You can create 3 different data directories and point your configuration files to them. You can then open 3 shells, and each shell will run its own server - you can point it to the right config file by using the ZOOCFG parameter, like: export ZOOCFG=zoo_replicated3.cfg Or you can run them all in one shell and provide the config file path as a parameter to zkServer.sh.
Here's an example of zoo_replicated3.cfg, pointing to the zookeeper3 data directory: tickTime=2000 dataDir=/Users/shralex/zookeeper-3.4/zookeeper3 initLimit=5 syncLimit=2 clientPort=2791 server.1=localhost:2780:2783:participant server.2=localhost:2781:2784 server.3=localhost:2782:2785:participant Alex No, it's not for testing. I'm not familiar with Curator's TestingCluster. I would like to simulate the production environment. On Apr 22, 2013, at 3:25 PM, Jordan Zimmerman <[email protected]> wrote: > If this is just for testing, use the TestingCluster from Curator. > > On Apr 22, 2013, at 3:23 PM, Drew Kutcharian <[email protected]> wrote: > >> Hi Guys, >> >> For development purposes I would like to run a 3 node Zookeeper ensemble on a single machine (Mac OS X). What's the cleanest way to do this? >> >> Thanks, >> >> Drew >
