You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for your note. I ust did a quick test on an HDP setup running on Centos7/java 8 and was able to install Zeppelin without having to do this workaround.
Installing is not the problem. try to reboot the node and see if the service comes up again.
During install, the install is able to create the pid directory. But the service restart expects the dir to be there and on centos7, this is now a tmpfs meaning it needs to be recreated every boot and the zeppelin user that starts the service has no permission to write in the default /var/run.
It's not dependant on the java version. It is a followon result of moving from upstart to systemd.
The same problem has occurred in my environment on CentOS7.
It looks "ambari-zeppelin-service/package/scripts/master.py" file's problem.
Maybe, In the "start" function, it needs to create the pid directory before executing "zeppelin-daemon.sh".
Even though CentOS 7 is not yet supported by this, there is an easy solution for this issue.
Carefull: Make sure you stop the zeppelin service before making the below changes!
Go to the config page of the Zeppelin service in Ambari and search for
(It's under Advanced zeppelin-env).
Change the value to
Where is the UID of the zeppelin user (find it using id -u zeppelin).
For me, the end result was /var/run/user/1008/zeppelin-notebook
The service should start normally now on CentOS 7.
The text was updated successfully, but these errors were encountered: