JDK-8241385 : java/awt/image/BufferedImage/ICMColorDataTest/ICMColorDataTest.java runs too long
  • Type: Bug
  • Component: client-libs
  • Sub-Component: 2d
  • Affected Version: 15
  • Priority: P3
  • Status: Closed
  • Resolution: Duplicate
  • Submitted: 2020-03-20
  • Updated: 2025-04-04
  • Resolved: 2020-03-20
Related Reports
Duplicate :  
JDK-8233028 - Investigate slow performance of ICMColorDataTest on Windows x64
Description
java/awt/image/BufferedImage/ICMColorDataTest/ICMColorDataTest.java
has a timeout of 300000 ! I wonder if the author thought that was
milliseconds, since 300,000 seconds is a very long time.
The test really does run a long time. In one of the FASTEST runs I
can see on a test job on Windows it took about 1hr 20mins.
[2020-03-19T09:07:00,184Z] TEST: java/awt/image/BufferedImage/ICMColorDataTest/ICMColorDataTest.java
[2020-03-19T09:07:00,184Z]   build: 2.294 seconds
[2020-03-19T09:07:00,184Z]   compile: 2.294 seconds
[2020-03-19T09:07:00,184Z]   main: 2514.554 seconds
[2020-03-19T09:07:00,184Z] TEST RESULT: Passed. Execution successful

Sometimes it runs slow enough that the entire jtreg job is timed out and killed

I don't know why this test needs to run so long but I don't think we can afford a test which probably is increasing the wall clock run time for the entire job from 30 mins to 90 minuts
Comments
We need this test to be dialled back - even at the expense of not being as thorough - to run in < 600 seconds on the slowest of systems because it is the long pole by a really long way. Probably this means dialing it back to 1,000 iterations instead of 300,000 iterations. If you can further reduce the max heap and not have the test fail, then that would be one way to keep it a stress test. We'll want to backport this to older releases too.
20-03-2020