The creation of the ephemeral znode `/controller` is probably a bit different from the creation of other ephemeral znode. The broker which creates the ephemeral znode `/controller` is explicitly specified in the znode data. Thus the old approach, which reads the broker id from the controller znode after seeing `Code.NODEEXISTS`, seems OK. And that old approach seems to handle the network connection loss scenario described here. I am wondering if we can use the old approach since its logic looks simpler. What do you think?
[ Full content available at: https://github.com/apache/kafka/pull/5101 ] This message was relayed via gitbox.apache.org for [email protected]
