UBE stays on TIMW

We recently migrated to an iseries model 8203-E4A and upgraded to V5R4M5 from a 550 O/S V5R4M0. We are on ERP 8 SP 23. We are noticing that some UBEs go on TIMW status and stay at this state for a long time and then eventually complete successfully. I viewed the joblog and did not find any error messages. This did not happen prior to the hardware migration. Do you have a clue what is causing this to happen?

Hi Cherry,

I'm not sure if this is the same as what you're experiencing with a TIMW condition but in normal circumstances what happens is that when the job is submitted in QBATCH, the job passes control to the SQL server which then takes control and runs the job. The original jobs waits (CNDW) until control is passed back to it from the SQL server when it is complete. To change it we added following line under the [DB SYSTEM SETTING] section of the iSeries jde.ini file and then restarted the JDE services:

sqlServerMode=0

We are currently on V5R4 and it was like that for some time. We then had different problem and added the line as per Oracle. Doing so did not fix our other problem but the jobs reverted back to running in QBATCH instead of passing control to the SQL Server to run the jobs. In V5R4, the sqlServerMode is by default 1 ("ON") from an i5/OS operating system perspective so you need to force JDE to switch it "OFF". I have not removed the line from the ini file.

Hope it makes sense and helps.

You can also view the following link which talks about the jobs going in CNDW (not similar to your situation) which explains what I mentioned above:

https://support.oracle.com/CSP/main/article?cmd=show&type=NOT&id=656138.1

Cheers,
Misael

Thanks for the reply. The UBEs are in active state in the subsystem QBATCH but they go to TIMW status for a minimum of 3 minutes and then end normally. The resulting reports are all correct. The end user sees the 'P' status for a long time while I see TIMW on the iseries active job screen.

Hi Cherry,

That is very strange indeed and totally different to my explanation. It could be anything that is causing this. Just what comes to mind:

Check the logs on the iSeries?
You could use ServerAdministrationWorkbench to turn on logging for the kernels on the iSeries?
Try turning on debugging when submitting the UBE?
Check memory pool sizes and allocations?
Relevant PTFs loaded and applied?

TIMW, if you don't already know, means that the job is waiting for a time interval to end. It could be the job is waiting for something before continuing. Does this happens with all UBEs or just some. Should the UBE take 3 minutes or should it be quicker than that?

Sorry for all the questions but like I said it could be any number of things causing the TIMW condition.

It will be interesting to see what fixes your problem.

Kind regards,
Misael

I did a trace job and sent it to IBM. They have narrowed it down to MODULE JDESMTP.I do not have my SMTP server started in the iSeries and have not changed the server or the client JDE.ini.

I think IBM have just pinned it down for you. If you have the address book configured with email addresses for users the system is attempting to send out job completion emails. If the JDE.INI is loaded with the default settings pointed to some Oracle Denver server the UBE process is hanging trying to reach the SMTP server. It times out after a standard delay.

Taking the SMTP server name out of the JDE.INI completely might do the trick.

Hi Cherry,

I think Justin is right. You could remove the SMTP name in the JDEMAIL section out of the jde.ini and restart the services.

Turning on debug for the job should give you an indication on what the system is trying to do.

The other option is to set up the STMP server on the iSeries and see if that is in fact your problem. Was the SMTP server set up on the previous system?

I have attached a PDF document you can use to set up email on the iSeries. See page 8 Setting up iSeries 400 to be an e-mail server.

Hope it helps.

Kind regards,
Misael

The issue has been resolved. We have the default mailserver name (mail.JDEdwards.com) in our iSeries JDE.ini. OpenDNS name resolution however, generates a web page which requires resolution to succeed. As a result, the iSeries succeeded and resolved the name instead of failing, and then tried to send email status messages through the firewall, which blocked the traffic, and iSeries continually retried for 3 minutes, creating the mystery pause.
We responded by creating a dummy DNS zone internally for jdedwards.com with no "mail" record. The iSeries now promptly fails at the earliest step to resolve the name and quickly skips generating the unwanted status report email message as per normal.
Thanks for your help and I will be using that attachment to get SMTP configured on the iSeries