Troubleshooting WAS 8.5.5.19 Connection to DB2 on i

We have an app that runs on Websphere (v8.5.5 FP19). This app functions on
Server A, but has quirks. When attempting to perform process X on the app,
it takes a VERY long time to complete. On the developer's box, it has no
issue. We redeployed the app to Server B for testing under WAS 7.0.0 FP45,
and it works fine there. Both are Windows servers with minimal differences
in network configuration. This app opens connections to DB2 on i (V7R4, PTFs
current as of 6/18/21).



The development team though it was network related, so I went down my list
and performed traces on both Windows (Wireshark) and the IBM box (TRCCNN).
Both show the exact same thing: traffic flowing successfully, but then
several pairs of PSH,ACK packets with a "TCP ACKed unseen segment" message.
I'm 99% sure it is not network related, as both Server A and Server B are
successful in opening database connections, ping times are down to less than
1 ms, and routes are defined appropriately. I can't explain that string of
packets, however. Supposedly they are placeholders when viewing in Wireshark
that indicate data was not captured, but received by the sender of the ACKs.
They say the typical cause for this is a poor capture, but I don't really
think that is the case here.



We have scoured the SystemOut and SystemErr logs, enabled toolbox trace, and
examined the job logs for the connection pool. We have also used SQL
Performance Center in ACS to review the queries running at the time of the
bad process (Process X) in the app, but according to the developers, nothing
stands out and I/O time is acceptable.



Is there something else you all can recommend as far as logging is
concerned? I'm trying to get as much information out of this issue as
possible. I know the simple answer is that there is something wrong with the
WAS 8.5.5 configuration, but we can't figure out what it is.



TIA.



-----------------------------------------
Jacob Banda
-----------------------------------------