[BIGTOP-1061] Provide a jstack capturing script

• Type: New Feature
• Status: Open
• Priority: Minor
• Resolution: Unresolved
• Affects Version/s: 0.7.0
• Component/s: Init scripts
• Assignee: Unassigned
• Reporter: Harsh J
• Created: 04/Sep/13 05:30
• Updated: 08/Sep/13 02:31

Description

To grab a jstack, a common thing maintaining people and devs equally do, we currently have to do a 3-step process:

1. Know which user to run the jstack as (hbase, hdfs, mapred, etc.)
2. Grab the PID using a tool (ps, jps, etc.)
3. Run jstack with the right JAVA_HOME path, cause some systems don't have its bin on the PATH.

I feel we could make Bigtop improve this process of debug-material collection, mainly jstack first. Bigtop service scripts and env-files know the right user behind a daemon, Bigtop service scripts have access to the PID, and Bigtop also uses JVM path detection. Perhaps we could add something like 'service hadoop-hdfs-namenode jstack' to grab a NameNode jstack as the right user with the right PID, and drop it into the configured log location.