It is possible that 63039 is used by other processes, as Mac can not detect
whether this port is used or not from my experience.
And can you reproduce this issue consistently ? Because each time zeppelin
will chose a random port as long as you don't specify the portRange, so if
you reproduce it consistently, it might due to other issues.
小野圭二 <onoke...@gmail.com>于2018年3月2日周五 下午1:32写道：
> Hi there,
>
> I am so confusing now, let me know why if You know this, please.
>
> The env:
> Mac: High Sierra ver.10.13.3
> Zep: 0.7.3
> Spark: 2.2.0
> Java: 1.8
> notebook script: %pyspark print 'aaa' (just it for test)
>
> In this env, i got the error in log as
>
> ERROR ({pool-2-thread-2} Job.java[run]:184) - Job failed
>
> org.apache.zeppelin.interpreter.InterpreterException:
> org.apache.zeppelin.interpreter.InterpreterException:
> org.apache.thrift.transport.TTransportException: java.net.ConnectException:
> Connection refused (Connection refused)
>
>
> Indeed, before running notebook, RemoteInterpreterManagedProcess.java
> prepares a Port for callback to Thrift.
>
>
> INFO ({pool-2-thread-1} RemoteInterpreterManagedProcess.java[start]:130) -
> Thrift server for callback will start. Port: 63039
>
>
> I guess, the reason why the error was Zeppelin could not attache to the
> Thrift PORT.
>
>
> And my confusion is, when i run the same module on CentOS, it worked fine.
> CentOS: 7
> Zep: same
> Spark: same
> Java: same
> notebook script: same
>
> And there was no the Thift port description in the log file. From this
> consequence, the fine working was cause of did not use Thrift port on
> CentOS.
>
>
> Do You know actually there is different behavior between Mac and CentOS?
> And i am so appreciated if You know the workaround on Mac.
>
>
> Best,
> Ono
>
