[AMBARI-12548] Ambari Agent does not register / heartbeat if "df" command hangs

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.1.0
• Fix Version/s: 2.1.1
• Component/s: ambari-agent
• Assignee: Dmytro Sen
• Reporter: Dmytro Sen
• Created: 27/Jul/15 09:31
• Updated: 27/Jul/15 12:26
• Resolved: 27/Jul/15 10:21

Description

The agent executes the "df" command when it starts up.
In rare scenarios, this command can hang (can even hanging for several days without any timeout). When this happens the agent won't move forward to the process of registration, etc.
There's an ambari-server property called agent.check.mounts.timeout. This can be defined to force a timeout.
However, we should always have a default timeout on the agent script side, even if this property is not defined to avoid a situation where the agent is completely hung.

Issue Links

• links to: Review Board