CDH 5.11x RHEL OS upgrade

[#1]

Hi,

  Anyone recently upgraded RHEL 7x OS on CDH 5.11x that can share some information please? Few details about

- any gotchas that they ran into

- was the approach to upgrade each node of the cluster

- was the cluster taken offline for RHEL OS upgrade for the entire duration while upgrading all nodes of the cluster 

- etc

Best,

NaniSK

[#2]

@NaniSK,

Here are some general guidelines:

https://www.cloudera.com/documentation/enterprise/5-11-x/topics/cm_ag_upgrading_os.html

That page memntions JobTracker, but you can replace that with Resource Manager and it should still apply.

I have not upgraded the OS myself, so I don't have any extra tips.

[#3]

@bgooley, Thanks for your reply. Would you be able to guide me were I might be able to find dfs.heartbeat.interval and dfs.heartbeat.recheck.interval in CM 5.11.0 please? Idea is to increase this interval so that the block copy can be avoided if needed be. Best, NaniSK

[#4]

@NaniSK,

You can set this "NameNode Advanced Configuration Snippet (Safety Valve) for hdfs-site.xml" by adding:

<property>
<name>dfs.namenode.heartbeat.recheck-interval</name>
<value>600000</value>
</property>

with a value in milliseconds.

by default the value is 300000 (5 min).  The algoithm for checking will be 2x that so it ends up being 10 minutes.  600000 would be 20 minutes

After setting it, you can do a rolling restart to apply the change.