GitHub user manishgupta88 opened a pull request:
https://github.com/apache/carbondata/pull/2560
[HOTFIX] Removed file existence check to improve dataMap loading performance
**Problem**
DataMap loading performance degraded after adding file existence check.
**Analysis**
When carbonIndex file is read and carbondata file path to its metadata
Info map is prepared, file physical existence is getting checked every time
which in case of HDFS file system is a namenode call. This degrades the dataMap
loading performance. This is done to avoid failures for 2 scenarios
1. Compatibility with 1.3 version store where segment file contains
mergeIndex as well as index file name even though index file physically do not
exist after creation of merge Index file.
2. Handle IUD scenario where after delete operation carbondata file is
deleted but the entry still exists in index file.
**Fix**
Modified code to check for physical file existence only in case when any
IUD operation has happened on the table
- [ ] Any interfaces changed?
No
- [ ] Any backward compatibility impacted?
No
- [ ] Document update required?
No
- [ ] Testing done
Verified in cluster on 20 billion data
- [ ] For large changes, please consider breaking it into sub-tasks under
an umbrella JIRA.
No
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/manishgupta88/carbondata
query_slow_executor_pruning
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/carbondata/pull/2560.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #2560
----
commit 7081c8d480f580414583beef8798a5e3a208f549
Author: manishgupta88 <tomanishgupta18@...>
Date: 2018-07-25T14:18:41Z
Removed file existence check to improve dataMap loading performance
----
---
