Issues with Concurrent Connections VCMP i7800

Better check /var/log/ltm /var/log/kern and the tmm logs for a connection reaper because of memory issues or logs for the VIP.Also look for port exhaustion events or if the VIP cmp demoded  in the logs etc. Maybe also see if ePVA axceleration is in use and if attaching one connect profile to the VIP helps.

This can be also application/network bottleneck issue where the app can't handle the stuff so the server and app teams need to also do their jobs so tcpdumps could be needed to see if the TCP transmit window is getting increased of the latency etc as F5 TCP optimized profiles will help. The pools could use the same network and servers/apps so check this. An issue I have seen is with different Time_wait settings between the F5 and the servers.

https://support.f5.com/csp/article/K35612380

If you have BIG-IQ for central F5managment it will help as BIG-IQ can show you latency etc without tcpdumps:

https://clouddocs.f5.com/training/community/big-iq-cloud-edition/html/class3/module1/module1.html

You will not find an easy answer weithout deep investigation.