CPU increases in CICS regions when running with VANGUARD product

Most of the processing in the regions was occurring in VANGUARD module VCLFX2PC (PTF level VS62159).

VANGUARD support recommended PTF VS62457 as the resolution for this problem:

 ++PTF (VS62457)         /* VANGUARD INTEGRITY PROFESSIONALS */ .  
 ++VER (Z038)  
 FMID(AVSS212)  
 PRE(VS62159,VS62411)     
 COPYRIGHT: (C) COPYRIGHT VANGUARD INTEGRITY PROFESSIONALS 2015   
               LICENSED MATERIAL - PROGRAM PROPERTY OF VANGUARD

Stopping the started task VCLSTC1 also provided a quick circumvention. Contact VANGUARD INTEGRITY PROFESSIONALS if you need additional assistance.

Below are some of the diagnostics seen in a console dump of the CICS regions:

There are very few non-system tasks in the each region, just long-running tasks like CKAM, CKTI, etc.

IPCS command SYSTRACE PERFDATA confirms that in the window of MVS system trace covered, several CICS jobs were using most of the CPU.

I formatted MVS system trace and for each of the CICS jobs, sorted the system trace entries -- first filtering out everything but interrupt entries like EXT, I/O and CLKC, and then sorting by PSW address. The majority of PSW addresses were within Vanguard module VCLFX2PC (PTF level VS62159).

Here is how I sorted MVS system trace:

For each CICS job, format system trace entries for the selected address space ID (ASID). This can be done with IPCS command SYSTRACE asid(x'aaaa') where aaaa is the address space ID of the selected job.

Max to the bottom of the formatted output to make sure all system trace entries are included by entering the letter M followed by F8 key.

Then enter these commands, one at a time:

Report view

x all

find clkc all

find i/o all

find ext all

delete all x (this gets rid of everything but the selected entries)

sort 39 46 (column 39 is where the PSW addresses are, length 8)

Once you have only the sorted trace entries, page forward to see where the majority of PSW addresses are pointing.