Re: Custom TCP server after Restricted State (corrections)

Mark,

The "restricted state" term is just what our System's guy called it.
Specifically it's an unattended execution during an IPL. I am told that the error occurs after the Option 21 as well, since TCP is shutdown before the Option 21 backup.
I believe it is a safe assumption that it occurs during the TCP startup procedure, since I used ADDTCPSVR ... AUTOSTART(*YES) option.

I will make a note to try the job delay. My confusion comes from the fact that the HTTP server successfully starts before the TOMCAT server is started. I assume that with the HTTP server starting successfully that TCP is up and running.


Jason Abreu
Abreu Innovations, Inc.
jason.abreu@xxxxxxxxxxxxxxxxxxxx
http://www.abreuinnovations.com/

On 3/7/2011 12:48 PM, Mark S. Waterbury wrote:

Hi, again, Jason:

A few corrections to my earlier reply....

I wrote a small to test a SBMJOB of the QSH command, and was mistaken
about command substitution in this case. Sorry for any confusion caused.

When you say this only seems to "fail" after "bringing the system back
from restricted state" ... what exactly is the procedure used to bring
the system back up after any "restricted state" operations? After an
IPL, the system will automatically call the system start-up program
specified in the QSTRUPPGM system value, or the IBM-supplied default.
But, are you also calling this program when bringing the system back
from a "restricted state"?

How is TCP/IP getting re-started? If this is normally done via the "IPL
Attributues" (issue DSPIPLA to see the defaults), perhaps you may need
to issue STRTCP, STRTCPSVR, and / or STRHOSTSVR, after such a restart.
Some customers will include these extra commands into their custom
QSTRUP program, with a MONMSG CPF0000 to ignore any errors after each
one, and that way, they can just issue:

CALL QGPL/QSTRUP

to re-start all the subsystems, etc., after any "restricted state"
operations.

In the case where the IPL calls QSTRUP, those extra commands are
"redundant" (due to the IPL attributes) and so they may issue some error
messages, but the MONMSG will just ignore those errors.

Also, perhaps there are some other things that need to happen ... you
might want to add a small delay of say 5 minutes, to your batch job CL,
such as:
DLYJOB DLY(300)

to allow sufficient time for those other processes to start up.

HTH,

Mark