[REEF-1508] Retrieve Job Status (and error details) through Reef client

• Type: Wish
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: REEF.NET, REEF.NET Client
• Assignee: Unassigned
• Reporter: Andrey
• Created: 04/Aug/16 21:03
• Updated: 07/Aug/16 00:40

Description

This is probably common problem for reef jobs.
I am exploring ways to propagate IMRU job status (specifically error details) to job client.
we want to retrieve job execution details (failure details to start with) using job ID.
To start with, it would be nice to collect information from events received by driver (failed task, failed evalutor, driver fail action) and expose it to the client.

• one option is to use driver's http endpoint to provide status, but if driver shuts down or is aborted, all information is lost.
• another option we started brainstorm offline is to define interface (observable in driver .... IStatusMonitor?) and allow client to provide implementation for the interface.

any other ideas? thoughts on how to provide client access to job details?

Issue Links

• is related to: REEF-908 Design .NET REEF client API