[MESOS-9711] Avoid shutting down executors registering before a required resource provider.

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 1.8.0, 1.9.0
• Fix Version/s: 1.8.0, 1.9.0
• Component/s: agent
• Labels: agent mesosphere provider storage
• Target Version/s: 1.8.0, 1.9.0
• Sprint: Storage: RI-13 Sprint 44
• Story Points: 1
• Assignee: Chun-Hung Hsiao
• Reporter: Benjamin Bannier
• Shepherd: Greg Mann
• Created: 09/Apr/19 11:48
• Updated: 11/Apr/19 17:04
• Resolved: 11/Apr/19 17:04

Description

If an HTTP-based executor resubscribes after agent failover before a resource provider exposing some of its resources has subscribed itself the agent currently does not know how to inform the resource provider about the existing resource user and shuts the executor down.

This is not optimal as the resource provider might subscribe soon, but we fail the task nevertheless.

We should consider improving on that, e.g., by deferring executor subscription until all providers have resubscribed or their registration timeout is reached, see MESOS-7554.

Issue Links

• is related to: MESOS-9667 Check failure when executor for task using resource provider resources subscribes before agent is registered
• is related to: MESOS-7554 Add re-registration timeout for local resource providers
• relates to: MESOS-8467 Destroyed executors might be used after `Slave::publishResource()`.