You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@curator.apache.org by Kuldeep Bora <ku...@gmail.com> on 2014/04/07 12:28:42 UTC

LeaderLatch delay in acquiring latch

Hello,

I was trying the LeaderLatch recipe by attaching a LeaderLatchListener. I
am testing by running two different processes which connect to local
zookeeper. I noticed that the moment i kill one process there is a huge
delay until the `isLeader` method is called in second process. I also
checked the corresponding znode in zookeeper and it seems to retain the
temporary znode under the latch path for a really long time!

Are the nodes under the latch path not ephermeal?  What can be done to fix
this delay as it is impractically long for a ephermeal znode based latching
mechanism?

Regards