Application stops working immediately when 1 node goes down


We have deployed Couchbase in 3 node cluster and auto-failover in 60 seconds.

So while a node goes down, applications stops working and returns error due to Couchbase connection failure for the first 60 seconds (works after new topology is received after failover)

But is there any way we can avoid this downtime since our application is running on critical real-time data.

Any suggestions would be greatly appreciated.

@nidhks the way couchbase works the only way to reduce this time is to decrease the auto failover interval (in most recent versions you can get it as low as 5s). Other than that your other option is on the read-only side you can read from a replica which then is not strongly consistent anymore in general.

@daschl Thanks for the quick reply.