Skip to Content
0
Oct 06, 2020 at 01:00 AM

Available RFC Work processes reaching 0, All DIA processes getting occupied intermittently

629 Views

Hi Experts,

We are facing an issue where all DIA WPs are getting completely occupied and at times no work processes are available for RFC processing causing time-outs while using SICF services. This also causes increase in response time if seen from SMLG. The ABAP system is on kernel 753/601.

Massive Drops:

Below is my analysis on the issue so far.

SM21 for the issue duration

This hasn't come for a while and also parameters related to this were changed when the issue was observed last time.

http/security_session_timeout = 800

http/security_context_cache_size = 50000

SDF/MON:

ST02 Extended Memory detail

We have received multiple memory related dumps during the issue time-frame.

Looking at the SDF mon in detail as to what is utilizing all the resources, found some workprocesses with Status SEMAPHORE 39 with more than 1 sec runtime AND multiple processes occupied for sequential reads and deletes on table SECURITY_CONTEXT.

When SM66 was observed, work processes were in RED SEM39 status meaning in "waiting" status for the semaphore in question. Extended Memory, again!

"'HTTP Security Context Cache is full' messages are intermittent and occurred only during issue duration and not throughout the day."

Please can someone advise if I can check anything more related to memory and do I really need to increase it looking at the above.

Also, I got an idea to check one more thing related to memory - Shared memory and ...

Accumulated read locks have been increasing while I'm writing the question.

Please suggest anything you can. :-)

Regards,

Manoj Somkuwar

Attachments

mtbvj.png (266.5 kB)
0zb90.png (297.2 kB)
tvhbu.png (627.7 kB)
hbrgp.png (17.3 kB)
asiso.png (60.9 kB)
nryjo.png (207.5 kB)
hdnhk.png (218.6 kB)