[IMPALA-10269] Failed Task Blacklisting

• Type: Task
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Component/s: Backend
• Epic Color: ghx-label-1
• Assignee: Wenzhe Zhou
• Reporter: Wenzhe Zhou
• Created: 20/Oct/20 22:05
• Updated: 20/Oct/20 22:10
• Resolved: 20/Oct/20 22:10

Description

• If node “a” has a higher rate of task failures compared to the rest of the cluster, then node “a” should be blacklisted
• There should only be a specific set of failures that count against a node - e.g. query specific failures like reading corrupted files, or mem limit exceeded should not count
• This is similar to how Spark Executor Blacklisting works