xi freezes under load

[Former Member]

Hi all,

We have a troublesome situation on our hands in our XI test system.

We have a synchronic flow that works in the following manner:

Site calls XI using a webservice.

XI start a bpm and gets information using IDMS and a RFC call to another sap system

XI replies back to the site.

Somehow if we fire 10 - 30 parallel messages to XI it totally freezes up.

In SXMB_MONI we see the messages coming in but the status stays at "Log Version" and then the complete system freezes. Only way to get it working again is by a full restart (stopsap, startsap).

In the logs we see that the gateway stops working. And we get some error messages about a desolation row being full.

Here are the exact messages:

21.11.2007 10:58:04.528 ERROR Connection pool XIAlerting:DesolationRow: is exhausted. The current pool size limit (max connections) is 5 connections.

21.11.2007 10:58:04.776 ERROR JCoClient connection missingHost:sap_XIU_00|AdapterType:Mail|Sender:null:null|Receiver:null:null|Interface:null:null|MsgID:

Does anyone know what is happening here? Or does anyone have any idea why XI would totally freeze up under some load?

Regards,

Wietse

[Former Member]

FYI...

https://www.sdn.sap.com/irj/sdn/go/portal/prtroot/docs/library/uuid/526bd490-0201-0010-038e-d3ff7eb1...

Regards,

Sumit Gupta

[Former Member]

ok I found the setting.

It seemd it was alerting.MaxConnections in the Visual Admin. Under Sap XI AF Core service of the J2EE engine.

We have set it to a higher value now (23, just to make it an odd value so we could recognize the change if it showed up in the logs).

We do not get the desolationrow errors anymore.. But the system still dies when receiving 30 webservice messages at once.

JCO connection settings are high enough now as well.

[Former Member]

Hi Wietse,

please close this questione as answered as you made it.

and reward points if sth is useful.

Regards,

Sumit Gupta

[Former Member]

Hi Wietse,

IFyou got the answer for this thread then make it answered.

It is necessary as the rule of SDN.

Regards,

Sumit Gupta

[Former Member]

This problem is not fixed by any of the answers here. We did however find a solution.

I will write a post here within 1 or 2 days with what we did to conquer this problem for future reference for others that might run into this problem.

[Former Member]

Hi Wietse,

can you please let us know the answer.

as we are getting the similar problem.

Best Regards

Sachin Bhatt

[Former Member]

DearWietse Braan

For more information about JCo pool, refer to the SAP Library under SAP NetWeaver

>Application Platform (SAP Web Application Server) > Connectivity >Components

of SAP Communication Technology > Communication Between ABAP and Non-ABAP

Technologies > SAP Java Connector > SAP JCo Client Programming > Establishing

Connection to an SAP Server ® Connection Pools, or refer to the help of the SAP

NetWeawer Developer Studio for the description of the class JCO.Pool under SAP

NetWeaver Developer Studio Documentation ® SAP Web AS for Java Applications >

API WAS > Java Connector > com.sap.mw.jco > JCO.Pool.

Regards,

Sumit Gupta

• rewards points if useful**

[aashish_sinha]

Hi,

I am afraid that somehow your JCO setting are changed. that is by default there is 12 connections or i am afraid that somehow or intensionally you have changed it to 5 to see the affect.

please revert it to 12 and it will run fine. and stop restarting the server its not a good solution. check out for the pool settings.

regards

Aashish Sinha

PS : reward points if helpful

[Former Member]

Where can these settings be found in the Visual Admin?

Al JCO connection settings we can find are set to 10-20 or higher.

[aashish_sinha]

Check out JCO Pool Setting and increase it and try again !!!!

[Former Member]

where do i do that?

[Former Member]

Hi Wietse,

to avoid this problem in future.

use any standard extractor with user exit to capture some custom fields If I run the FIGL delta twice in same day (before midnight)

Standard FI datasources usually extract data from the previous day

with you problem just increase the pool size.

if not solving the issue then start the JAVA engine.

Regards,

Sumit Gupta

[Former Member]

Hi Wietse

Visual admin is a GUI which checks all clusters and cluster elements that are running on J2EE engine and CCMS agents use RFC to a CEN so check your

JCO connections in visual admin.

if the number of connnections is low then increase this limit and also check if any application is engaging your more connection due to malfunction.

hope it will be helpfull

Regards

sandeep Sharma

PS: if helpful reward points

[Former Member]

Have you checked the JCo Connections in the Visual Admin

Increase these connections to 10 to start with, see if that helps