HP3000-L Archives

November 2000, Week 3

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Sletten Kenneth W KPWA <[log in to unmask]>
Reply To:
Sletten Kenneth W KPWA <[log in to unmask]>
Date:
Mon, 20 Nov 2000 16:16:48 -0800
Content-Type:
text/plain
Parts/Attachments:
text/plain (74 lines)
Calling all user logging experts:

I might just as well start out by confessing that I am stumped....
here's the scenario:

After our in-house client-server app ran fine for a couple hours
this morning, all of a sudden our operations staff started getting
user logging errors (same with termulator apps going against
production database):

DBOPEN, MODE 1, ON <xxxxxx>
MAXIMUM USER COUNT PER LOG PROCESS REACHED
HEX DUMP OF STATUS ARRAY FOLLOWS:
ff92 000d 0000 0000 0000 0191 4163 cd08 0001 0000

I'm going to leave out a lot of <interlude> stuff, and jump right
to the summary:

(1)   Three weeks ago we merged two programs that had been
running on separate databases into ONE database.  That one
production database is the ONLY database using our
production log process.  Until today we had "No worries, mate"
(but have probably been slowly and incrementally gaining a few
more concurrent users over that three-week period).

(2)   Above error would "come and go", as people logged on
and off and got into and exited various programs.  At times of
extended inability to log more people on due to above error,
DBUTIL SHOW USERS indicated 127 - 128 active DBOPENs.

(3)   While I didn't get through doing up-and-down process
tree check / count for every one of those 127 - 128 DBOPEN
PINs, I checked maybe 30+ before the count started falling off
at the end of day shift;  never found more than one child
process for each.

(4)  IIRC off the top, with current CM user logging the limit per
user log process is 1100+ (can't believe I'm not sure of exact
number;  after all the IPROF / HPW / 3000-L threads I have
been involved with on this subject).

(5)   I looked at everything I can think of with GLANCE and
with SHOT (FAMily, TREE, etc.).   I don't see anything that
jumps out at me.


Two questions for the experts for right now:

(1)   Any know problem which can yield the "MAXIMUM USER
COUNT PER LOG PROCESS REACHED" error;  but which is
really caused by something else;  i.e.:  We have another
problem but are NOT really hitting the USER count per LOG
PROCESS limit ??.

(2)   Any straightforward way to get a real-time read of the total
list of USERs / PINs that a particular LOG PROCESS thinks it
being accessed by;  i.e.:  something that effectively does:
SHOWLOGSTATUS logid; USERS=ALL        ??
Thinking we have been way, way short of the 1100+ limit, we
have never bothered to look in to this...  now I'm looking hard...


We managed to suffer through the day by getting rid of a few
static clients, and now on swing shift the number of active
DBOPENs as fallen off to only 26....  but just like US election
for Pres, I fully expect our problems are NOT over....

SIDEBAR:  I haven't called HP yet...  Why ??....  Because our
usual ~!@#$%^&*-+= end-of-the-FY-quarterly-funding fiasco
only just today finally got straightened out to the point where I
got the money to start the P.O. process......   < major sigh > ....

Ken Sletten

ATOM RSS1 RSS2