Purge the Dead cases before Migration !

To avoid any migration (of cases) releated error , Please help to purge the dead cases and only migrate the live cases.


"swutil PROC RELEASE ... -m LIVE"

Happy Migration  :D No no Happy iProcess-ing !!

Error : (ORA-30036: unable to extend segment by 8 in undo tablespace 'UNDOTBS2')

UNDO Tablespace issue can take toll on Oracle database.

Simple reason for this problem is that UNDO Tablespace is running out of allocated disk space whilst attempting the procedure migration .That causes MIGRCASE instruction failure and the Oracle database would be required to rollback all of the changes it had made up until the UNDO failure.The MIGRCASE instruction would then be un-locked and another BG process would then be allowed to process that same MIGRCASE instruction which would start the migration process from the very beginning again.


1. How to find out the total cases which need to be migrated ?
Please run below query on iProcess database for a specific procedure .

select b.is_dead, b.major_vers, b.minor_vers, count(*)
  from proc_index a JOIN case_information b
    on a.proc_id = b.proc_id
 where a.proc_name = '<>'
 group by b.is_dead, b.major_vers, b.minor_vers
 order by b.is_dead, b.major_vers, b.minor_vers

To confirm on the migration error , you would also like to search "MIGRC" in the log files. You will find details in sw_warn.log file for case migration. Something similar to below :

CPLInstruction - Error> <-2>

Gather the Client Debug Logs

To validate the SSO to add users to Queues in iProcess, You can gather the client debug logs using below commands :

To see why that one procedure is listed twice, we would need to see Debug logs for an import/export.
Let's set environment debug.  

Steps:Log in as the the iProcess user.
Type this command at the Linux prompt:
DEBUG=ALL=255
export DEBUG

Then run the import and export commands, and validate  pstaffifc01.log file ffor errors, along with the plistP.txt file from this command

./plist -Pv > plistP.txt

Please make sure to stop the Debug, type this command, or log out and back in
unset DEBUG

Invalid Temporary Destination Error!!

When loads get high, sometimes, there are errors such as Invalid Temporary destination encountered
The tibjms.jar file is located in SWDIR/jmslib/ems. The checksum for the old version of tibjms.jar is as follows:

$ cksum SWDIR/jmslib/ems/tibjms.jar
1042607223 279905 SWDIR/jmslib/ems/tibjms.jar

If the checksum of the output is as above, you need to update it to the version used in BW services. The steps are as follows:

 Stop iProcess
 Backup the original file in SWDIR/jmslib/ems/tibjms.jar by renaming it to tibjms.jar.OriginalTechPlugin
 Overwrite the current file with the relevant tibjms.jar
 Start iProcess

Please review the tibjms.jar version by running the checksum and proceed accordingly.

SAL Session Timeout Value

Incase users are getting error after certain time accessing the system, sometime, there is an error stating that maximum user encountered.

In $SWDIR/seo/data/swentobjsv.cfg, there is a parameter called “SALSessionTimeout”. By default, this is set as 0, which disable the session timeout for inactive SAL sessions.

Initially we can recommend using atleast 900 (15 MINUTES for SALSessionTimeout). This value may vary based on the workload.

Segmentation Fault!!

In IPE version 10.6.x (Linux) , I have faced the segmentation fault error for so many times. Which happens when nscd thread is hung.

When attempting to startup iProcess Engine, either through swstart or failover script, it failed with “Segmentation fault” error. IProcess rely on the underlying mechanism for authentication. It is a known issue that sometimes, the nscd process hung and prevented iProcess Engine from starting up.

To resolve this, get the OS Admin to restart the nscd process and try again. We strongly recommend restarting the “nscd” daemon on periodically.

Index Rebuild Script for IPE database!!

For iProcess engine DB maintenance , you can schedule the Index rebuiding script on periodic basis.

Check the Index space before rebuilding
select sum(bytes)/1024/1024,segment_name from dba_segments
where tablespace_name='Data_Tablespace_Name'
and segment_type='INDEX' group by segment_name;

Rebuild the IndexDECLARE
cursor c_index IS
SELECT index_name
FROM all_indexes"
WHERE index_name not like '%AQ%'"
and tablespace_name = 'Data_Tablespace_Name';

BEGIN
FOR idx_Rec in c_index LOOP
EXECUTE IMMEDIATE 'alter index '||idx_rec.index_name||' rebuild online';
END LOOP;
END;
/

Check the status of Index Rebuild
select object_name,last_ddl_time from user_objects where OBJECT_TYPE='INDEX' order by last_ddl_time;

Check the Index space after rebuilding
select sum(bytes)/1024/1024,segment_name from dba_segments
where tablespace_name='Data_Tablespace_Name'
and segment_type='INDEX' group by segment_name;

Note : Ensure that index rebuild is not happening on AQ tables and AQ indexes.