On Tue, Sep 16, 2003 at 11:48:09PM +1200, David McNab wrote: > Hi, > > Something else that's arisen from my maintenance work with pyFreenet. > > With the older Build 5013, FCP's FEC commands and data transfer work > fine.
This is the same bug that I and some others have been exeperiencing -
the last guy tio report it (i forget his name now) was told that the
bug is known. It is, as far as I can tell, NIO related - the
in.read() (line #3018 in the cvs i have) of RecieveInputStream.read()
blocks forever when it shouldn't, in certain cases when it is called with
an empty accumulator. I presently blame this on a JVM bug, because the
code as read seems all good to me, however I was unable to assertain this
for sure, or come up with an acceptable fix.
This problem seems to come and go randomly, sometimes it will not work for
ages, and then it will cease to function for no good reason - additionally,
a large percentage of the population cannot reproduce this at all (I do
have, however, debug level logs of it). Then, one day, without upgrading
any of your node or your JVM, everything will start working for no reason.
And then, a few weeks later, it will stop again. This seems to effect
all FCP based apps if you're effected (I tested with FIW and Frost as well)
Short term, if you are really desperate, you can place a Thread.sleep(200);
line after the in.read(), and it will hackishly make the probelm go away
by ensuring that the data isn't read faster than the accumulator can be
fed. However, I have been thus far unable to enact a more reliable fix
when it has been working.
-- jj
pgp00000.pgp
Description: PGP signature
_______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
