This is not correct, 3 is a minimum for redundancy. If 1 goes down, the other 2 can still form a quorum (as there are more than half of them remaining). There is likely a configuration issue in the original posters case. Either that, or some sort of networking issue, in which the remaining 2 nodes cannot communicate with each other.
On Fri, Jun 13, 2014 at 9:35 AM, James A. Robinson <[email protected]> wrote: > 3 is the minimum you can run, if you lose one > the remaining 2 cannot reach consensus on > who should be the leader. > > So my understanding has been that one > should probably run at a minimum of 5 > nodes. > > > > On Thu, Jun 12, 2014 at 1:07 PM, Lahiru Gunathilake <[email protected]> > wrote: > > > Hi Users, > > > > I have followed the cluster guide and created a zookeeper cluster with 3 > > nodes. After a successful start of the cluster I have created few nodes > and > > everything looked fine. I see one leader and two followers as we expect. > > > > Now I killed the leader process in machine one and I am waiting for rest > of > > the two processes to pick a new leader and start serving my requests, > but > > it seems two followers are simply trying to connect to each other and > > throwing errors and whole cluster is not working. Here is my zoo.cfg > file. > > I tried to restart my leader to see whether cluster comes up > automatically, > > neither is happening. Am I starting too less server nodes or what is the > > issue with my setup. > > > > tickTime=2000 > > initLimit=10 > > syncLimit=5 > > dataDir=/home/sgg/zookeeper/zookeeper-3.4.6/data > > clientPort=2181 > > server.1=149.160.172.65:2888:3888 > > server.2=156.56.179.104:2888:3888 > > server.3=gw110.iu.xsede.org:2888:3888 > > > > Can someone please correct me if I am doing something wrong. > > > > Thanks > > Lahiru > > -- > > System Analyst Programmer > > PTI Lab > > Indiana University > > >
