We have a process that runs on our ERP server every 2 hours that takes it offline for about 3-4 seconds. The main access to the server is Telnet based. We have a few people that seem to hit it about every time. If the software is idle, it doesn't affect it, if you hit a key it locks up until it has a chance to retry (I'm guessing) and it usually about 45-60 seconds. I put the STM in front of it hoping that it would help, and it seemed to at first but doesn't seem to be helping as much anymore. We also noticed a HUGE speed increase using the STM so we want to keep it in front. Are there connection settings that I can tweak to try to get it to retry faster?
We only have a single node and I have health monitoring disabled. I have node_so_nagle: set to yes.