This customer site is using DB2 for z/OS v7. The application is using DB2 plans only, no packages yet. The online system is very active, and during online, most of the locking is row locking , to allow more concurrency.
There were 2 incidents we observed that caused the application having many contentions, deadlocks and slow access.
Incident 1 - an online REORG was done on all the tablespaces, then on weekend, while CICS regions were shutdown, a REBIND of all the plans were executed. After this, the application had a systemwide locks suspended and extremely slow performance. We saw this problem during CICS online window, we did a BIND on the plans, and that fixed the issues. After the BIND Plans, the application ran smooth and fast again.
Incident 2 - wednesday night, drop a tablespace, recreate it as one of the column in the middle of the table needed to be increased from 2 characters to 10 characters. After the recreate, data loading, grant access, a BIND on the affected plans were performed. This was done during CICS shutdown too.
the next day, again, observed systemwide slowness, some affected plans not running fast enough, causing deadlocks. Resolution was during the CICS online window, we BIND the plans again to solve the slow and deadlocks issues.
I am trying to figure out the rootcause here , Incident 1 - why was it a REBIND after a REORG not working OK , DB2 seem stucked and confused with the access path, things all jammed up. Incident 2 - why wasn't the first BIND did right after the DROP TS/RECREATE TS tasks were done. The online seem stucked until a 2nd BIND was performed.
Does anyone encounter this kind of issues before ? any ideas , please...