You have "." instead of ":" in a few places. > server.3=s1.2890:3890 > server.3=s2.2890:3890 > server.3=0.0.0.0.2888:3888
Is this causing the problem? Also, the message you asked about is coming from the quorum cnx manager, which connects servers when they're trying to elect a leader. I'd need to see the server logs to be able to say something more meaningful. -Flavio > On 24 Nov 2014, at 11:52, craig w <[email protected]> wrote: > > I am trying to get Zookeeper running on Kubernetes (http://kubernetes.io/) > and as a result each server in the quorum has a slightly different looking > config file. For example, in the 3-node quorum where I have 3 hosts (named > s1, s2 and s3): > > s1 zoo.cfg > ----- > server.1=0.0.0.0:2888:3888 > server.2=s1:2889:3889 > server.3=s1.2890:3890 > > s2 zoo.cfg > ----- > server.1=s2:2888:3888 > server.2=0.0.0.0:2888:3888 > server.3=s2.2890:3890 > > s3 zoo.cfg > ----- > server.1=s3:2888:3888 > server.2=s3:2889:3889 > server.3=0.0.0.0.2888:3888 > > As you can see the server.x lines are a bit different. With Kubernetes, > each host will have a proxy that forwards traffic the appropriate host. For > example, the proxy on s1 would have an entry that says "all traffic > destined for port 2888 will go to s1:2888", "all traffic going to 2889 will > go to s2:2888". > > When I start up the zookeeper servers in their logs I see things like: > > logging INFO "Have smaller server identifier, so dropping the connection: > (3, 1) and then WARN "Connection broken for id 1, my id = 1, error = > java.io.EOFException" > > Any ideas how as how to debug further to figure out what's going on? > > I am using Zookeeper 3.4.5. > > Thanks, > craig
