Hello Tim Armstrong,
I'd like you to reexamine a change. Please visit
http://gerrit.cloudera.org:8080/4113
to look at the new patch set (#6).
Change subject: IMPALA-3905: Add single-threaded scan node.
......................................................................
IMPALA-3905: Add single-threaded scan node.
Adds a new single-threaded scan node HdfsScanNodeMt that
materializes tuples in the thread calling GetNext().
The new scan node uses the HdfsScanner::GetNext() interface,
which is currently only implemented for Parquet.
As before, I/O is performed asynchronously via the I/O manager.
The new scan node is enabled if the mt_dop query option is
set to a value greater than 1. Otherwise, the existing
multi-threaded scan node is used.
The changes are mostly a refactoring of the existing multi-threaded
scan node to separate out the common code between the existing
multi-threaded scan node and the new single-threaded one.
Summary of changes:
- Move code from hdfs-scan-node.h/cc into a new hdfs-scan-node-base.h/cc
- Add new single-threaded scan node in hdfs-scan-node-mt.h/cc
- Both scan nodes inherit from HdfsScanNodeBase
- Rework the allocation of templates tuples such that the memory is drawn
from a new mem pool in the scanners, and that each scanner clones the
partition exprs contexts. Before, the memory was taken from the parent
scan node's mem pool, and there was only one instance of the partition
exprs contexts. Their access was protected under a lock, however, not
in all instances, so their use was not always obviously correct.
The change in this patch makes thread safety obvious and helps move
a lock into the multi-threaded scan node which would otherwise have
to remain in the HdfsScanNodeBase class.
- Simplify a couple of loops with C++11 for-each
Testing: A private core/hdfs run passed. I ran TPC-H/DS and test_scanners.py
on ASAN several times locally.
Change-Id: I98cc7f970e1575dd83875609985e1877ada3d5e0
---
M be/src/exec/CMakeLists.txt
M be/src/exec/base-sequence-scanner.cc
M be/src/exec/base-sequence-scanner.h
M be/src/exec/exec-node.cc
M be/src/exec/hdfs-avro-scanner.cc
M be/src/exec/hdfs-avro-scanner.h
M be/src/exec/hdfs-lzo-text-scanner.cc
M be/src/exec/hdfs-lzo-text-scanner.h
M be/src/exec/hdfs-parquet-scanner.cc
M be/src/exec/hdfs-parquet-scanner.h
M be/src/exec/hdfs-rcfile-scanner.cc
M be/src/exec/hdfs-rcfile-scanner.h
A be/src/exec/hdfs-scan-node-base.cc
A be/src/exec/hdfs-scan-node-base.h
A be/src/exec/hdfs-scan-node-mt.cc
A be/src/exec/hdfs-scan-node-mt.h
M be/src/exec/hdfs-scan-node.cc
M be/src/exec/hdfs-scan-node.h
M be/src/exec/hdfs-scanner-ir.cc
M be/src/exec/hdfs-scanner.cc
M be/src/exec/hdfs-scanner.h
M be/src/exec/hdfs-sequence-scanner.cc
M be/src/exec/hdfs-sequence-scanner.h
M be/src/exec/hdfs-text-scanner.cc
M be/src/exec/hdfs-text-scanner.h
M be/src/exec/scanner-context.cc
M be/src/exec/scanner-context.h
M be/src/exprs/expr-context.h
M be/src/runtime/tuple.h
29 files changed, 1,816 insertions(+), 1,468 deletions(-)
git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/13/4113/6
--
To view, visit http://gerrit.cloudera.org:8080/4113
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings
Gerrit-MessageType: newpatchset
Gerrit-Change-Id: I98cc7f970e1575dd83875609985e1877ada3d5e0
Gerrit-PatchSet: 6
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Alex Behm <alex.b...@cloudera.com>
Gerrit-Reviewer: Alex Behm <alex.b...@cloudera.com>
Gerrit-Reviewer: Tim Armstrong <tarmstr...@cloudera.com>
