[YARN-445] Ability to signal containers

• Type: Task
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: nodemanager
• Labels: BB2015-05-TBR
• Assignee: Unassigned
• Reporter: Jason Darrell Lowe
• Created: 02/Mar/13 18:04
• Updated: 09/Sep/15 19:06

Description

It would be nice if an ApplicationMaster could send signals to contaniers such as SIGQUIT, SIGUSR1, etc.

For example, in order to replicate the jstack-on-task-timeout feature implemented by MAPREDUCE-1119 in Hadoop 0.21 the NodeManager needs an interface for sending SIGQUIT to a container. For that specific feature we could implement it as an additional field in the StopContainerRequest. However that would not address other potential features like the ability for an AM to trigger jstacks on arbitrary tasks without killing them. The latter feature would be a very useful debugging tool for users who do not have shell access to the nodes.

Issue Links

• is depended upon by: YARN-896 Roll up for long-lived services in YARN
• is depended upon by: MAPREDUCE-5783 web UI update to allow people to request thread dump of a running task.
• is depended upon by: MAPREDUCE-5784 CLI update so that people can send signal to a specific task
• is duplicated by: YARN-1619 Add cli to kill yarn container
• is required by: MAPREDUCE-5044 Have AM trigger jstack on task attempts that timeout before killing them
• relates to: YARN-4131 Add API and CLI to kill container on given containerId
• relates to: HADOOP-9752 Latest Ubuntu (13.04) /bin/kill parameter for process group requires a 'double dash kill -0 -- -<pid>
• relates to: MAPREDUCE-5387 Implement Signal.TERM on Windows
• relates to: MAPREDUCE-5044 Have AM trigger jstack on task attempts that timeout before killing them

Sub-Tasks

• Provide ContainerManagementProtocol#signalContainer processing a batch of signals: Open