curator-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jordan Zimmerman <jor...@jordanzimmerman.com>
Subject Re: Leader Selection and failing node
Date Mon, 04 Nov 2013 16:30:25 GMT
You need to give the clients all 3 instances. Otherwise, the clients won’t have a server
to connect to when you stop node_1.

-JZ

On Nov 4, 2013, at 7:19 AM, Sznajder ForMailingList <bs4mailinglist@gmail.com> wrote:

> Hi
> 
> I ran the LeaderSelectorExample / Example classes on a real Zookeeper chorum of three
nodes
> - node_1
> - node_2
> - node_3
> 
> I start ZK on theses nodes by using:
> bin/zkServer.sh start command on each one of three nodes (Is it necessary? Could I start
only one of them? )
> 
> 
> Then, I am launching the LeaderSelectorExample by giving it node_1:2181 as connect string,
on each one of the three nodes.
> I made two tests:
> 
> - I stop the process on node_1 , the two other nodes continue to take the leadership
one after the other, as usual.
> 
> - I stop ZOOKEEPER on node_1 by calling 'bin/zkServer.sh stop' on node_1.
> In this case, immediatly the current leader prints 
> "relinquishing leadership"
> 
> But **no one** takes the leadership then. And all the processes stay stuck...
> 
> 
> Thanks for your help
> 
> Benjamin
> 


Mime
View raw message