Hello Techpyassa, Sometimes old gossip information tends to echo around for quite a bit longer than intended. I'm unsure how long the LEFT messages are supposed to be echoed for, but if you want to force the removal of a removed node from gossip, you can use the Assassinate Endpoint JMX command. On larger clusters, running this command synchronously across all machines may be required. Instructions on Assassinate Endpoint can be found here:
https://gist.github.com/justenwalker/8338334 If you're planning on recommissioning the same node, upon bootstrapping the gossiped message should change to a JOINING message overwriting the LEFT message. I've personally never checked `nodetool gossipinfo` before recommissioning a node and typically only ensure the node does not appear in `nodetool status`. Hope that helps, Joaquin Casares Consultant Austin, TX Apache Cassandra Consulting http://www.thelastpickle.com On Sun, Sep 25, 2016 at 2:17 PM, Laxmikanth S <techpya...@gmail.com> wrote: > Hi, > > Recently we have decommissioned nodes from Cassandra cluster , but even > after nearly 48 hours 'nodetool gossipinfo' still shows the removed nodes( > as LEFT). > > I just wanted to recommission the same node again. So just wanted to know > , will it create a problem if I recommission the same node(same IP) again > while its state is maintained as LEFT in 'nodetool gossipnfo'. > > > Thanks, > Techpyaasa >
