concurrent mode failure - CMS & RMI
Hi,
I've been struggling with this concurrent mode failures for quite some time and I would be very nice if somebody could answer some more or less basic questions:
1.) If I scedule a full gc when using CMS do I get an concurrent mode failure? In my log I get one like every 60 seconds even as more than 50% of the heap is free, so I guess it's due to RMI?!
2.) If 1.) is true, what can I do? I can disable explicit gcs, but will there remain uncollected garbage or will there be any other disadvantage? I mean SUN won't do this ugly FullGC-scheduling-feature for fun, do they?
3.) If I need explicit gcs (or if I at least suppose it's better to schedule it manually in the night), what can I do in that case? As far as I understood the sun.rmi.dgc.server.gcInterval just set's a maximum interval! Actually I put it to a very high value, but I get the FullGC with a concurrent mode failure every 63 seconds, at memory usage where the CMS should not even kick-in.
Thanks a lot for any help,
Robert.
