cancel
Showing results for 
Search instead for 
Did you mean: 

PI 503 - Service Unavailable in SXMB_MONI

suwandi_cahyadi
Contributor
0 Kudos

Dear Experts,

Lately we have been facing a 503 - Service Unavailable error in SXMB_MONI.

The symptoms that I've observed is started from the Adapter Engine becoming slow (Java Part). Because when I go to the RWB, message monitoring, choosing the Adapter Engine from the component takes more time than usual.

After sometime that the adapter engine is processing slowly, the messages in integration engine become stuck in SMQ2, trying reprocessing the message ends up with 503 - Service Unavailable. Sometimes it is ICMP service unaivalable.

We then solve this problem by restarting the PI system.

What I want to as is. What is this 503 - Service Unavailable error? Which part should I check?

Is there any tools that can be used to monitor the Java stack of PI? Can we use SolMan? Like the utilization, etc...

Thank you,
Suwandi C.

Accepted Solutions (1)

Accepted Solutions (1)

markangelo_dihiansan
Active Contributor
0 Kudos

Hello Suwandi,

Please check this document under section 2.2.4 HTTP Response Code 503: Service Unavailable.

Hope this helps,

Mark

suwandi_cahyadi
Contributor
0 Kudos

Hi Mark,

Sorry for forgetting to mention that the interfaces are asynchronous.

Does the document apply for asynchronous interfaces too?

Thank you,

Suwandi C.

former_member184720
Active Contributor
0 Kudos

Hi Suwandi - i dont think it's bacause of a perticular interface.

Can you/basis team analyze the performance on the java stack?

refer to the section 6.5 J2EE Engine Bottleneck in the below document.

http://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/f0b96675-87cf-2c10-b489-dee0ac03f...

markangelo_dihiansan
Active Contributor
0 Kudos

Hello Suwandi,

Yes, this is applicable for async interfaces too since they would also have to pass the same pipeline/steps as with sync interface.

Regards,

Mark

suwandi_cahyadi
Contributor
0 Kudos

Hi Hareesh,

Thank you for the reply. I will look into the document.

Do you know the TCode or tools that can be used to analyze the performance on the Java Stack? Is it in the document at the Analyzing the Adapter Framework (AFW) part?

Because this becomes happening more often. Seems the Adapter Engine gets "exhausted" easily.

Thank you,

Suwandi C.

former_member184720
Active Contributor
0 Kudos

NWA is the best place.. and you can install willy introscope to analyze J2EE engine.

Also for channel issues, you can use XPI inspector .. a very good tool provided by SAP.

Answers (1)

Answers (1)

suwandi_cahyadi
Contributor
0 Kudos

Hi Experts,

Based on the given documents. For the following scenario that occur to me:

There's a master data interface from SAP to other system with the scenario IDoc to File. This master data interface will be triggered by a background job that runs at night. Sometimes the master data could be massive that it stuck at the adapter engine and impacts other transaction interfaces at the next day

I think of the following solutions:

- Increase the maximum File thread to 10. Currently there is only 5 maximum thread. (is it okay to double the maximum thread?)

- Increase the maximum concurrency of the Master Data receiver adapter to 5. Currently it is only 2. Because the master data could be massive, increasing the maximum concurrency could faster the processing at the receiver adapter that often becomes the bottleneck.

- Create Message Prioritization at the AAE (Component monitoring > Adapter Engine > Message Prioritization) to put the master data to low priority while other business critical interfaces to high priority. (For the interfaces that is not configured as low nor high, would they be treated as if they were in normal priority?)

Any concern or advise would be appreciated.

There's no transport system for increasing the maximum file thread and the message prioritization right?

Do we configure it directly at the target PI system?

Actually, another thing that I want to ask is that. The background job for the master data creates 900 IDoc which ends up creating 600k of files. So, we can say that those 600k of files created almost at the same time. Is this something normal for PI could still handle or it depends on the hardware and the performance tuning configuration?



Thank you,

Suwandi C.