T.R | Title | User | Personal Name | Date | Lines |
---|
540.1 | | UTRTSC::utoras-198-48-118.uto.dec.com::JurVanDerBurg | Change mode to Panic! | Mon Apr 28 1997 10:28 | 6 |
| What's the customer trying to do? /FOREIGN means that the application
driving the tape is responsible for volume swithing, and that has not
changed.
Jur.
|
540.2 | New Mount Code... | XDELTA::HOFFMAN | Steve, OpenVMS Engineering | Mon Apr 28 1997 10:54 | 10 |
|
V7.1, and the "Redhawk" kit on V6.2, is a new MOUNT design, and this
work has caused some illegal qualifier combinations to be rejected,
combinations that were previously permitted but were meaningless or
were illegal. The work was intended to resolve a number of subtle
(and not-so-subtle) problems seen with the older MOUNT design, and
particularly around its use with shadowing.
Recode the procedure to use the correct syntax.
|
540.4 | | KERNEL::ANDERSONS | | Mon Apr 28 1997 11:28 | 11 |
| Thanks for the prompt reply,
The customer feels the need to mount multiple volumes /FOREIGN and select the
volume labels to be used. As this was not a meaningless combination of Mount
qualifiers, why is it now 'illegal'.
Also, do you have any ideas on how the customer should do the above?
Thanks
Scott.
|
540.5 | Details, Please | XDELTA::HOFFMAN | Steve, OpenVMS Engineering | Mon Apr 28 1997 12:36 | 6 |
|
There is a known bug around ISO9660-format CD-ROM processing and the
new MOUNT code that is being ECO'd -- if this is a different problem,
please provide the specific details around just what the customer is
up to here, and I can pass these along to the MOUNT maintainers.
|
540.6 | | AUSS::GARSON | DECcharity Program Office | Tue Apr 29 1997 00:05 | 9 |
| re .4
What difference between MOUNT/FOREIGN and MOUNT/FOREIGN/NOAUTO did the
customer perceive in previous versions of VMS? There is some suggestion
(but I certainly can't verify it) that /AUTO is implemented by the ACP
and hence would have been ignored if used with /FOREIGN in previous
versions anyway.
What application is handling the magtape?
|
540.7 | customer reply... | KERNEL::ANDERSONS | | Tue Apr 29 1997 11:06 | 24 |
| 1. The command files that used the MOUNT/FOREIGN/NOAUTOMATIC had not been
changed prior to the upgrade of VMS V7.1. This was working OK under VMS
6.1 and under 7.0.
2. We need the above mount command so as to make the tape available for an
RMU backup of an RDB relational database. As the databases are very
large, a full backup may require two DAT tapes, the routine requests the
tapes be loaded, and their volume labels identified. The backup specifies
these labels explicitly, so as to avoid the backup requesting an
additional volume, where the volume label would be overridden by the
backup. We make use of the volume labels as a control as to what is
physically on these tapes, and therefore cannot let VMS write its own
label.
i.e. vol1 = 123400, vol2 = 123500,
and not vol2 ending up as 123402. !!
As far as we are concerned, it was WORKING before and must remain so now
and in the future. Therefore we need this resolved as soon as possible.
Does this help?
Scott.
|
540.8 | "ASAP"? | XDELTA::HOFFMAN | Steve, OpenVMS Engineering | Tue Apr 29 1997 11:11 | 5 |
|
: As far as we are concerned, it was WORKING before and must remain so now
: and in the future. Therefore we need this resolved as soon as possible.
The customer said the magic works -- where's the IPMT?
|
540.9 | | AUSS::GARSON | DECcharity Program Office | Tue Apr 29 1997 23:50 | 22 |
| re .7
>This was working OK under VMS 6.1 and under 7.0.
Sorry to labour the point but was it actually working or was it just
not producing an error message? i.e. did the customer ever verify what
the labels getting written were? What is the difference in V7.0 or
earlier between MOUNT/FOR and MOUNT/FOR/NOAUTO as far as observable
effects are concerned?
The fact that they are using the tapes with RMU/BACKUP makes some kind
of sense i.e. that RDB might really have the tape end logic within
it. (Unfortunately that might complicate any fix since we may not know
what functionality Oracle was/is relying on, if any, and what it was/is
doing.)
