Eben Yong wrote:
> To backup our HP3000, we use Hiback from HICOMP.  Recently we have
> experienced sporadic failures when attempting to restore data from the dlt
> drive across the network to another HP3000.  For example, we backup data
> on production HP3000; then attempt to restore the data to a development
> HP3000.  Below is a snippet from the $stdlist, which displays the error.
> I looked at available patches that potentially target this problem and the
> only thing I found for MPE/iX 6.5 is patch MPEMXF6A.  Would anyone care to
> comment on this?  Any suggestions as to the advisability of installing the
> patch?  Thanks as always.

The symptoms you would see, if you needed the MPEMXF6A patch, would
most likely be a system abort 1730, when a VSM table becomes full.
On an (otherwise idle) system you would need to restore over 200
GB of data (in pretty large files) to hit this.

A way of checking if some OS table resources are hit high is
to use Threshold Manager (THMGR).

However I think MPEMXF6 is unrelated to the error shown below.
VSM subsys error -300 basically means the program hit an invalid
virtual address. I have no idea what causes that. Since you say
it is sporadic aka intermittent, it can be rather difficult to
troubleshoot.

It appears that HIBACK is making use of the so-called
TurboStore API for network backups (as do/did Legato,
Veritas, HP Omniback aka DataProtector).
HP has recently found a buffer handling problem in this
API, and fixed this in patch MPEMXH6.
On MPE 6.5 (I assume this is what you have since you refer
to the -A version of MPEMXF6) this patch is still Beta. It
has also been superseded by two more beta patches, MPEMXK1
and MPEMXJ6.

If you are willing to test if the API fix helps with your problem
then ask your HP RC for the beta patch; I would recommend to
use/request the latest one: MPEMXJ6-A.

>  STORE ABORTED BY THE ENTERPRISE CLIENT DUE TO AN ERROR (S/R 3305)
>  /AIHWORK/DATA/HEALTHJ5.001 NOT RESTORED: RESTORE FORCED TO TERMINATE
> WHILE RESTORING FILE
>  FILES RESTORED :                             198 FILES FAILED TO
> RESTORE :                      1  **** INTERNAL TRAP  Data PAGE
> Fault                            [VSM] Undecoded status.info = -300 ABORT:
> HIBACKX.PUB.HICOMP NM SYS   a.0231de70 ts3k_restore_next_file+$180
>  NM PROG  f7.0002d154 P5_restore+$898
>
> Patch Name: /mpe-ix_patches/c.65.00/MPEMXF6A
>    Description: RESTORE/COPY of large amounts of data can cause VSM table e
>    Patch Size : 1376 sectors

HTH,
Goetz

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *