[
https://issues.apache.org/jira/browse/HIVE-17833?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16210423#comment-16210423
]
Prasanth Jayachandran edited comment on HIVE-17833 at 10/19/17 1:13 AM:
------------------------------------------------------------------------
Addressed review comments. Removed pom.xml changes. Will update it once tez
makes a release until then this patch can't be committed anyway.
2 counters one for #files that will be read by input and other for #directories
(did not call it partitions as we are not checking with partition spec).
#directories comes from jobConf via mapred.input.dir. So only #files has to
make sure the path in split is a file. There can be OrcSplit with delta
directory is not account for in any of these counters (expensive to do list
status on each delta directories here).
was (Author: prasanth_j):
Addressed review comments. Removed pom.xml changes. Will update it once tez
makes a release until then this patch can't be committed anyway.
2 counters one for #files that will be read by input and other for #directories
(did not call it partitions as we are not checking with partition spec).
#directories comes from jobConf via mapred.input.dir. So only #files has to
make sure the path in split is a file. There can be OrcSplit with delta
directory which don't account for in any of these counters (expensive to do
list status on each delta directories here).
> Publish split generation counters
> ---------------------------------
>
> Key: HIVE-17833
> URL: https://issues.apache.org/jira/browse/HIVE-17833
> Project: Hive
> Issue Type: Bug
> Affects Versions: 3.0.0
> Reporter: Prasanth Jayachandran
> Assignee: Prasanth Jayachandran
> Attachments: HIVE-17833.1.patch, HIVE-17833.2.patch
>
>
> With TEZ-3856, tez counters are exposed via input initializers which can be
> used to publish split generation counters.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
