-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/10699/
-----------------------------------------------------------
(Updated April 22, 2013, 9:27 p.m.)
Review request for Flume.
Changes
-------
Good catch Hari! Fixed.
We need unit tests for the timeZone stuff, I filed FLUME-2018 to track that.
Description
-------
This patch changes the behavior of the BucketPath.escapeString() function to
only ever call clock.currentTimeMillis() once during an invocation of the
method. This prevents a race condition that can cause unexpected results in the
interpolated paths.
This addresses bug FLUME-2014.
https://issues.apache.org/jira/browse/FLUME-2014
Diffs (updated)
-----
flume-ng-core/src/main/java/org/apache/flume/formatter/output/BucketPath.java
971c75c
flume-ng-core/src/test/java/org/apache/flume/formatter/output/TestBucketPath.java
9cfefc0
Diff: https://reviews.apache.org/r/10699/diff/
Testing
-------
Added unit test for the previous condition. All tests pass.
Here is some output from running the new unit test without the patch:
Running org.apache.flume.formatter.output.TestBucketPath
Tests run: 6, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 0.87 sec <<<
FAILURE!
Results :
Failed tests: testDateRace(org.apache.flume.formatter.output.TestBucketPath):
Race condition detected expected:<02:[5]0> but was:<02:[0]0>
Tests run: 6, Failures: 1, Errors: 0, Skipped: 0
Thanks,
Mike Percy
