Hello Everyone, We are using Zookeeper 3.5.5 in our environment. We don’t take frequent snapshots as our snap count value is 10M.
When we restart our cluster, follower node tries to read the snapshot from the disk which is too old while the latest transaction are in txlog. While the leader node after reading snapshot takes a new snapshot from txlogfile. Followers tries to get the difference from leader node but somehow I don’t see updated znode in follower instance which is causing issues in our application. Can anyone suggest how I can Ensure that leader should send the complete snapshot to follower or is there a way each instance can take a snapshot while the instance is stopped so that next time when they start they should read the latest snapshot file. Or each follower should load the last txlog ? The immediate resolution we are following is to remove the data dir for follower and start the instance. Doing this way it takes complete copy from leader. Thanks Srikant Kalani Sent from my iPhone
