Support Questions

Find answers, ask questions, and share your expertise
Announcements
Celebrating as our community reaches 100,000 members! Thank you!

Cloudera Director 2.1 on aws with centos 7: ntpd not started

avatar
Contributor

hello,

 

I have installed a cluster on EC2 (centos 7), CDH: 5.8, Services: HDFS, Hive, Hue, Impala, Oozie, Sqoop 2, YARN, ZooKeeper.

ntpd was not started on the cluster nodes, in order to fix it I started and registered it as service.

 

Is it on purpose? This means I should have my images with ntpd already registered as service or are you planning to add this in the CDH installation scripts?

 

Johann

1 ACCEPTED SOLUTION

avatar
Super Collaborator

Hi kerjo,

 

We're actively looking into the issue of ntpd on CentOS 7 and RHEL 7. What we understand is that for those OS versions, chronyd has replaced ntpd as the preferred time synchronization daemon. It's likely that the AMI(s) you are using have chronyd installed and that is causing a conflict.

 

If chronyd is running on the cluster nodes despite how ntpd behaves, then you should be in the clear. You won't need ntpd running.

 

Otherwise, until we adjust our instance normalization logic, you can define an instance template bootstrap script for the cluster nodes that installs and starts chronyd, and makes it enabled. I think that even though Director's normalization logic will subsequently still attempt to work with ntpd, chronyd will still run and be unaffected. You can find information about bootstrap scripts in the sample configuration files, documented here: www.cloudera.com/documentation/director/latest/topics/director_cluster_config.html.

 

yum install chronyd
systemctl enable chronyd.service
systemctl start chronyd.service

Hopefully that will get things going for you. If you still end up with neither ntpd or chronyd running on the cluster nodes, or if Director reports problems working with ntpd due to the conflict, let us know and we can investigate some more.

 

Bill

View solution in original post

2 REPLIES 2

avatar
Super Collaborator

Hi kerjo,

 

We're actively looking into the issue of ntpd on CentOS 7 and RHEL 7. What we understand is that for those OS versions, chronyd has replaced ntpd as the preferred time synchronization daemon. It's likely that the AMI(s) you are using have chronyd installed and that is causing a conflict.

 

If chronyd is running on the cluster nodes despite how ntpd behaves, then you should be in the clear. You won't need ntpd running.

 

Otherwise, until we adjust our instance normalization logic, you can define an instance template bootstrap script for the cluster nodes that installs and starts chronyd, and makes it enabled. I think that even though Director's normalization logic will subsequently still attempt to work with ntpd, chronyd will still run and be unaffected. You can find information about bootstrap scripts in the sample configuration files, documented here: www.cloudera.com/documentation/director/latest/topics/director_cluster_config.html.

 

yum install chronyd
systemctl enable chronyd.service
systemctl start chronyd.service

Hopefully that will get things going for you. If you still end up with neither ntpd or chronyd running on the cluster nodes, or if Director reports problems working with ntpd due to the conflict, let us know and we can investigate some more.

 

Bill

avatar
Contributor

Thanks for the feedback, the point is that after the configuration of the cluster performed by Cloudera Director there was no time synchonization setup.

 

Cheers

Johann