Reading data, inserting data in a table which is in sybase database. Tried odbc, sybaseoc stages. When we run the job from sequencer we get Abnormal termination without any reason. If we re-run the same job without any changes it completes without any issues. Below is the error message we get when job aborts.
From ODBC Stage we get this:
tgtinsDFMSeGLHdrIns..xfmFMS: |tgtinsDFMSeGLHdrIns..t_eGL_interface_hdr: DSD.BCIConnect connecting to SYB_DR_FMStrns, call to SQLConnect failed.
SQLSTATE=08S01, DBMS.CODE=7500
[DataStage][SQL Client][ODBC][DataDirect][ODBC Sybase Wire Protocol driver]Unexpected Network Error. ErrNum = 73|
From Sybase Stage we get this:
srcDFMSeGLCntlChk..t_eGL_Control: Sybase Version 'Sybase Client-Library/12.5.1 ASE Edition/P-EBF11360/DRV.12.5.1/RS6000/AIX 4.3.3/BUILD1251-004/OPT/Mon Aug 11 09:32:15 2003'
Abnormal termination of stage srcDFMSeGLCntlChk..xfm detected
Just wandered if anyone came across something like this.
DataStage
Moderators: chulett, rschirm, roy
If running the job manually from Director works, but using a Sequencer to execute the job fails, then the only variable would seem to be the job parameters. If the Sequencer is running out of a scheduled execution, then maybe there's a specific userid associated with running scheduled jobs (verify in DS Administrator for your project).
Both the Sequencer and Director call the same API to attach and run a job. The API doesn't care who does it, but what does matter are the job parameters. The Sequencer has the list of parameters and their values hard-coded into the job when it's designed. The Director shows you the defaults internal to the job.
Can you verify the parameters being used, as well as confirm if running the Sequencer is via a scheduled run or are you running it yourself?
Both the Sequencer and Director call the same API to attach and run a job. The API doesn't care who does it, but what does matter are the job parameters. The Sequencer has the list of parameters and their values hard-coded into the job when it's designed. The Director shows you the defaults internal to the job.
Can you verify the parameters being used, as well as confirm if running the Sequencer is via a scheduled run or are you running it yourself?
Kenneth Bland
Rank: Sempai
Belt: First degree black
Fight name: Captain Hook
Signature knockout: right upper cut followed by left hook
Signature submission: Crucifix combined with leg triangle
Rank: Sempai
Belt: First degree black
Fight name: Captain Hook
Signature knockout: right upper cut followed by left hook
Signature submission: Crucifix combined with leg triangle
Let's fill in more details. Is the Sequencer doing more work than just running this specific job? Is it simultaneously running other jobs?
When you say that the job works outside the Sequencer, but fails when called from the Sequencer, is it 100% of the time? Your initial post gave me that impression.
Now that you say "sometimes", that means your issue is most likely instantaneous server load. If you have a lot of jobs running and one of those jobs fails intermittently, but when re-ran runs okay, than the missing piece of information was that you were running other things at the same time.
You can be exceeding the memory limitations of the client software or ODBC driver, you could be slamming the server with so much i/o that DS jobs freak out trying to manage their threads because they have internal timeouts.
Can you clarify your Sequencer job and also use something like top, glance, or if on Solaris 2.8+ prstat to monitor your server load?
When you say that the job works outside the Sequencer, but fails when called from the Sequencer, is it 100% of the time? Your initial post gave me that impression.
Now that you say "sometimes", that means your issue is most likely instantaneous server load. If you have a lot of jobs running and one of those jobs fails intermittently, but when re-ran runs okay, than the missing piece of information was that you were running other things at the same time.
You can be exceeding the memory limitations of the client software or ODBC driver, you could be slamming the server with so much i/o that DS jobs freak out trying to manage their threads because they have internal timeouts.
Can you clarify your Sequencer job and also use something like top, glance, or if on Solaris 2.8+ prstat to monitor your server load?
Kenneth Bland
Rank: Sempai
Belt: First degree black
Fight name: Captain Hook
Signature knockout: right upper cut followed by left hook
Signature submission: Crucifix combined with leg triangle
Rank: Sempai
Belt: First degree black
Fight name: Captain Hook
Signature knockout: right upper cut followed by left hook
Signature submission: Crucifix combined with leg triangle
In the same sequencer there are 3 jobs.
1st job: Looks up a table which has 200 records
2nd job: Inserts 200 records
3rd job: Updates 1 record
When we run the sequencer it aborts majority of the time. everytime at different place (it could be lookup, update, or insert). If we re-run without any changes it completes successfully. The amount of data we are loading is not much.
We tried to run this job at different times. This job is in our stage environment. We ran this when there is nothing running on the box (server was idle).
1st job: Looks up a table which has 200 records
2nd job: Inserts 200 records
3rd job: Updates 1 record
When we run the sequencer it aborts majority of the time. everytime at different place (it could be lookup, update, or insert). If we re-run without any changes it completes successfully. The amount of data we are loading is not much.
We tried to run this job at different times. This job is in our stage environment. We ran this when there is nothing running on the box (server was idle).
Is this a new installation of DS on this server?
Is Sybase installed on the same machine?
Is this a new installation of Sybase?
I'm asking because we're probably beyond the realm of it's something you're doing. With an idle server, DS should behave well. If it's a new install, maybe there's a configuration that's been overlooked, such as kernel parameters, ulimit, etc.
If it's an existing and stable installation of both DS and Sybase, then what I would do as a last check is go into the DS jobs and turn off row-buffering, don't check anything. Even the project default may enable buffering, so just don't check anything. Recompile and see if that changes anything.
Other than that, you'll need to get Tech Support involved because they would know of specific issues related to your release, platform, and stages.
Is Sybase installed on the same machine?
Is this a new installation of Sybase?
I'm asking because we're probably beyond the realm of it's something you're doing. With an idle server, DS should behave well. If it's a new install, maybe there's a configuration that's been overlooked, such as kernel parameters, ulimit, etc.
If it's an existing and stable installation of both DS and Sybase, then what I would do as a last check is go into the DS jobs and turn off row-buffering, don't check anything. Even the project default may enable buffering, so just don't check anything. Recompile and see if that changes anything.
Other than that, you'll need to get Tech Support involved because they would know of specific issues related to your release, platform, and stages.
Kenneth Bland
Rank: Sempai
Belt: First degree black
Fight name: Captain Hook
Signature knockout: right upper cut followed by left hook
Signature submission: Crucifix combined with leg triangle
Rank: Sempai
Belt: First degree black
Fight name: Captain Hook
Signature knockout: right upper cut followed by left hook
Signature submission: Crucifix combined with leg triangle
We have been using DS for one year, not new install.
Yes. Sybase Client is installed on the same server as DS server. we are accessing sybase server which is on another unix server.
Sybase is not new installation. We have been using for many years.
I will see by turning off row buffering helps the situation.
Thank you Kenneth for taking time to help us. I will contact tech support...
Yes. Sybase Client is installed on the same server as DS server. we are accessing sybase server which is on another unix server.
Sybase is not new installation. We have been using for many years.
I will see by turning off row buffering helps the situation.
Thank you Kenneth for taking time to help us. I will contact tech support...