[IMPALA-8571] Make query-hook-execution more robust and observable

• Type: Sub-task
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: Frontend
• Epic Color: ghx-label-7
• Assignee: radford nguyen
• Reporter: radford nguyen
• Created: 22/May/19 17:30
• Updated: 13/Feb/20 21:59

Description

The execution of {{QueryEventHook}}s currently has some drawbacks in the design:

• exceptions thrown from hooks are simply logged by the Frontend and ignored
• hooks that hang will forever block an executor thread (currently fixed-size threadpool)
• an unbounded queue is used for scheduled hook tasks, which means that slow/hanging hooks may cause the queue to grow and grow until an OutOfMemoryError is thrown
• metrics around hook execution are not captured (which would really help in debugging/diagnosis)

These are all points that should be addressed/improved in this ticket

Code Review: https://gerrit.cloudera.org/c/13748/

Issue Links

• causes: IMPALA-8914 Collect and publish QueryEventHook metrics
• is caused by: IMPALA-8473 Refactor lineage publication mechanism to allow for different consumers
• supercedes: IMPALA-8573 Implement timeout for query hook execution