since you reported "But all command files had finished unsuccessfully with an error "File not found", where exactly did you see this error message? It isn't exactly the same as the message that is displayed in the show entry status of a retained job if the device has changed.

I was able to reproduce the "File not found" error.

$ submit/que=sigma$oper disk$itrcdemo:[demo]showtime/noprint/hold/log=sys$scratch/retain=always
Job SHOWTIME (queue SIGMA$OPER, entry 1681) holding
$ copy _$4$LDA4:[DEMO]SHOWTIME.COM;1 _$4$LDA4:[DEMO]SHOWTIME.COM;/log
%COPY-S-COPIED, _$4$LDA4:[DEMO]SHOWTIME.COM;1 copied to _$4$LDA4:[DEMO]SHOWTIME.COM;2 (1 block)
$ pu _$4$LDA4:[DEMO]SHOWTIME.COM/log
%PURGE-I-FILPURG, _$4$LDA4:[DEMO]SHOWTIME.COM;1 deleted (1 block)
$ ren _$4$LDA4:[DEMO]SHOWTIME.COM;2 ;/log
%RENAME-I-RENAMED, _$4$LDA4:[DEMO]SHOWTIME.COM;2 renamed to _$4$LDA4:[DEMO]SHOWTIME.COM;1
$ sho ent/ful 1681
Entry Jobname Username Blocks Status
----- ------- -------- ------ ------
1681 SHOWTIME JON Holding
On available batch queue SIGMA$OPER
Submitted 17-JAN-2008 20:35:13.88 /KEEP /LOG=DSA1200:[USERS.][JON.SCRATCH].LOG; /NOPRINT /PRIORITY=100 /RETAIN=ALWAYS
File: _$4$LDA4:[DEMO]SHOWTIME.COM;1
$ ! here everything looks ok...
$ sho sym dw
DW == "DIRECTORY/SIZE=ALL/DATE/WIDTH=(FILE:38,SIZE:7)"
$ dw _$4$LDA4:[DEMO]SHOWTIME.COM;1

Directory _$4$LDA4:[DEMO]

SHOWTIME.COM;1 1/1 17-JAN-2008 20:35:50.20

Total of 1 file, 1/1 block.
$ ! but watch this...
$ set ent/rel 1681
$ sho ent/ful 1681
Entry Jobname Username Blocks Status
----- ------- -------- ------ ------
1681 SHOWTIME JON Retained on error
%RMS-E-FNF, file not found
-RMS-E-FNF, file not found
-SYSTEM-W-NOSUCHFILE, no such file
On available batch queue SIGMA$OPER
Submitted 17-JAN-2008 20:35:13.88 /KEEP /LOG=DSA1200:[USERS.][JON.SCRATCH].LOG; /NOPRINT /PRIORITY=100 /RETAIN=ALWAYS
File: _$4$LDA4:[DEMO]SHOWTIME.COM;1
Completed 17-JAN-2008 20:37:40.92 on queue SIGMA$OPER
$


Did you possibly run a command procedure to "resequence" the versions? At least one version I have seen backup instead of rename to avoid the files revision data from being changed, However it does change the File ID, and that must remain the same for the queue manager to be able to find the file.

[David Beorn · Accepted solution]

Check me if I'm wrong, but queue manager (as well as ALL RMS/Files-11 file operations, I believe) are actually performed using the file-id - so you can change not only the version number but the name as well (even "rename" it to another drectory), and operations continue on as always. Of course you can't copy it to another disk or copy/delete and rename another file "in under it" - because it creates a new file ID. I supposed if you tweaked it somehow so the file-ID WAS the same (can DMU(?) do that?), then it would probably work...

Dave

[Jess Goodman]

If the batch job command procedure could not be opened you would see this error in the batch log file:

Error opening primary input file SYS$INPUT

That line is followed by the text part of the RMS error message that explains why the batch file could not be opened. So if the batch file was deleted the complete log file would be:

Error opening primary input file SYS$INPUT
File not found
SYSTEM job terminated at 12-MAR-2008 17:08:11.64

But if the problem was with the device name changing or not being mounted then you get:

Error opening primary input file SYS$INPUT
Device not ready, not mounted, or unavailable
SYSTEM job terminated at 12-MAR-2008 17:13:00.48

If the log file device is not mounted the job won't run either and there is no log file to check obviously, but in this case you shouldn't see a "file not found" error either in accounting or on SHOW ENTRY if the job was retained.

Sorry, but this issue sounds more like some user error than a system problem.