We (zab and I) are still trying to track down a bug causing NIO nodes
to freeze after some hours. This appears to be caused by a rare JVM bug,
whose effect is to cause notify/wait synchronization to break - a
notification doesn't get delivered, and the waiting thread waits
forever. If you have had your NIO node freeze, we would like to know
a) What JVM you were using (e.g. Sun 1.4.2)
b) Whether it was the server or the client VM
and ideally
c) Send us a debug level log

After the workaround, if the same bug is triggered, you will see, in the
log and sometimes on stderr:

Waited more than 1 minute to dequeue <some number> in queue, <some
number> millis since enqueued last item, <some number> maximum waits so
far - could indicate serious JVM bug


If so, we really would like to be told about this too, including the
items above.

If you run a reasonably busy node for 48 hours with the NIO code, and it
does not crash or hang like this, we want to know about that too.

We especially welcome reports on nodes running 1.4.2 on M$ Windows.
-- 
Matthew J Toseland - [EMAIL PROTECTED]
Freenet Project Official Codemonkey - http://freenetproject.org/
ICTHUS - Nothing is impossible. Our Boss says so.

Attachment: pgp00000.pgp
Description: PGP signature

Reply via email to