Data Dictionary Problems

We are having a very strange problem. Sometimes the Data Dictionary on the Enterprise server gets hung up. What is happening is that we have multiple jobs running at the same time and the jobs go into a mutex wait. When looking at the threads the item that is holding the mutex is a lock on the DD files.
If we end all the jobs and then delete DDDICT and DDTEXT, then the jobs will continue running.
We have gone weeks without this happening and then we have it happen a lot. We got this error twice this weekend. This is happening in both our testing and production machines.

Thank you
Ellen
ES V5R2M0 Iseries PTF Level TC03364
Citrix
ERP8.0 SP21_Q1
Many 'Fat' Clients

mutex wait is caused by deploying a package while OneWorld UBE's are running on the AS/400. Job queues must be held and wait for UBE's to finish processing before deploying a server package.

I did checked that no one is deploying packages at the time the mutex wait is happening.


Thank you
Ellen
SP21_Q1
ES - V5R2M0
Citix
Many 'Fat Clients'

to know for sure, display the joblog of the Package Install Kernel and use F1 on the last message in the log.

To get everything out of mutex wait, hold the job queues you use for oneworld ube's, kill the oneworld ubes in mtxw status, end oneworld services, restart the services, release the job queues.



Jean Driscoll
AS/400(V5R2) Co-existent Xe SP22, Update 7/A73Cum12
WWW.JDETips.com

Re: RE: Data Dictionary Problems

We tried ending the jobs, bouncing services (endnet, clripc, strnet) and reran the jobs the mutex waits reappears. Only if we end jobs, end services, delete the Data Dictionary from the path code on the Enterprise Server, CLRIPC, bring up the services we were then able to run the jobs.

That is how I believe the problem is with the DD on the Enterprise Server. Also in the job log of the job in the mutex wait there are locks on the DD.

Ellen
V5R2M0 TC03364
SP21_Q1
Many Fat clients
Citrix

Re: RE: Data Dictionary Problems

We just experienced this same issue. We had dropped services, cleaned up sql pkgs, temp files, ddtext, dddict, glbltbl spec files and came up with jobs in semaphore/mutex wait.

We even IPLed the box and still got the same error.

The jdedebug had a lock on ddtext.

Aug 1 12:42:38 ** UNLOCK: Total locks for ddtext after operation: 0
Aug 1 12:42:38 ** Locking ddtext in WRITE mode.
Aug 1 12:42:38 ** LOCK: Total WRITE locks for ddtext after operation: 1

THanks to your info, we had a fix....

I just opened a call. Did you have any resolution to this????