cluster forget of failed, disconnected nodes results in "handshake" nodes

[hokie...@gmail.com]

Hi Everyone,

In my testing of Redis cluster in version 3.0.5, I am finding that killed nodes remain in the cluster in failed, disconnected states. I attempted to manually remove them via cluster forget. However, these failed master and slave nodes in failed, disconnected state are replaced by nodes of type handshake, also in disconnected state.

Any ideas as to why cluster forget is yielding handshake nodes in disconnected state?

--John

[Tuco]

Are you running cluster forget <nodeId> on all nodes, or only on one node. you need to run it on all the nodes, ,as per this.

You could execute a shell script by  something like 

for ip in {127.0.0.1}; do for port in {7001..7003}; do redis-cli -h $ip -p $port CLUSTER FORGET <nodeIdOftheNodeyouwanttoforget>; done; done;