infinispan usage

[Russell Beall]

Hi Paul,

If you are still hunting for a more immediate solution, for your servers that are still using Terracotta, I highly recommend upgrading the EventingMapBasedStorageService object to use:
org.terracotta.collections.ConcurrentDistributedMap

instead of the ConcurrentHashMap.

I have released this to production and it is performing beautifully.

In testing this had better garbage collection performance for sure, however, I will add that I don't entirely believe it will handle your spurious two-minute garbage collection issues (even on small heaps).

If you like, I can send you the very simple patch info.  The patch can be used with unclustered nodes as well since that object simply defaults back to the ConcurrentHashMap object in an unclustered environment.

Regards,
Russ.


On Sep 7, 2011, at 11:34 AM, Paul Hethmon wrote:

> cluster. I'm running about 60 or 70 Shib clusters right now. Not all of
> them are configured for session support and hence need clustering, but
> enough of them. So I should be able to get away with only having to change