[YARN-5644] Define exit code for allowing NodeManager health script to mar

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 3.0.0-alpha2
• Component/s: nodemanager
• Labels: supportability
• Assignee: Yufei Gu
• Reporter: Ray Chiang
• Created: 13/Sep/16 16:06
• Updated: 13/Sep/16 17:23
• Resolved: 13/Sep/16 17:23

Description

Done as a alternate design to YARN-5567. Define a specific exit code for the health checker script (property yarn.nodemanager.health-checker.script.path) that allows the node to be blacklisted.

As discussed in the latter part of YARN-5567, the current design requirements are:

1. Ignore all exit codes from the script
   1. except the newly defined error code which will mark the NodeManager as UNHEALTHY
   2. This allows any syntax or functional errors in the script to be ignored
2. Upon failure (or multiple recorded failures):
   1. Store the status in the metrics2 state on the NodeManager
   2. Allow the RM to blacklist the NM or allow the jobs to drain

Issue Links

• duplicates: YARN-5635 Better handling when bad script is configured as Node's HealthScript