I've seen those database deadlocks now too, since going to 3.1  and a while
ago when I called it in, IBM said things were normal.




Virginia Hysock <Virginia_L_Hysock AT CSC DOT COM> on 08/14/98 10:32:18 AM

Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>

To:   ADSM-L AT VM.MARIST DOT EDU
cc:    (bcc: Julie Phinney/Green Bay/Humana)
Subject:  ANR0390W Database Deadlock





OS390/MVS 2.4 ADSM 3.1.1.3 Server

Hi all,

     First off - thanks to all those who responded to my notes early last
week
regarding the problems with insufficient memory in running my MVS server
under
3.1.1.3.  This has turned into a Sev 1 problem with IBM, and there does
seem to be
a memory leak around the database prefetch functionality.  I will be
applying a
zap on Saturday and am hoping this will alleviate the memory problems.  I
will let
you know how things turn out.  Details are in PMR60835 (can the general
public
access this record?  I'm not sure ...).

     Anyway, I've also been seeing sporadic ANR0390W messages regarding a
database
deadlock condition, followed up by ANR9999D of various types (this is a
generic
message with varying details depending on the service being discontinued to
free
the deadlock), then followed by additional messages indicating what service
was
aborted (ANR2183W, ANR0986I, ANR1093W, ANR1042I to name a few).  One of the
fix
attempts on the above-noted memory problem was to reduce the size of my db
bufpool.  The cache hit ratio has, of course, dropped drastically to the
92-93%
range.  I'm wondering if this could cause an increase in these database
deadlocks.
They only seem to occur when I have multiple tape functions going at once
(migration, reclamation, backups, etc.).  I checked the info in IBMLINK,
but the
only 3 listings against this message were old.

     Anybody out there have any experience with this under the 3.1 version?
Any
ideas??  TIA,

                              Ginny