Support Questions
Find answers, ask questions, and share your expertise
Announcements
Alert: Welcome to the Unified Cloudera Community. Former HCC members be sure to read and learn how to activate your account here.

NameNode Heap size

Highlighted

NameNode Heap size

Explorer

Team

I got below alert, Seems NN heap size was occupied too much. how to resolve this issue

The variance for this alert is 1,143MB which is 27% of the 4,224MB average (845MB is the limit)

4 REPLIES 4
Highlighted

Re: NameNode Heap size

Rising Star

@suresh krish

These two links contain most of the information that is needed for tuning the Namenode heap.

Alerts come from Ambari, so if you want to tune that instead of Namenode heap, you can do that in Ambari.

https://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.4.2/bk_installing_manually_book/content/ref-809...

https://community.hortonworks.com/articles/14170/namenode-garbage-collection-configuration-best-pra....

Highlighted

Re: NameNode Heap size

Cloudera Employee

@suresh krish

There are three parameters on which is alert is triggered :

Growth Rate for Critical ,Growth Rate for Warning and Minimum Heap

Minimum Heap : if the heap usage of namenode is less that configured value then it will not trigger any message.

Growth Rate for Warning : We give a value for this parameter in %. Which means that if the deviation in heap usage of namenode is above the % of "Average of Namenode Heap Usage" then it should trigger Warning alert.

For Ex : The variance for this alert is 1,143MB which is 27% of the 4,224MB average (845MB is the limit)

Growth Rate for Critical : 50%

Growth Rate for Warning : 20 %

Minimum Heap 100 MB

Namenode Average heap usage : 4224 MB

Variance : 1143 MB

This means that 1143MB is the deviation of heap usage which is less than 27% of 4,224MB which is Average heap usage for your cluster.

So 1,143MB is greater than 845MB, hence it will trigger an alert.

======================

We cannot really tune the NN heap based on this alert as this shows only the daily/weekly variance. It could happen if there was too much activity on the cluster just for that day, also I have seen this coming when we restart namenodes.

Please analyze the NN log and the GC log to see if there are any really long GC pauses and consistent over heap utilization, If everything looks normal I would just reconfigure the thresholds for these alerts in ambari. Please check this link to see if the NN heap meet the recommendations.

You can configure the ambari alert thresholds as per your requirements by analyzing Average namenode heap usage on your cluster. You need to identify or you can set a benchmark for heap usage and configure this alerts as per it.

Re: NameNode Heap size

Explorer

I recently upgraded to Ambari 2.5 . Is it a new feature

Highlighted

Re: NameNode Heap size

Cloudera Employee

No, It is available in Ambari 2.4 also, just verified it.

Don't have an account?
Coming from Hortonworks? Activate your account here