The errors you listed may not be serious, as long as they are transient.
When you say 2 of the brokers are not responsive, are they issuing fetch
requests to the 3rd broker (look at the request log)? During a restart of
the whole cluster, brokers that are started later may not have any leader
and thus won't take any request from the client. You will need to run the
leader balance tool.


On Mon, Aug 26, 2013 at 10:12 PM, Vadim Keylis <[EMAIL PROTECTED]>wrote:
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB