Laszlo Gaal has uploaded a new change for review.

  http://gerrit.cloudera.org:8080/4914

Change subject: IMPALA-4172: Switch to BlockLocation methods for disk IDs
......................................................................

IMPALA-4172: Switch to BlockLocation methods for disk IDs

This change enables Impala to use BlockLocation#getStorageIds,
the new call in Hadoop 3 for getting HDFS data block location
info from the NameNode. This call supercedes the old
getFileBlockLocations call which will be  deprecated in Hadoop-3.

The presence of BlockLocation#getStorageIds is determined dynamically.
This is necessary because the implementation was backported to
various Hadoop-2 releases after it appeared in Hadoop-3, so simple
version checking is not suitable to decide whether the call can be
used. In cases where both getFileBlockLocations and
BlockLocation#getStorageIds are supported, the code prefers the
latter for performance reasons: getStorageIds does not have to
query the DataNodes for this information.

BlockLocation#getStorageIds returns disk IDs as UUID-based strings,
which would be too expensive to ship around and store. This patch
maps these strings to small integers, preserving compatibility with the
existing representation of diskIDs.

Change-Id: Ibbff94cef9a9db7b3945f8e7b0286866d2cc3b61
---
M fe/src/main/java/org/apache/impala/catalog/HdfsTable.java
1 file changed, 155 insertions(+), 42 deletions(-)


  git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/14/4914/1
-- 
To view, visit http://gerrit.cloudera.org:8080/4914
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: newchange
Gerrit-Change-Id: Ibbff94cef9a9db7b3945f8e7b0286866d2cc3b61
Gerrit-PatchSet: 1
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Laszlo Gaal <laszlo.g...@cloudera.com>

Reply via email to