Redis redis 3.2.8 cluster opens too many TCP connections

HI, I am running Redis Server 3.2.8 in cluster mode. I run 3 nodes on 1 host and 3 nodes on 2nd host. They form cluster. I see that around 400 odd socket connections are opened between these 2 hosts. I find it too much & in the past i had seen these connections reaching around 30k which caused complete downtime of the database.

Host 1 : [root@localhost REDIS_SCRIPTS]# ps -eaf | grep redis root 12273 1 0 Nov09 ? 00:03:42 redis-server 172.18.85.7:2240 [cluster] root 12296 1 0 Nov09 ? 00:03:44 redis-server 172.18.85.7:2241 [cluster] root 12328 1 0 Nov09 ? 00:03:41 redis-server 172.18.85.7:2242 [cluster] root 19038 5951 0 18:12 pts/0 00:00:00 grep --color=auto redis [root@localhost REDIS_SCRIPTS]# [root@localhost REDIS_SCRIPTS]# netstat -anp | grep redis | wc -l 353 [root@localhost REDIS_SCRIPTS]#

Host 2 :+1: [root@localhost guest]# ps -eaf | grep redis root 3299 1 0 Nov09 ? 00:03:28 redis-server 172.18.87.10:7002 [cluster] root 19542 18320 0 18:11 pts/1 00:00:00 grep --color=auto redis root 30569 1 0 Nov09 ? 00:03:30 redis-server 172.18.87.10:7000 [cluster] root 31887 1 0 Nov09 ? 00:03:29 redis-server 172.18.87.10:7001 [cluster] [root@localhost guest]# [root@localhost guest]# netstat -anp | grep redis | wc -l 429 [root@localhost guest]#

How do we figure out the reason behind these connections ?

Comment From: trevor211

Maybe it's easier and not error prone to establish 2 connections between 2 redis nodes.