Page 1 of 1

OSH is running into infinite loop

Posted: Wed Oct 20, 2010 8:57 am
by sohasaid
Oracle:
Client version: 10.2.0.1.0
Server version: 10.2.0.1.0

DataStage: 8.0.1.0
SuSE Linux 10.1 64-bit

Oracle Enterprise stage
Reading/inserting from/into Oracle

OSH process hangs (100% CPU time), DS client irresponsive

Trials:
1- running dsjob -run PROD_ETL test2
Same results

2-
OSH process:
strace -p <OSH pid>
...
times(NULL) = -2090015179
times(NULL) = -2090015179
times(NULL) = -2090015179
times(NULL) = -2090015179
times(NULL) = -2090015179
times(NULL) = -2090015179
times(NULL) = -2090015179
times(NULL) = -2090015179
times(NULL) = -2090015179
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
times(NULL) = -2090015178
...
Which indicates that OSH is running into infinite loop.

I don't think it's related to the Oracle server as even when we tried to connect to an Oracle SID with invalid hostname and we got the same result (infinite loop in OSH), it should have give invalid hostname error instead.


Thank you

Posted: Wed Oct 20, 2010 9:03 am
by ArndW
What does the job do (i.e. which stages) and does it have before/after logic in the job or any of the stages?

Posted: Thu Oct 21, 2010 2:41 am
by sohasaid
1- They are several jobs, extracting data from Oracle 10G database, make no or minor transformation and then loading into DB2 after a CDC stage. (it didn't hang in a particular stage, because it didn't even start it)

2- The job Director shows no abort messge, it's just hangs in:
main_program: orchgeneral: loaded
orchsort: loaded
orchstats: loaded

and then gives no further progress.

3- It doesn't have a before or after logic, except that some of them are issuing a Refresh MQT's command after running.

4- We have been tried creating a simple rowgenerator job, loading into a dataset and the job didn't hang, the problem is happening everytime it tries to extract or load from/into Oracle.

However, as i mentioned in the first post, we think it has nothing to do with the Oracle server.

Thank you in advance.

Posted: Thu Oct 21, 2010 3:20 am
by ArndW
I would tend to disagree with you that the Oracle connection or stage is not the culprit. Do you have any specific SQL you use in the jobs, or nonstandard stage settings? I don't think that SuSE is a supported platform, otherwise you could approach your support provider.

Have you tried using ODBC for the Oracle connection?

Posted: Tue Jan 25, 2011 5:13 pm
by dojunk1
Hi,

Would this link help in your case?
http://www-01.ibm.com/support/docview.w ... wg21430101