I tried to explain the issue in this stackoverflow question, but I will try one more time.
Our redis version is 3.2.8 and we have a default keepalive (which is 300 seconds for that version). We are using redis cluster mode. Our client port (6379) connections are having keepalive set, which can be seen in netstat -o command.
tcp 10.10.111.122:6379 10.10.111.121:50461 ESTABLISHED 2571/redis-server keepalive (103.28/0/0)
but for cluster bus port (higher port, 16379), it is not enabled:
tcp 10.10.111.122:16379 10.10.111.121:41227 ESTABLISHED 2571/redis-server 1 off (0.00/0/0)
Is there something I can do to enable keepalive for cluster bus port because currently we have a large number of connections hanging, or at least it looks like it?
Comment From: rajiv2205
I hope you found the solution for this. If not, tcp keepalive connections setting should also be there for cluster bus. Is there any reason, this setting is not present in redis.conf?