impala-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tim Armstrong (Code Review)" <>
Subject [Impala-ASF-CR] IMPALA-5890: Abort queries if scanner hits IO errors
Date Tue, 12 Sep 2017 16:14:04 GMT
Tim Armstrong has posted comments on this change.

Change subject: IMPALA-5890: Abort queries if scanner hits IO errors

Patch Set 8:

File be/src/exec/

Line 180:     state_->LogError(ErrorMsg(TErrorCode::SEQUENCE_SCANNER_PARSE_ERROR,
Unfortunately it looks like when abort_on_error=1 we depend on this status wrapping to add
a filename and offset for some parse errors, e.g. "Table schema is not a record". I think
the filename is pretty essential to understand and fix any errors so I think we should be
careful not to drop it.

Ideally I think all parse errors from the scanners would just include that context when original
constructed. Not sure how much of a project it would be to go through all the sequence scanners
and fix that. It's much cleaner using LogOrReturnError().

Maybe in the meantime we should just do:

    if (!status.IsCancelled() && !status.IsMemLimitExceeded() && !status.IsDiskIoError())
    state_->LogError(ErrorMsg(TErrorCode::SEQUENCE_SCANNER_PARSE_ERROR, ...)
File be/src/runtime/

Line 544:     return Status(TErrorCode::DISK_IO_ERROR,
> Should we re-use DISK_IO_ERROR for those, even though they're technically n
I think it makes sense to use DISK_IO_ERROR for now if we interpret its meaning broadly as
"an error that prevented the disk I/O manager completing a request range".

To view, visit
To unsubscribe, visit

Gerrit-MessageType: comment
Gerrit-Change-Id: I44dc95184c241fbcdbdbebad54339530680d3509
Gerrit-PatchSet: 8
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Lars Volker <>
Gerrit-Reviewer: Dan Hecht <>
Gerrit-Reviewer: Joe McDonnell <>
Gerrit-Reviewer: Lars Volker <>
Gerrit-Reviewer: Tim Armstrong <>
Gerrit-HasComments: Yes

View raw message