db2 hangs with no messages
Aleksander Grzebyta
What could be a reason when db2 hang in the way that I cannot call "db2
connect". The console simply hangs and does not return any error. The CPU
utlization is low. When I call "db2 force application all" the CPU
utlization rises to about 70% for user processes. After "db2 force" the db2
hang also on "db2 stop" command.
What , in general, can be a cause of such a behaviour of db2 when it hangs
during calls not event returning any error message.
sincerely Olek
Knut Stolze
Have a look at the db2diag.log. Maybe the connect just takes so long because
the db crashed before and has to be recovered upon the connect.
--
Knut Stolze
DB2 Spatial Extender
IBM Germany / University of Jena
Larry Menard
occurs on the first connect after a database was shut down uncleanly.
I would also suggest that in future requests for help, please include at
least your platform , Operating System and DB2 level information.
--
Larry Menard
IBM Workstation Database (DB2) Performance Team
Defender of Geese and of All Things Natural
"Aleksander Grzebyta" <aleksander...@SPAMtalex.com.pl> wrote in
message news:aqvqb0$3nt$1...@news.onet.pl...
Aleksander Grzebyta
Użytkownik "Larry Menard" <lme...@ca.ibm.com> napisał w wiadomości
news:ar062s$2pmc$1...@tornews.torolab.ibm.com...
> Knut's suggestion sound likely... this is called "crash recovery", and
> occurs on the first connect after a database was shut down uncleanly.
>
> I would also suggest that in future requests for help, please include
at
> least your platform , Operating System and DB2 level information.
Hello !
IBM DB2 7.1 sp 6
Sun Solaris 7
6 CPU
8GB RAM
The described behaviour occurs in the time when DB runs for few days and
finally encounters performance problems causing us to brutally kill it since
all connect and admin commands hang with no answer.
sincerely Olek
Larry Menard
If the connections are hanging *before* the database crash, then it's not
crash recovery (I think Knut and I both assumed the hangs were occurring
*after* a db crash.)
Is there anything written to the "db2diag.log" file when the hang
happens, or when the performance problems first begin? I wonder if you are
running out of some kind of heap. Are you using any Stored Procedures, and
if so, are they FENCED?
--
Larry Menard
IBM Workstation Database (DB2) Performance Team
Defender of Geese and of All Things Natural
"Aleksander Grzebyta" <aleksander...@SPAMtalex.com.pl> wrote in
message news:ar0abo$li1$1...@news.onet.pl...
Aleksander Grzebyta
> Oleg,
>
> If the connections are hanging *before* the database crash, then it's
not
> crash recovery (I think Knut and I both assumed the hangs were occurring
> *after* a db crash.)
Yes. It is before the carsh, but
... the term of crash : DB simply does not respond and does not perform its
work so we are forced to kill it.
>
> Is there anything written to the "db2diag.log" file when the hang
> happens, or when the performance problems first begin? I wonder if you
are
These are the error codes that appear in db2diag.log:
0xFFFF877E
0xFFFF8905
0xFFFF8B16
Only the first one is unclear for me. The remaining two are "by design"
since the application catches the exceptions that corresponds to these
errors.
############################################################################
#######
2002-11-14-09.26.04.797887 Instance:marcinc Node:000
PID:5801(db2agent (MB)) Appid:*LOCAL.marcinc.021109003010
base_sys_utilities sqleriar Probe:201 Database:MB
Error tokens : 34 4
Data Title:SQLCA PID:5801 Node:000
sqlcaid : SQLCA sqlcabc: 136 sqlcode: -727 sqlerrml: 1
sqlerrmc: 4
sqlerrp : sqlrropn
sqlerrd : (1) 0xFFFF877E (2) 0x00000000 (3) 0x00000000
(4) 0x00000000 (5) 0x00000000 (6) 0x00000000
sqlwarn : (1) (2) (3) (4) (5) (6)
(7) (8) (9) (10) (11)
sqlstate:
############################################################################
#######
> running out of some kind of heap. Are you using any Stored Procedures,
and
> if so, are they FENCED?
No, there are no Stored Proc. This is all Java application where DB2 is
accessed from IBM WebSphere App Server that serves the CMP EJBs.
There is also standalone JDBC application connecting to DB2.
The sympthom I have found are massages in /var/adm/messages having the
following form:
############################################################################
#######
Nov 14 09:25:16 system-name DB2[5826]: DB2 (user.000) base_sys_utilities
sqleriar reports: probe id 201 with error 2055 and alert num 0
Nov 14 09:25:16 system-name DB2[5826]: extra symptom string provided:
Error tokens :
Nov 14 09:25:16 system-name DB2[5826]: data: 31ff4d41 5243494e
432e5a41 50595441 1.USER.ZAPYTA
Nov 14 09:25:16 system-name DB2[5826]: data: 4e4941
NIA
############################################################################
#######
There are also snapshots activeted and colledcted every 10 minutes. During
today crash the last snapshot that is collected is at 9:17 am, the two next
snapshots ( taken at 9:27, 9:37) are empty and then the DB2 was killed.
sincerely Olek
Nanchil
we are also facing the same problem. The db2 commands hang without
any messages. we are in
IBM DB2 7.1 sp 6
Sun Solaris 7
6 CPU
8GB RAM
Nanchil
"Aleksander Grzebyta" <aleksander...@SPAMtalex.com.pl> wrote in message news:<ar0abo$li1$1...@news.onet.pl>...
Fan Ruo Xin
Does the db2diag.log shows the same information each time when your db2 hang. If
possible, you can take a trace.
Regards,
FRX
Joerg Ammann
we had a similar problem (had to hard kill ALL db2-processes) after
upgrading to Db2 7.2 FP 5 on AIX.
the workaround to the problem was to eliminate all SQLJ from our java
programs.
does somebody know if this is fixed in newer fixpacks?
joerg