orphaned QZDASOINIT jobs --> SQL below machine interface (MI); broken TCP connection

Really appreciated this comment from Rochester Support. One day our
system spawned more than 100 extra QZDASOINIT jobs during a period when
TCP connection was dropping to web server.

Message ID . . . . . . : TCP2617


Message . . . . : TCP/IP connection to remote system 10.0.0.39 closed,

reason code 1.

Cause . . . . . : The TCP/IP connection to remote system 10.0.0.39 has
been
closed. The connection was closed for reason code 1. Full connection

details for the closed connection include:

- remote IP address is 10.0.0.39

Reason codes and their meanings follow:

1 = TCP connection closed due to expiration of 10 minute FINWAIT2
timer.

Yes, it is certainly possible that a broken TCP connection can orphan

a QZDASOINIT job. For example, a client connects and sends a SQL
request. The QZDASOINIT job passes this request to SQL code which goes
below the > MI (machine interface) to satisfy the request from the
database. While the job is below the MI running the query, the
connection to the client drops. Sometime later the job tries to come up
above the MI to go back to the client > > but the client is long gone as
the connection was broken.

You can tell this is the case by looking at the jobs callstacks. If

there is no callstack, then the job is stuck below the MI and should be
ended.

Brad Menges, Advisory Software Engineer
Rochester Support Center - System i5 Performance