reason for huge amount RFC connection caused by SAPJSP ?

[Former Member]

Hello,

We have double stack NW7.0 with UME in ABAP, portal, BI.

ST03 shows more than 10 mln request daily caused by SAPJSP user. It calls BAPI_USER_GET_DETAIL, SYSTEM_RESET_RFC_SERVER and PRGN_J2EE_TELNO_CONVERT ... functions, transferring a lot of data and consuming CPU system resources.

But we have no more than 200 users daily and no more than 60 concurrent users using portal and very few applications.

What my be the reason for such strange for me behavior ?

How to check what it responsible (I mean which components and applications) on Java side for those SAPJSP calls ?

Any hints,

best regards,

Michal

[Former Member]

Hi All,

thank you for your help, I have learned something.

Fortunately we have found wrong written customer program. Problem is solved.

best regards,

Michal

[Former Member]

Hi Michael

We are facing a perormance issue due to enormous "SYSTEM_RESET_RFC_SERVER" calls to the seever. Please help me find out which URL in extranet is calling this function to server???

Regards

Manikiran G

[Former Member]

Hi Michal,

We have this same architecture and yes, there are numerous calls to the ABAP UME backend because every time a user authenticates the ABAP stack is checked as you say. Here are a couple of things that may be helpful to you:

1. Increase the capacity and times on your UME Cache (SAP note 1118873). See Support Desk > User Management > Cache monitor. We increased the caching times as per the [bold] entries below. Obviously, size this for your environment:

ume.cache.acl.permissions.default_caching_time = 3600 [86400]

ume.cache.acl.permissions.initial_cache_size = 100 [1000]

ume.cache.group.default_caching_time = 3600 [86400]

ume.cache.group.initial_cache_size = 500

ume.cache.notification_time = 0

ume.cache.principal.default_caching_time = 3600

ume.cache.principal.initial_cache_size = 500

ume.cache.role.default_caching_time = 3600 [86400]

ume.cache.role.initial_cache_size = 500

ume.cache.user.default_caching_time = 3600 [86400]

ume.cache.user.initial_cache_size = 500 [2500]

ume.cache.user_account.default_caching_time = 3600 [86400]

ume.cache.user_account.initial_cache_size = 500 [2500]

2. Tune your PCD Cache using the PCD cache optimiser

3. Make sure your TLN Navigation Caching is switched on and working

4. Check your logfiles for UME pool exhaustion error messages. If needed, raise the UME pool size on the UME ABAP configuration. Make sure you're using a load-balanced connection to your ABAP backend.

Hope this helps,

Marc

[Former Member]

Hi Marc,

thank you, it's very helpfull.

But we have constant load during nights, weekends when there is no or few working people.

This load not coming from jlaunch itself but each second 250 calls from J2EE to SAPJSF ...

We are investigated Srivastava's advice, but haven't find anything, there is lack of our knowledge.

best regards,

Michal

[Former Member]

Hi Michal,

What patch level are you on? A couple more ideas to look at:

1. Are you using any RTC? Check SAP Note 864801 for removal of RTC components if you aren't using it...

2. Do you have any CCMS or GRMG monitoring that is hammering your heartbeat service and maybe is not configured correctly?

3. It's worth checking your Java system reports in /nwa to see what are your most expensive components. You might - as we did - find a few surprises.

4. Try using the Diagtool for a few seconds to log what's happening in the system. Again we found this quite illuminating.

Thanks,

Marc

[g_srivastava]

Hi Michal,

As I was going through same concept these days.The reason for such huge mails are because there is a problem in proper configuration of the work item.Work item is created using workflow.There may be many reason within workflow for this though i think 'Loop until' has been handeled with container.Please check it out the work item.

To stop the mails you need to go to SM50 and terminate the session.

Hope it be useful for your query.

Have a best day ahead.

[Former Member]

Hi Srivastava,

thank you, but could you please give more details about mails you have mentioned. What mails ? I do not observe and have any mails.

I have no process in SM50 I am able to kill. Those request coming from Portal for authentication reason (SAPJSF user), but I am unable to find their source.

best regards,

Michal

[g_srivastava]

Hi Michal,

I was talking about the workflow event.If the work item is not configured properly, in conditional loop this is the general problem we face.I am sorry to specify word mail.Though I meant that when we configure a task it may be mails or anyother event.Mail being the general terminology so I used it.Just let me know if you have defined this event through a workflow.

Have a best day ahead.

[Former Member]

Hi Srivastava,

because I am new to workflow could you please give me some hints, how and where can I look for cause.

best regards,

Michal

[g_srivastava]

Hi Michal,

I think you should read this document which will definately help you understand what i was saying.

Link: [workflow|https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/com.sap.km.cm.highlightedcontent?documenturi=%2flibrary%2fnetweaver%2fnetweaver-developers-guide-2004s%2fnetweaverdevelopersguide2004s_otherdocuments%2fworkflow_tutorial.pdf]

Have a best day ahead.