VMS and EMC BCVs

[Tim Nelson]

Got another one for the VMS gurus.
Anyone ever implement EMC BCV volumes for backups on VMS ?

Scenario:
BCV volume split from orional volume
BCV volume mounted
during mount system complains ( as it should ) %MOUNT-I-REBUILD, volume was improperly dismounted; rebuild in progress
This is no different that if I was doing this in the UNIX world. I must FSCK the filesystems before I can mount. I do not worry about this here as I am sure the applications are either shutdown or quiesced.
If the same quiescence occurs with the application on the VMS side can I feel equally as safe ?
Looking for others experiences here if any ..

Thank you !

[Uwe Zessin · Accepted solution]

Yes, a BCV is a crash-consistent copy of the original disk. If you can shutdown the application you might even be able to dismount the parent disk on OpenVMS before you do the split.

[Tim Nelson]

You are the man Uwe..

Thanks again and again for sharing the VMS knowledge !

If I get anything special from EMC like %FLUSH options to flush the buffers I will post it. The Managing Shadowing book has some good discussion on this type of topic but does not have too many specifics.

[Keith Parris]

The "rebuild" that VMS is doing here involves much, much less than in the UNIX world. All VMS is doing is finding areas that were proactively marked by VMS in the disk allocation bitmap on disk as being in-use because they were in the extent cache and might become allocated, and marking those areas as free so they can be used and thus providing more free disk space on the disk. It also fixes up disk quotas at the same time.

The file system metadata itself is not at risk here, as it is with FSCK.

[Antoniov.]

Hi Tim,
because you known unix world you can find in follows links the equivalent vms commands of unix:
http://www.think-forward.com/tips/Ivmsunix.htm
http://www.ctstateu.edu/help/unix/vms2unix.html
http://wwwvms.mppmu.mpg.de/vmsdoc/UNIX_VMS_CMD_XREF.HTML
I used theese links for my collegues.

Antonio Vigliotti

[Tim Nelson]

Here is something interesting that I found.
If you happen to be running a DCL script through a scheduler or a spawned job to sync a bcv group and the bcv group is large. i.e. 14 pairings.
You may run into exeeded quota problems when executing the symmir establish command.
The command will die with a %SYSTEM-F-EXQUOTA, process quota exceeded
If so make sure your bytlm quota is set to at least 150.

There is more to this story if someone wishes to know or if the above does not make any sense.

[Uwe Zessin]

I have checked some systems and they have PQL_MBYTLM = 100000. Is it possible that you are missing some digits?

[Antoniov.]

Uwe,
I guess Tim looked at BIOlm value in SYSUAF that's about 150.
Tim,
if your quote is bytlm you must change another value in user authoritation.
Both values make sense:
BIOLM is the maximum number of buffered I/O operations, such as terminal I/O, that can be outstanding at one time.
BYTLM is the maximum number of bytes of nonpaged system dynamic memory that a user's job can consume at one time. Nonpaged dynamic memory is used for operations such as I/O buffering, mailboxes, and file-access windows.
I guess PQL_MBYTLM value, in SYSGEN, is more important in this case as Uwe suggested.

Antonio Vigliotti

[Tim Nelson]

I modified the run switches to use /authorize so my "scheduler" job would read the UAF quotas. Once I did this I noticed all jobs spawned by the "scheduler" now listed 150 under the direct i/o limit and buffered i/o limit when doing a show proc/quota instead of the default 100.
I am sure I mixed up the quota names.

Thanks !!