File and JDBC CCs not polling

[former_member433211]

Hi,

We are currently facing a issue with the Sender File and JDBC channels, after a high consumptions of application threads in several server nodes these adapters stop polling, we need to restart the whole J2EE cluster in order to reestablish the polling.

We are running a PI 7.0 SP15.

Has someone faced the same issue?

Thanks in advance.

Edited by: Ruben Hopkins on Feb 26, 2009 5:26 PM

[Former Member · Accepted solution]

Ruben,

Were these CC working earlier?

I would suggest you to check the sender side CC monitor from RWB to know the exact error message ( ofcourse after confirming that they are active :)..... )

Venkat.

[former_member433211]

The problem was solved by SAP, a fix in the code of the AE Scheduler was needed. You can see the details about this issue here:

Note 1317686 - Adapter Framework Scheduler may fail to remove a job

Thanks to all for the help.

[former_member433211]

During the application threads consumption peak the defaultrace if showing a Mapping that fails several times, what type of Thread the Mappings use, Application or System?

Thanks.

[Former Member]

Mappings has to use application threads and not system threads.

-VIjendra

[former_member433211]

Hi,

Based on your experience which can be the immediate downsides of increasing Heap Size, Application Threads and System Threads in a Server Node? if these values are incremented the Dispatcher values should not also be increased accordingly?

Thanks.

[prateek]

Could you try my comments here first

Regards,

Prateek

[former_member206760]

Hi Hopkins,

we had the same problem on production server and file channel was not polling .. we had 4 server nodes

and we applied notes and the problem got solved

801926 - additional paramters in file adapter

1083488 - File/JDBC sender adapter stoppedd polling

do not forget to use the following paramter in the advance mode of the file adapter

clusterSyncMode value lock

Giving points is another way to say thanks

Edited by: Tarang Shah on Feb 27, 2009 11:29 AM

[former_member433211]

Hi,

The issue occurs in server nodes with 4GB heap and 2GB heap alike. The application threads are already in 450. We are running 14 server nodes in total plus 3 more for a De-central adapter engine. The funny thing is that in the 14 server nodes (Central and Application Instances) we are running only 36 Sender File CCs and in the De-central Adapter Engine instance with only 3 server nodes we are running 120 Sender File CCs without problems.

My theory is that somehow the Enqueue Server under "conflict" situations between the server nodes locks the File and JDBC services globally and after the nodes went down in the consumption of Application Thread these locks are not being release.

Any ideas?

Thanks in advance.

Edited by: Ruben Hopkins on Feb 27, 2009 9:38 AM

[Former Member]

Hi Ruben,

• Increasing the number of threads set in the system,

• Increasing the Java heap size

• minimizing the UDFs used in the mappings

can really solve your problem or atleast can minimize the number of restarts.

-Vijendra

[former_member433211]

Hi,

What is the name of the server that should be restarted for the Adapter Engine?

Thanks.

[Former Member]

Hi,

As of now temporary solution could be restarting your Adapter Engine.

As an alternative to avoid these types of problem, Please request your basis person to increase number of threads.

This is just a suggestion. Please dig this forum for more clues.

Since this is production issue, please donot take chance. There is not harm in restarting Adapter Engine in these types of situations.

Thanks

Subbu

[former_member433211]

Hi,

But if we restart the adapter framework service, all channels will go down right?

Thanks.

[Former Member]

Hi,

Stop/Start the File, JDBC adapters and adapter framework in visual admin. Stop/Start the cache in visual admin, then it will work.

Thanks,

Jogula Ramesh