[AURORA-370] The killTasks RPC should not wait until tasks are killed

• Type: Story
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.5.0
• Component/s: Client, Scheduler
• Assignee: Maxim Khutornenko
• Reporter: Maxim Khutornenko
• Created: 29/Apr/14 20:06
• Updated: 16/May/14 03:43
• Resolved: 08/May/14 00:51

Description

The killTasks RPC is currently waiting for a task to get killed actively blocking the client thread. With the killTaskMaxBackoff set at 1 minute the delayed killTasks call effectively destroys any multiprocessing attempts on the client.

Move the kill wait to the client potentially reusing the existing JobMonitor.

Issue Links

• blocks: AURORA-350 Parallelize updates to speed up deploys