• Bug
  • Status: Closed
  • 2 Major
  • Resolution: Not a Bug
  • ehcache-core
  • cdennis
  • Reporter: ihrytsyuk
  • September 29, 2010
  • 0
  • Watchers: 0
  • July 27, 2012
  • March 22, 2011



We noticed DiskStore data file expanding (when DiskStore is used) under high concurrent load.


  • Cache configuration:
<cache name="test.METHOD_CACHE"
  • We have 4 predefined elements. 600 threads (150 * 4) invoke put/get operation on the cache with 300ms timeout. Each thread works with only one predefined element.

Our results: * test.METHOD_CACHE.data’s size is 400M after 5 hour of testing. But we expected size not more than 0.7M. * in case we decrease number of threads to 8 (2 * 4), all works as expected. We tested it only for 5 minutes. * in case we increase diskExpiryThreadIntervalSeconds to 60 - file size expands unpredictable over time. That is why we believe the issue isn’t related to diskExpiryThreadInterval.

Test application is attached.


Fiona OShea 2011-02-22

This has not been reproduced. Should we close and re-open if it happens again?

Chris Dennis 2011-02-23

I could not reproduce this on my laptop, nor on one of the 8-way dev machines, but that doesn’t mean it doesn’t happen. Since this is a user submitted issue, I’m going to ask the reporter if he can try to reproduce using 2.4.0, and then let him get back to me.

Chris Dennis 2011-02-23

I could not reproduce this on either my 2-core development machine or using an 8-way Linux box. If this is easily reproducible for you, could you give me some more information on the hardware you are using to reproduce it, and see if it still happens with the new 2.4.0 release.

Ivan Hrytsyuk 2011-02-24

Run attached application for 17 hours *on new machine*. After test test.METHOD_CACHE.data’s size was *9.6M*, not 400M as was on previous machine.

So it looks like issue is connected with hardware.

Current machine:

  • Ubuntu 10.04
  • 4GB RAM
  • Intel(R) Core(TM) i5 CPU 650 @ 3.20GHz
  • HDD 320GB ATA WDC WD3200AAKS-61L9A0

Fiona OShea 2011-03-22

Closed based on Ivan’s feedback