SWIFT 2003 ETD Library - hanging on this.prepareInputData() method -

[cshubhra]

Hi,

We use eGate version 4.5.2 and were using Swift2001 libraries till Nov. On Nov, we upgraded to Swift 2003 libraries.

After that, one of our collaborations, which has ELS is intermittently getting hung while trying to call this.prepareInputData().

We tried to upgrade our stcjcs.jar, changed from Standard IQ to JMS IQ, and even tried to upgrade to EGate Version 4.5.3 and we are consistently getting hung. Seebeyond does not have any solution for us...

Some statistics...

1. Our production hangs after processing about 200-350 messages.

2. Development, after applying several patches, hang after 700-1000 messages for both Standard and JMS iqs.

3. 4.5.3 hangs after 2200 - 3000 messages

Things we tried already...

1. Increased heap memory size for JVM
2. Used multiple versions of stcjcs.jar, and other associated jars

Did someone face the same/similar problem and have a solution?

We are on Solaris 5.8

----------------------
-------Shubhra--------

______________________
______________________

 

[sundeepra]

conversion from Standard IQ to JMS iq

when collaboration script is being loaded the following method is invoked

this.prepareInputData();

the output is transferred to JMS iq and the messaging structure when converted from e*Gate 4.2 to V 4.3

keep the max messaging sequence at 3000,for this use the following code in ur collab script ---

set message_sequence(data-payload)= 3000;
com.stc.common.collabService.JCCollabControllermpl.translate;

set_max_handler(0);

stccmd $port_name = "" ; stack_address = "" ;

type this in the client installation of e*Gate manager
and when the schema is called

sendMessageToeGate("message"

when the e*Gate is polled the messages are transmitted thru
the above mentioned syntax...

keep me posted!!!

 

[sundeepra]

what was the heap size limit that u gave in JVM

coz the range is an integer between 0 and 2147483647.

 

[cshubhra]

This is the response I got from Seebeyond

Shubhrajit,
Based on research by Development, the swift librairies for mt502 is generated by the old compiler and the etd uses push mode. We have experienced hanging issues for push mode etds in 4.5.3. They believe this could be causing the problem at your end.

They want you to try to regenerate the etd by using the SSC wizard from the mt502.ssc which comes with the 2003 swift library and use that particular ETD in your collaboration rule.
By doing this, the newly generated ETD will be using the pull mode.

Please let me know whether that suggestion resolves the hangong issue or not.

This has solved the problem.
BTW I had tried these before and it did not solve the problem. I am documenting these for others

The major things we tried and has not worked (Latest first)

1. Remove Swift 2001 templates from environment.
2. Increase stack memory size for threads ( both UNIX Kernel and JVM )
3. Use JMS IQs instead of STC standard.
4. Upgrade to EGate 4.5.3
5. Have only one collaboration in the problem eway. (Production has two)
6. Increase available heap memory to the process.
7. Upgrade executables and Jar files in 4.5.2. (Tried multiple combinations)
8. Increase IQ re-org frequency and clean IQs.

----------------------
-------Shubhra--------

______________________
______________________