By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
454,237 Members | 1,228 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 454,237 IT Pros & Developers. It's quick & easy.

db2fmp

P: n/a
I have a problem with that the db2fmp process are hung , with truss can
I see this :

/4: nanosleep(0xFC4CFBF8, 0xFC4CFBF0) = 0
/4: semop(1441803, 0xFC4CFC78, 1) Err#11 EAGAIN
/4: nanosleep(0xFC4CFBF8, 0xFC4CFBF0) = 0
/4: semop(1441803, 0xFC4CFC78, 1) Err#11 EAGAIN
/4: nanosleep(0xFC4CFBF8, 0xFC4CFBF0) = 0
etc ...

Our application runs java stored procedures which are called
concurrently and this seems to cause problem for the db2fmp process
because after a while the application doesn't respond.

My first thought was that there wasn't enough db2fmp's so I changed
NUM_INITFENCED and
FENCED_POOL but the problem perisist.
Current values in dbm cfg are :
Keep fenced process (KEEPFENCED) = YES
Number of pooled fenced processes (FENCED_POOL) =
MAX_COORDAGENTS
Initial number of fenced processes (NUM_INITFENCED) = 2
Max number of existing agents (MAXAGENTS) = 200
Max number of coordinating agents (MAX_COORDAGENTS) = MAXAGENTS
..
db2fmp for java , is it one for each connection or ?
/Roger

Sep 5 '06 #1
Share this Question
Share on Google+
1 Reply


P: n/a
ka******@hotmail.com wrote:
I have a problem with that the db2fmp process are hung , with truss can
I see this :

/4: nanosleep(0xFC4CFBF8, 0xFC4CFBF0) = 0
/4: semop(1441803, 0xFC4CFC78, 1) Err#11 EAGAIN
/4: nanosleep(0xFC4CFBF8, 0xFC4CFBF0) = 0
/4: semop(1441803, 0xFC4CFC78, 1) Err#11 EAGAIN
/4: nanosleep(0xFC4CFBF8, 0xFC4CFBF0) = 0
etc ...

Our application runs java stored procedures which are called
concurrently and this seems to cause problem for the db2fmp process
because after a while the application doesn't respond.

My first thought was that there wasn't enough db2fmp's so I changed
NUM_INITFENCED and
FENCED_POOL but the problem perisist.
Current values in dbm cfg are :
Keep fenced process (KEEPFENCED) = YES
Number of pooled fenced processes (FENCED_POOL) =
MAX_COORDAGENTS
Initial number of fenced processes (NUM_INITFENCED) = 2
Max number of existing agents (MAXAGENTS) = 200
Max number of coordinating agents (MAX_COORDAGENTS) = MAXAGENTS
.
db2fmp for java , is it one for each connection or ?
/Roger
Are these db2fmp (C) or db2fmp (Java) processes?
(& use ps -ef |grep db2fmp to check parent PID)

We have an open problem with db2fmp (C) dying (orphans) now on AIX.
(db2 8.1 FP10 on AIX 5.3.0-TL05, 32bit DB2, 64bit AIX, 32bit Java 1.4)

dbx shows:

(dbx) th
thread state-k state-u ..function
$t1 wait blocked $PTRGL
$t2 run blocked _global_lock_common
$t3 zombie terminated sqloAppWaitOnThread
$t4 zombie terminated sqloAppWaitOnThread
$t5 zombie terminated sqloAppWaitOnThread

(dbx) where
pth_spinlock._global_lock_common(??, ??, ??) at 0xd0110420
rec_mutex._rec_mutex_lock(??) at 0xd0335b24
cuexit.unatexit(??) at 0xd034909c
....dlclose at 0xd03ee180

According to DB2 support, this is a known problem in AIX,
when multiple terminating threads call dlclose() and exit().
AIX-team is undecided whether to do a design change in locking,
or go the Doc-APAR route and ignore it. In DB2-land they are
preparing IY87538 to avoid the problem, regardless of AIX direction.
(targeted to ship with FP14 if testing goes well).

There are also several recent APARS for db2fmp (Java) problems.
Can you divulge your OS & DB2 levels, and if you see any relevant
error messages in the db2diag.log ??

Eric
Sep 5 '06 #2

This discussion thread is closed

Replies have been disabled for this discussion.