Problem Migrating from RAC to single node db

[Laura Albrecht]

Hi guys.  This is an unusual one - not sure anyone has ever gone in this direction - usually it would be going from single node to RAC, but we strive to be different.  :-)

We are in the process of migrating our DEV db from a RAC cluster to a single node db.  Variety of reasons for this - I won't go into that.  We brought the system down normally and then the DBA's did the migration using Datapump schemas export / import.  They had no problems with the migration.

However, when we went to bring up the system, the JWP's wouldn't start.  We found this strange message in the log.

U00003367 No primary work process is registered in the database; next attempt in '60' seconds.

Has anyone ever seen this before?  We weren't quite sure what to make of this because obviously there had been a PWP in the original db.

I opened up a ticket with support to see if there were any settings specifically pointed toward being a RAC that could be causing a problem, but they said no.

Was just wondering if anyone out there has attempted this and if you ran into any issues.  This is an Oracle db.

Thanks in advance.
Laura


------------------------------
------------------------------
Laura Albrecht
Enterprise Scheduling Lead
Takeda Pharmaceuticals LLC
------------------------------
------------------------------

[Christoph Rekers]

Hi @Laura Albrecht

If you shut down your system what does this statement return?

SELECT MQSRV_Name, MQSRV_System, MQSRV_TcpIpAddr, MQSRV_TcpIpPort, MQSRV_restendpoint,
(CASE mqsrv_type
WHEN 1 THEN 'CP'
WHEN 2 THEN 'WP'
WHEN 4 THEN 'PWP'
WHEN 16 THEN 'JWP'
WHEN 32 THEN 'JCP'
-- else str(mqsrv_type)
END) "Process Type" FROM MQSRV WHERE MQSRV_Name IN (SELECT OH_Name FROM OH WHERE OH_OType='SERV')
order by "Process Type"


Cheers
Christoph 



​

------------------------------
----------------------------------------------------------------
Automic AE Consultant and Trainer since 2000
----------------------------------------------------------------
------------------------------

[Laura Albrecht]

@Christoph Rekers when I run that query - I get no results.  

If I just query the OH table where OH_TYPE = SERV I do get results.

​

------------------------------
------------------------------
Laura Albrecht
Enterprise Scheduling Lead
Takeda Pharmaceuticals LLC
------------------------------
------------------------------

[Pascal Osthus-Bugat]

Hello Laura,
The most common reason for that problem is that the Oracle link definition either in the tnsnames.ora or in the ucsrv.ini has not been correctly updated.
For example if the database hostname changed, has it been correctly updated either in the tnsnames.ora or in the [JDBC] section the AE configuration file?

Kind Regards

Pascal

[Laura Albrecht]

The system comes up.  So we can even login into client 0, etc.  So the tnsnames.ora and ucsrv.ini files are good.  It's just the JWP's that were not coming up.  Unfortunately I was not on hand when the maintenance / work was being done.  I think I am going to try again tonight and troubleshoot more then.  I'll update here tomorrow on what I find.  Thanks.

------------------------------
------------------------------
Laura Albrecht
Enterprise Scheduling Lead
Takeda Pharmaceuticals LLC
------------------------------
------------------------------

[Laura Albrecht]

So, we tried connecting to the system again tonight.  This time I brought the processes up individually (normally we use a script).  

So, first started a WP.  This came up fine and I assume became the PWP.  Got the ready for run message.  No problems.

Next, I started a JWP.  This appeared to start up fine, but once I looked closer - it was very strange.

I would have expected, at this point to have a WPsrv_log_001_00.txt and a WPsrv_log_002_00.txt file.  However, I didn't.  It seemed like suddenly the 2nd WP process overtook the initial WP process.  So that's where the message for "No PWP registered on the db" message is coming from.  When the 2nd process - a JWP - starts and essentially seems to kill the first WP - there is no more PWP on the system.

I have no idea how this is happening.  I tried starting the processes via the service manager dialog as well.  Same thing.  It is quite peculiar.  I'm still mulling what could be causing this strange behavior, but wanted to update this thread in case anyone has any ideas.  

I'm not sure what else to check / look into.  All support has suggested is me migrating the db using the AE utilities (unload / load) which I'm not really keen on, nor do I think is necessary.  We've used Oracle utilities all the time to copy db's and I've never seen this type of issue / behavior.

I'm going to ask the DBA's to check / verify the access of the db userid, but that's all I can think of at this point.

------------------------------
------------------------------
Laura Albrecht
Enterprise Scheduling Lead
Takeda Pharmaceuticals LLC
------------------------------
------------------------------

[Laura Albrecht]

That was it @Pascal Osthus-Bugat.  We had updated the tnsnames.ora file.  Since the db name, etc. was all the same we didn't think we needed to do anything else, but since the server name changed we also needed to update the entry in the ucsrv.ini file.​  Thanks!!

------------------------------
------------------------------
Laura Albrecht
Enterprise Scheduling Lead
Takeda Pharmaceuticals LLC
------------------------------
------------------------------

[Anicet TAGNE]

Hi @Laura Albrecht ​,
JWP's are java process. Make sure that you use the correct jar file in the folder ae/bin/lib.
Have you also switch Oracle version and the Oracle Client on AE host machine ?