Kent,

This is a known error, (I can't remember the apar), such that when a volume
becomes emptied, not all the pointers get deleted, and an audit volume
fix=yes does not correctly delete them either.

It is a volume-related problem, so if you mark the tapes involved "unav",
or if they are scratch, remove them from the robot inventory, the problem
will be alleviated although it has not gone.

** Consider VERY carefully upgrading at the moment. **

The upgrade to any v5 on a significant size db will take hours, since it
involves building a new db table.  There are expiry problems with system
objects at releases above 4.1 and below 5.1.5.1. (they don't get deleted,
and just build up-and-up-and-up), and expiry performance problems deleting
all the built-up inventory of system objects at 5.1.5.1.

We consider it ESSENTIAL that anybody upgrading to v5.1.5.1 should do a
complete trial run using a copy of their live database on a test server.
Include in the testing: expiry running against datamove processes (and a
few bu sessions) to check for server stalls.

 You will need to run "cleanup backupgroups", and check your actlog for
"audit required" type errors.

We still have a 50GB db at 4.1.4 but do not dare upgrade it due to the
problems we have experienced with upgrading our others.

The words "of", "worms" and "can" spring to mind, not necessarily in that
order.

Regards

Dave
--
+44 (0) 20 7608 7140

Qui custodiet ipsos custodes



                      Kent Monthei
                      <Kent.J.Monthei@G        To:       ADSM-L AT VM.MARIST 
DOT EDU
                      SK.COM>                  cc:
                      Sent by: "ADSM:          Subject:  URGENT - ANR0102E 
asalloc.c(####): Error 1 inserting row in table
                      Dist Stor                 "AS.Segments" during Migration 
/ what is table 'AS.Segments' ??
                      Manager"
                      <[email protected]
                      .EDU>


                      01/11/2002 15:27
                      Please respond to
                      "ADSM: Dist Stor
                      Manager"






Solaris 2.6 / TSM 4.1.2  (I know..........we're upgrading to Solaris 8 ,
TSM 5.1 very soon!)

        DB - 50GB, PctUtil 78%
        Log - 4GB, LogMode=Normal, PctUtil=1%
        DISKPOOL - 138GB, PctUtil 87%, PctMigr 39% (currently)

We're getting migration process failures.  We start 4 migration processes
at a time, but lately only the first 2 run to completion successfully. The
others fail almost imediately after starting, with this error-pair:

        ANR0102E asalloc.c(####): Error 1 inserting row in table
"AS.Segments".
        ANR1032W Migration process ### terminated for storage pool
DISKPOOL - internal server error detected.

For failed processes, MigrationProcesses=4,MigrationDelay=0 settings cause
the server to immediately start new migration processes, which also fail
with the same error.  This fail/restart cycle continues until the original
two Migration processes finally finish, eventually leaving nothing left to
migrate, and the cycle finally stops.

Until today, we could avoid this by reducing the DISKPOOL
MigrationProcesses setting to 2 or 1.  However, this morning all migration
processes reported the error.  None ran to successful completion and we had
to intervene.

At this point, we cannot migrate date from this pool & need some expert
advice a.s.a.p.


Where's the problem, and how can I fix it to break this cycle?

-rsvp, thanks

Kent Monthei
GlaxoSmithKline