BDA Cluster Checks with "mammoth -c" Result in Very High Disk Consumption on /tmp Causing NameNode Health to Become "Bad". (Doc ID 1683113.1)

Last updated on APRIL 30, 2017

Applies to:

Big Data Appliance Integrated Software - Version 2.3.1 to 3.0 [Release 2.3 to 3.0]
Linux x86-64

Symptoms

Running several iterations of the BDA cluster verification with "./mammoth -c" to resolve test errors, results in the following:

1. Very high disk consumption in /tmp of Node 1 of the cluster . This can be measured by:

a) Reporting the disk space usage of /tmp with:


2. The disk consumption can become so high that /tmp on Node 1 of the cluster becomes 100% full.

3. This results in the NameNode on Node 1 of the cluster going into safemode.  This causes Cloudera Manager (CM) to report NameNode health as "Bad".

4.  The DataNode on Node 1 gets into "Bad" health.

5. Deleting files in /tmp, brings the cluster into "Concernng" rather than "Bad" health.

6. Cluster services do remain available throughout.

Cause

Sign In with your My Oracle Support account

Don't have a My Oracle Support account? Click to get started

My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms