Did you solve this problem? I faced the same problem.
On 5/25/07, Gaston Dombiak <[EMAIL PROTECTED]> wrote: > > Hey Richard, > > We observed the exact problem. It seems like SocketIoProcessor is > failing to flush sessions and traffic is kept in > SocketSessionImpl#writeRequestQueue. Moreover, if you leave the server > running for a long time this will end up in an OOM. I'm still trying to > get to the bottom of this problem but I'm trapped in some other tasks as > well so I'm not able to debug this problem further. > > Any help is greatly appreciated. :) > > Thanks, > > -- Gato > > > -----Original Message----- > From: Richard Lowe [mailto:[EMAIL PROTECTED] > Sent: Thursday, May 24, 2007 12:44 PM > To: [email protected] > Subject: Packets drying up... > > Hi, > > I'm using Mina 1.1 under Java 5 for both my server and client. I'm > sending roughly 4 packets a second from the client to the server and > from the server back to the client, each one tends to be around 48 bytes > > or so. > > My TCP send and receive buffers are set to 128 bytes currently. > > It works great for a while (maybe five minutes), but then the packets > seem to dry up suddenly. > > The stream of packets from the server to the client start dribbling > through and I only get one packet every 5 seconds or so. > > At first I thought that they might be getting delayed on the Internet > somewhere, but it appears that isn't the case. > > 1. I ran TCPDump on the client and it really is only receiving a packet > every 5 seconds (almost exactly 5 seconds - regular as clockwork). > 2. I then ran TCPDump on the server and it's only *sending* one packet > every 5 seconds (again, like clockwork). > > My application log reveals that everything has been passed to the > IoHandlerAdapter write method - there's nothing left for my application > to send, so the data is getting queued somewhere deeper. > > I'm scratching my head here. The server is Windows 2000 and the client > is Mac OS X. > > If anybody has any suggestions, they would be most greatly appreciated! > > Many thanks, > Richard. > -- > > > > > > > > >
