CA7ONL Task could experience an hang condition in submission, with the message :
CA-7.780 POSSIBLE CA-7 HANG CONDITION. CODE = 1
and only performing a Task Cancel and Restart (TYPE=ERST), the situation could be reset back to normal.
How to find a possible cause and bypass for this kind of problem
Release : All
Component : CA-7
This can happen when the use of CCI terminals is very high and a lot of DEMAND commands are issued at the same time. In addition it could be that the DEMANDed jobs would probably coming from the same JCL data set.
Each of these jobs requires a buffer in CA-7 storage and the size of this buffer is the DCB blksize value of the JCL library where the job resides.
In this scenario it is easy to have a spike in the use of the internal CA-7 storage, even if the buffers are dynamically gotten and freed as needed because there are many DEMANDed jobs at the same time.
It is possible to check the CA-7 pools dimension using the command
and search for NSP pool regarding this situation, but there is no report that shows the usage of the CA-7 pools.
Usually this type of processing is "normal", but depending on the many things that CA-7 does in a certain situation/environment, it could cause this kind of problem.
As it is normal to execute many concurrent DEMANDs, in order to avoid this problem, the best thing might be to decrease the blksize of the JCL data set, maybe from 32720 to 16000.