We are seeing that during a high traffic tests we are seeing the members are being marked as DOWN form haproxy logs.
We are trying to debug the issue by collecting tcpdumps during high traffic case.
Is there a way to isolate health check messages from the tcpdump. This is a TCP health check.