Dan writes:

> Hi Gang,
>
> We have started to get some unusla errors during our online backup on a 7.0
> pp2 box.  This is a test and development system, so getting quiesce errors
> is an ongoing issue for us.  We have recently started getting some new
> errors however, and had a system abort during our backup the other evening.
>
> After getting tons of quiesce errors in the store stdlist, we are now also
> seeing the following:
>
> NOT STORED: TOO MANY FILES ATTACHED TO SHADOW
> and then
> NOT STORED: UNEXPECTED SHADOW ERROR -19 ENCOUNTERED
>
> We opened a call for the system abort we had on Thursday (8/12) evening
> (SYSTEM ABORT #773 FROM SUBSYSTEM #122).  Then last night, we had our
> online backup abort again complaining of the following:
>
>    4748   /DDP/DATABASE/MDSFILE0 NOT STORED: TOO MANY FILES ATTACHED TO
> SHADOW
>    4749   /DDP/DATABASE/INVDBE0 NOT STORED: FILE IS PART OF AN ALLBASE DB
> ENVIRO
> NMENT
>    4750   USE "PARTIALDB" OPTION FOR PARTIAL DATABASE STORE/RESTORE
>    4751   /DDP/DATABASE/MDSFILE0 NOT STORED: FILE IS PART OF AN ALLBASE DB
> ENVIR
> ONMENT
>    4752   USE "PARTIALDB" OPTION FOR PARTIAL DATABASE STORE/RESTORE
>    4753   UNKNOWN SHADOW SUBSYSTEM ERROR -80 ENCOUNTERED (S/R 911)
>    4754   SHADOW NOTIFY SYNC POINT ROUTINE ENCOUNTERED ERROR (S/R 3022)
>    4755   /NSD/USERVOL/files9/A.18539.1092447966.hdr NOT STORED: UNEXPECTED
> SHAD
> OW ERROR -19 ENCOUNTERED
>    4757
>    4758
>    4759   :else
>    4760   STORE aborted because of error. (CIERR 1090)
>           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>    4761   REMAINDER OF JOB FLUSHED.
>           ^^^^^^^^^^^^^^^^^^^^^^^^^
>    4762   CPU sec. = 908.  elapsed min. = 137.  FRI, AUG 13, 2004,  9:17 PM.
>
>
> Has anyone seen anything like this before ???
>
> TIA,
> Dan Barnes

If you have a call to your support provider for the SA773 you probably want one
for this too. Stores aborting is never a good thing even on a development system.

There is a known problem using PARTIALDB and DATE >= that could cause the
shadow error -19.  That occurs when part of a database qualifies to be stored and
part does not. However,  in your case the file it claims to be in error doesn't look like
it is part of a database "/NSD/USERVOL/files9/A.18539.1092447966.hdr".

The cause of the problem is that the shadow subsystem uses a files UFID or unique
file identifier. It is possible, although rare for portions of  UFID's to match those of
other files. In the cases of PARTIALDB it was found that two adjacent sets had a
similar UFID. This was caused when the database was enabled for "autodefer". Doing
this will detach each set from the transaction manager and this causes each set to be
assigned a new UFID. Because this is happening in quick succession, set by set, two
adjacent sets wound up with similar (not duplicate) UFIDs.  When STORE found that
one qualified and the other did not it confused the two and that generated the -19 error.

Storing and then restoring the reported file  will change its UFID and correct the problem.
I honestly don't know enough about the problem to know if this could occur with any files
or if it was specifically due to the PARTIALDB selection criteria. But it's worth trying to
store and restore the reported file to see if that resolves it.

The shadow error -80 has also been seen rarely in the past (afaik). The only documented
cause I could find had to do with a STORE ONLINE=END and having a temp CM
KSAM file saved as permanent at the sync point. I'm sure there may be other reasons
for this particular error but this was the only description I could locate.

hope this helps,

Bill
hp/vCSY

===========================================
  Reply to: bill . vcsy -at- comcast . net
===========================================

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