BSSV Timeout Error

Hi All,
We are having Business service for generating Work orders(WO) and this is running fine in Production. Recently we had requirement from external system to generate mass WOs and when a huge batch of thousands are executed in CRP some of the request fails with timed out error as below:

*ERROR* CallObject@1917935: COSE#1000 Request timeout: timeout after 90000ms host AS400:6014(7653) SocID:2042 PID:5757 BSFN:CreateServiceLineWO user:BSS_PY Env:JPY812

This is happening due to the kernel zoombie and i am trying to find balance between the input and what the BSSV server & Enterprise server can take on.

Did anybody ever face such issue?. If so i would like to discuss what possible changes you made to overcome this issue.

We are on 8.12 TR 8.97 AS/400 & OAS with two clusters of BSSV for CRP with load balancing.

Hi,
open server manager url and go to your BS stanza.

Under configuration go to Runtime Settings and increase your Timeout Threshold (you will find the same value of your timeout exception).

Let me know if this solve your problem.

Regards.

Bruno Condemi

I have the some problen with as some users are disconnect randomly, sometimes queries remain processing, please can you advice any change on sttings

Yes we have the setting as 90000ms is this the maximum threshold or we can go beyond this? Can somebody tell me what timeout you have set to overcome large jobs.

Chan

Chan,

You mention the kernel in question is a zombie. That implies it is crashing. Is there anything in the log for that kernel? Since there is more activity with the new load, perhaps you have a memory leak that wasn't evident before.

Craig

Craig,
The Callobject logs doesnt have any trace of crash/leaks but i get timeout error on the BSSV logs. The function is never executed and the request is thrown out from BSSV.

Chan

Any chance a package is being deployed at that time? We see the timeout get reached if a large update package (that includes BSFNs of course) is deployed. Just a guess.

Yes, you need to increase the value.

My setting is 300000ms.

No we have a scheduled deployment of package so there is no deployment during testing.

Craig I think zombie kernel is an effect of timeout.

Thanks Bruno did you do any timeout change in jdeinterop or jdbj?

A kernel will become a zombie if it stops processing without unregistering. For example, a memory violation that crashes the process. The client (BSSV server in this case) that does not receive a net response within it's timeout threshold will report the timeout error. But that is the client's exception, not the kernel.

Is it possible that a BSFN is actually taking longer than the timeout value?

It's true but I was aspected some kind of "JdeNetAssociatedKernelException" in this case (coming from another invoke) but it's no the case.

However it's easy to verify if timeout on bssv layer cause zombies.

The setting I said resolved my issue (big ship confirm from bssv) so I didn't changed others settings.

If your issue persist we can compare our settings (I'm not at customer site now but if you need I'll check).

Bruno,

That would be cool to find out. Though I would be surprised to find a kernel stop working because one of its client sessions reached a timeout threshold. Thanks, as always, for your input.

Craig

Craig,
Thru logs it clearly says that the error is due to the timeout. I couldnt get any trace of crash or other error in callobject.
I refered to Oracle KG last week it does has reference to the timeout error resulting of same issue so its no surprise.
I only wanted to ensure that i am not missing any other INIs settings required to overcome this.

Anyway thanks all for the help.

Chan