How to update NameNode Configuration with HA enabled?

[#1]

I need to update NameNode Heap Size because we keep receiving NAMENODE_PAUSE_DURATION alerts. 

I'm going to change this configuration in cloudera Manager: HDFS -> Configuration -> NameNode -> Java Heap Size of Namenode in Bytes. 

I believe Cloudera Manager will ask to restart the NameNode service. Will the restart break the clients?

Or I should do manually.

• Restart StandBy after updating the configuration.
• Manually failover to StandBy.
• Restart Active,
• Then manually failover to Active.

I'm running CM/CDH 5.3.2 with HA enabled. 

[#2]

Hi! I'm going to implement the same procedure. Did it work for you?

[#3]

@szamasz

If you have enabled automatic failover you dont need to manually failover , just stop the namenode that you gonna the change the configuration wait for the other namenode to become active so that you are not disrupting the users after that again start the namenode that you stopped before. 

[#4]

Hi,

You can try performing "Rolling Restart" of your service. 

For details refer, Rolling Restart - "During a cluster restart, to avoid having your NameNode (and thus the cluster) be unavailable during the restart, Cloudera Manager forces a failover to the standby NameNode."

Hope that helps.

[#5]

@bwang,

Your steps are fine, but it appears that you are concerned about clients.

HDFS client code is written in a way that allows it to fail over to another NameNode.  While we cannot predict all circumstances, clients should be able to detect the namenode state and use the new active NameNode more or less seemlessly.