Does RMU itself contain any qualifiers to get the desired tape label
behaviour (as BACKUP does)? e.g. does RMU/BACKUP/LABEL=(...) have the
desired effect? (Have the customer read the description.) While the
customer may want to beat up on Digital for making an incompatible
change, it may make more sense for them to discuss with Oracle whether
Oracle RDB can already do what they want.
|
540.10 | /Label=('tape_label_list') | KERNEL::ANDERSONS | | Thu May 01 1997 08:58 | 32 |
| I think the following extract from the customers backup command file illustrates
the problem......
$ RMU/BACKUP/ONLINE - ! Full Backup
/LOG/NOREWIND/BLOCK_SIZE=49152 -
/Label=('tape_label_list') - ! May contain 1 or more volume labels
/Tape_expiration="''keep'" -
divorced$db - ! Logical for .RDB database
tape:'backup_set'.rbf
$ RMU/BACKUP/ONLINE/INCREMENTAL=COMPLETE - ! Incremental Backup
/LOG/NOREWIND/BLOCK_SIZE=49152 -
/Label=('tape_label_list') -
/Tape_expiration="''keep'" -
divorced$db -
tape:'backup_set'.rbf
$ RMU/BACKUP/AFTER_JOURNAL/QUIET_POINT/FORMAT=NEW_TAPE - ! Journal Backup
/LOG/NOREWIND/BLOCK_SIZE=49152 -
/Label=('tape_label_list') -
/Tape_expiration="''keep'" -
divorced$db -
tape:'backup_set'.AIJ
(.7) I have just raised an IPMT so I hope that we can give the customer a fix or
at least a good explanation as to the reasoning behind the change.
Thanks.
Scott.
|
540.11 | Works Fine Without... | XDELTA::HOFFMAN | Steve, OpenVMS Engineering | Thu May 01 1997 09:24 | 4 |
|
I've done a number of multi-volume RMU tape backups over the years, and
have never needed nor used that MOUNT command qualifier for them.
|
540.12 | Well how did you do it? | KERNEL::ANDERSONS | | Thu May 01 1997 10:06 | 16 |
| Steve could you please expand on .11.
Looking at the RMU backup commands used in .10, how do you suggest the customer
mounts multiple volumes while giving the volume labels.
If we go back to .0 you will remember that me original request was
'My customer is not happy with the change and would like us to suggest an
alternative command to have the same result.'
If you have send suggestion, then I would appreciate any contructive help that
you could give.
Thanks.
Scott,
|
540.13 | MOUNT Documentation Clear As Mud... | XDELTA::HOFFMAN | Steve, OpenVMS Engineering | Thu May 01 1997 10:37 | 15 |
|
I would expect RMU to handle to handle the volume switches and the
volume labeling itself, either checking the labels or not, depending
on the number of labels specified via the RMU /LABEL qualifier. (If
the customer wants to check all labels, then the RMU/BACKUP command
in the procedure needs to specify a sufficient number of labels.)
MOUNT/NOAUTOMATIC is for ACP-mounted tapes. If you look at MOUNT
/AUTOMATIC, you will see references to the use of the magtape ACP,
and to the ability to use multiple tape drives and multiple volumes
and to switching. If you look at /FOREIGN, you will see that no ACP
is used, nor is an ACP provided.
In other words, remove /NOAUTOMATIC, and specify enough /LABELs.
|
540.14 | | AUSS::GARSON | DECcharity Program Office | Sat May 03 1997 22:20 | 21 |
| re .10
>I think the following extract from the customers backup command file
>illustrates the problem......
It doesn't illustrate it sufficiently for me.
The first paragraph in my .9 does not appear to have been answered i.e.
while it is true that a combination of qualifiers that was previously
accepted is now not accepted, is this of any consequence? What
difference did it previously make? Since /AUTO apparently only applies
to tapes controlled by the ACP did RMU really take any note of whether
/AUTO or /NOAUTO was in force if MOUNT even made that information
available for a tape mounted /FOREIGN, or, from a users perspective, is
there any observable difference?
Can the customer explain what functionality over and above
RMU/BACKUP/LABEL is required?
Can the customer explain what goes wrong if they just MOUNT/FOR the
tape and do the RMU/BACKUP?
|