[
https://issues.apache.org/jira/browse/DRILL-4025?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14992829#comment-14992829
]
Mehant Baid commented on DRILL-4025:
------------------------------------
[~jnadeau] We aren't changing the behavior of checking if the cache file is in
sync with the actual data. That check is done a couple of line earlier in the
code
[ParquetFormatPlugin.readBlockMeta()|https://github.com/apache/drill/blob/master/exec/java-exec/src/main/java/org/apache/drill/exec/store/parquet/ParquetFormatPlugin.java#L229].
What we are avoiding in my patch is the additional ls in
[FileSelection.init()|https://github.com/apache/drill/blob/master/exec/java-exec/src/main/java/org/apache/drill/exec/store/dfs/FileSelection.java#L141]
to populate the FileStatus in the case it is null. However, I will run a small
test to confirm this and report the result.
> Don't invoke getFileStatus() when metadata cache is available
> -------------------------------------------------------------
>
> Key: DRILL-4025
> URL: https://issues.apache.org/jira/browse/DRILL-4025
> Project: Apache Drill
> Issue Type: Bug
> Affects Versions: 1.3.0
> Reporter: Mehant Baid
> Assignee: Mehant Baid
> Attachments: DRILL-4025.patch
>
>
> Currently we invoke getFileStatus() to list all the files under a directory
> even when we have the metadata cache file. The information is already present
> in the cache so we don't need to perform this operation.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
