IBM Mainframe Forum

Hi,
We have a sandbox system where the system often goes to X-Clock (i.e., it hangs). Have gone through some manuals where the solution suggest to purge the disabled task in CICS region. But i have gone through the same. Its perfect and i cant find any task in disabled mode. what could be the reason for this. Any suggestion as solutions. Can anyone help me on this. Apologies if m wrong.

Regards,
V

It sounds like you are talking about a CICS problem here - probably a loop. Can you look at the system with TSO and by going into SDSF, sorting CPU% D, look at which job is eating up your CPU? Do you have a monitor for CICS? With it you could look in your CICS and see what is looping. Try entering CEMT I TASK from the console. You really did not describe the problem clearly.

Hi Mogan,
Thanks for your post.
I have tried in prior what you have now adviced i.e., to monitor the task using CEMT I TASK from where i cant find any task getting into disabled state(i.e., because of looping) But still the TSO session hangs. Also have Gone through the SDSF and tried the same thing what you have mentioned. Everything is normal. What could be the reason? Hope i'm clear. Apologies if m wrong


Regards,
V

A monitor for CICS would be something like Omegamon or Mainview -- tools that are external to CICS but allow access to the region. Using CEMT I TASK is an internal to CICS monitoring method.

If the system is clocked, something is using up the CPU. If you suspect a CICS program it may well be looping with no CICS commands in the loop (hence CEDF would be useless to help debug the program). Such loops are very tough to debug without access to some sort of monitoring tool since invoking the program locks up the system. Once the program is identified, you have have to use visual inspection of the code to find the loop.

And why do you think the task should be in disable state, if it is looping it is running and goobling CPU. If you do not have a monitor you will just have to try killing the active tasks. That you even get a response from your CEMT requests shows that CICS is not in a total loop. Look for a task (not a CICS system task) that just keeps running and running with the same taskid. Why don't you post the response to your cemt i task?

Thanks again mogan for your post!!
Yes here is the response of my CEMT I TAS command from a CiCS terminal.
Tas(0000092) Tra(CKAM) Sus Tas Pri( 255 )
Sta(SD) Use(CICS1USR) Uow(C74CF1F8BB29FC84)
Tas(0017710) Tra(T077) Fac(0182) Sus Ter Pri( 001 )
Sta(TO) Use(TCHN077 ) Uow(C74F7045B900F984) Hty(FCPSWAIT)
Tas(0020996) Tra(CEMT) Fac(0015) Sus Ter Pri( 255 )
Sta(TO) Use(A201802 ) Uow(C74F9638EFCFB744) Hty(ZCIOWAIT)
Tas(0023169) Tra(CEMT) Fac(0026) Sus Ter Pri( 255 )
Sta(TO) Use(A251806 ) Uow(C74FAC984D001D04) Hty(ZCIOWAIT)

This was taken when i found when my system has gone into a X-clock condition. All the task that runs here are user tasks. Please help me out on this. Apologies if m wrong.

Transactions starting with a 'C' are (or should be) system transactions, which leaves you with only one possible culprit -- the T077 transaction. When the CEMT was done, the transaction was waiting for a file access.

Thanks robert for your reply!!!
I have a small doubt like the transaction was just suspended which means that the user has not terminated the transaction and the system is not processing the transactions and records related to that( this is up to what i know. pls correct me if m wrong plz don take it as a sin) how can we come to a conclusion like T077 will be the culprit. I m not able to understand that. Pls guide me on this. apologies if m wrong.

Regards,
V

Probably 95 to 99% of the time when you do a CEMT, the transactions will show as suspended. A task shows up as active when it is using the CPU -- and since most tasks spend most of their time waiting (for a terminal response, for a file read / write / delete / update / start browse / end browse, for a TSQ read / write, for a TDQ read / write, ...) they usually show suspended in CEMT.

Your CEMT showed 4 tasks -- two CEMT tasks, one CKAM task, and one T077 task. Since CEMT tasks don't go into hard loops, and CKAM (whatever that is) is a system task (note the user name is the CICS region name), it is very unlikely to be in a loop -- although if it is that's the easy fix since you could report the loop to IBM and THEY would resolve the loop and issue a fix for the program. The only other possible culprit would be T077, which is a user written program and hence much ore likely to be looping.

Now, T077 is in FCPSWAIT status. A quick Google search will tell you that FCPSWAIT is waiting for a private string. I would find out which file the wait is for (code inspection will usually help there) and have the CICS support group add some strings to the defintiion of that file (and find out why the current value is not enough).

Thanks a lot for your time and support!! got cleared!!!

Regards,
V