RFC 7038914: VM could throw uncaught OOME in ReferenceHandler thread
peter.levart at gmail.com
Thu May 9 08:56:08 UTC 2013
On 05/09/2013 10:35 AM, Peter Levart wrote:
> In my experience, the thread dies between 10-20ms after interrupting,
> so waiting for about 500ms is enough I think.
But only if the test is started with low max. heap setting. When trying
with large heaps (2G+), a long GC pause can ruin Thread.sleep() enforced
ordering. The test also works with heaps as low as 16m, so perhaps it
should be started with -Xmx64m or even -Xmx32m...
More information about the hotspot-gc-dev