I don't know how your specific server works, but we have some EDI servers that spawn a gazillion threads, but only show 5 jobs in their subsystem. Starting and ending them always takes a while-- they do shut down all of those spawned jobs, then the subsystem jobs end.
We know (now) to give them extra time to end or start.
Paul E Musselman
From: MIDRANGE-L <midrange-l-bounces@xxxxxxxxxxxxxxxxxx> On Behalf Of Steinmetz, Paul via MIDRANGE-L
Sent: Wednesday, February 13, 2019 10:36 AM
To: 'Midrange Systems Technical Discussion' <midrange-l@xxxxxxxxxxxxxxxxxx>; 'Web Enabling the IBM i (AS/400 and iSeries)' <web400@xxxxxxxxxxxxxxxxxx>
Cc: Steinmetz, Paul <PSteinmetz@xxxxxxxxxx>
Subject: ENDTCPSVR SERVER(*HTTP) HTTPSVR(BRCWEBLIVE) issue - instance jobs not ending
I have an AJS scheduled job that recycles an HTTP instance once a week.
Runs every Monday morning, 3:30 AM.
10 ENDTCPSVR SERVER(*HTTP) HTTPSVR(BRCWEBLIVE)
20 DLYJOB DLY(80)
30 STRTCPSVR SERVER(*HTTP) HTTPSVR(BRCWEBLIVE)
Since we upgraded/migrated from P7-8205 to P9-9009, this process is failing.
The ENDTCPSVR for the instance is not ending the instance jobs, immediately.
Most are still running a minute later, so when the STRTCPSVR runs, STRTCPWVR fails with a TCP7410.
Message . . . . : HTTP Server instance BRCWEBLIVE is already active.
Cause . . . . . : The server is trying to use an instance file that is
already in use.
What's interesting, following the STRTCPSVR, all the instance jobs then do end immediately, all within one second.
IBM support thinks the STRTCPSVR is calling a background job that forces the end of the instance jobs.
No proof of this as of yet.
Another variable to this is the length of time the instance jobs are running.
When I tested the process with the instance running for 24 or 48 hours, everything works fine.
V7R3, latest cum, 2 months back on groups.
No errors found in any logs.
I have an open PMR with IBM.
Any thoughts from the group.
IBM i Systems Administrator