[DRILL-6495] Fragment error message profile dumped into log file. Why?

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.13.0
• Component/s: Functions - Drill
• Assignee: Unassigned
• Reporter: Chun Chang
• Created: 13/Jun/18 22:11
• Updated: 13/Jun/18 22:11

Description

When a query fails for some reason, we dump the following gigantic json profile into drillbit.log. Why do we do this? Has anyone found this info useful? It completely clutters the log file. If the profile contains crucial information, I recommend finding an alternative.

2018-06-13 14:47:31,094 [24de6f10-0e19-1769-a136-76b15ce5b832:frag:2:7] INFO o.a.d.e.w.fragment.FragmentExecutor - 24de6f10-0e19-1769-a136-76b15ce5b832:2:7: State change requested CANCELLATION_REQUESTED --> FINISHED
2018-06-13 14:47:31,094 [24de6f10-0e19-1769-a136-76b15ce5b832:frag:2:7] INFO o.a.d.e.w.f.FragmentStatusReporter - 24de6f10-0e19-1769-a136-76b15ce5b832:2:7: State to report: CANCELLED
2018-06-13 14:47:31,095 [24de6f10-0e19-1769-a136-76b15ce5b832:frag:2:7] WARN o.a.d.exec.rpc.control.WorkEventBus - A fragment message arrived but there was no registered listener for that message: profile {
state: CANCELLED
minor_fragment_id: 7
operator_profile {
input_profile {
records: 4096
batches: 1
schemas: 1
}
operator_id: 10
operator_type: 29
setup_nanos: 0
process_nanos: 2786560384
peak_local_memory_allocated: 54050816
wait_nanos: 47930232
}
operator_profile {
input_profile {
records: 4096
batches: 1
schemas: 1
}
operator_id: 8
operator_type: 10
setup_nanos: 1705434
process_nanos: 379744084
peak_local_memory_allocated: 105226240
wait_nanos: 0
}
operator_profile {
input_profile {
records: 0
batches: 0
schemas: 0
}
operator_id: 12
operator_type: 42
...
...