Hi Gunnar, I believe I see something from the logs: 1) goto line 190 in the log, time index 09:00:59.357 2) search backward in the log file for the znode you are getting "nonode" on, this: /gyrex/prefs/cloud/org.eclipse.gyrex.http/applications/org.eclipse.gyrex.http.equinoxhttpservice.application 3) time index is now 09:00:58.968 (the most recent earlier operation on this znode), notice the operation being performed on the znode - it's explicitly being deleted by the client.
This (3) is the cause of the log message from (1). So from what I see in the log it looks like expected behavior from the ZK view of the world. Does this make sense to you or am I missing something? Regards, Patrick On Thu, May 19, 2011 at 11:57 PM, Gunnar Wagenknecht <[email protected]> wrote: > Am 19.05.2011 21:14, schrieb Patrick Hunt: >> I'd also suggest creating a JIRA to track this issue: >> https://issues.apache.org/jira/browse/ZOOKEEPER > > Filed as > https://issues.apache.org/jira/browse/ZOOKEEPER-1065 > > I also attached a debug log there. > > Thanks, > Gunnar > > -- > Gunnar Wagenknecht > [email protected] > http://wagenknecht.org/ > >
