cancel
Showing results for 
Search instead for 
Did you mean: 
Read only

SAP PI Cache Problem

Former Member
0 Likes
998

Dear all,

I am developing different scenarios with BPMs accessed by SOAP (Web Services) within SAP-PI.

I am having a lot of problems with some of the XSLT split mappings programs. What's happening is that we have developed various mappings and made changes to them and upon activating them it's possible that they are being incorrectly managed by the J2EE server cache e.g. wrong versions being kept in the cache. The consequence of this is that we obtain different results for the same data input.

Executing the services sequentially, with the same input parameters, we obtain results with different data in a random form that coincides with the different versions of the mappings that we have published to the server.

Could there be an error with the way the java cache works?

I am checking different issues regarding the SAP PI Caché manual ("How to Handle Caches in SAP XI 3.0.pdf"), but I cannot find the problem. Doing testing, I am getting in different moments, following errors:

1. This error occurs sometimes:

com.sap.aii.af.service.cpa.impl.exception.CPADirectoryCacheException: Failed to check secure connection configuration. Reason: com.sap.aii.af.lib.sld.SLDException: Failed to read SLD instance name. Reason: HTTP response code: 503 (Service Unavailable)

at com.sap.aii.af.service.cpa.impl.cache.directory.DirectoryAccess.getDirectoryURL(DirectoryAccess.java:134)

at com.sap.aii.af.service.cpa.impl.cache.directory.DirectoryAccess.getDeltaRefreshURL(DirectoryAccess.java:191)

at com.sap.aii.af.service.cpa.impl.cache.directory.DirectoryAccess.getDeltaCacheUpdateXML(DirectoryAccess.java:440)

at com.sap.aii.af.service.cpa.impl.cache.CacheManager.performCacheUpdate(CacheManager.java:491)

at com.sap.aii.af.service.cpa.impl.cache.CacheManager$CacheUpdateRunnable.run(CacheManager.java:322)

at com.sap.engine.frame.core.thread.Task.run(Task.java:73)

at com.sap.engine.core.thread.impl5.SingleThread.execute(SingleThread.java:144)

at com.sap.engine.core.thread.impl5.SingleThread.run(SingleThread.java:242)

Caused by: com.sap.aii.af.lib.sld.SLDException: Failed to read SLD instance name. Reason: HTTP response code: 503 (Service Unavailable)

I did all checks that came in the manual, but to no avail.

Thanks in advance.

Regards

Accepted Solutions (0)

Answers (3)

Answers (3)

Former Member
0 Likes

Hi!

According to my Knowledge

1. BPM need to use in final case because it tooks lot of performance issues in RUN time because IN BPM while running your scenario the process is always switch from ABAP stack BPM Process Engine to JAVA Stack for executing the mapping whch you configured in the java stack and again it needs to swtich towards the Process Engine and finally again it needs to swtich towards java stack to ID part through Interface determintation step....

In going all these it takes lot of Load on server and suppose if it is handling very huge data defnitely you ill face performance issues.

2. Also I think there is some authentification issue while running it means defalutly even though u configured any adapter on your user id by default all those adapters run by using ADMINs authentifications check whether it was locked or some other problem.

3. Also check once CACHE thorugh SXI_CACHE and also u can check via RWB-->Cache Monitoring...

Regards::

Amar Srinivas Eli

Edited by: Amar Srinivas Eli on Mar 9, 2009 4:17 PM

Shabarish_Nair
Active Contributor
0 Likes

what does SXI_CACHE show?

Usually the resource not found error has been seen to disappear on restarting the java stack. Can you try that too?

Former Member
0 Likes

This "ramdom" errors could be due to amount of maximum java memory heap of the different J2ee servers?

Former Member
0 Likes

Did you already tried transaction SXI_CACHE to do a full or delta cache refresh?

Former Member
0 Likes

I did it, and is not the problem

Former Member
0 Likes

hm... then I really think - as mentioned above - you may need to restart the java stack.

Shabarish_Nair
Active Contributor
0 Likes

illogical but what about pulling the plug.

A server restart and then see if it makes any difference. Else we will have to have more troubleshooting.