I have tried this solution but the effect is not good.
在 2013-03-26 16:48:34,"Emmanuel Lécharny" 写道:
>Le 3/26/13 9:44 AM, Jeff MAURY a écrit :
>> Even if you running on a single processor, as you are writing into file, I
>> would recommand to use a pool of threads to write onto files.
>> You curre
thanks,I will try another solution, A thread is responsible for receiving UDP
messages, one thread is responsible for writing documents, through the queue
communication between threads.
At 2013-03-26 16:44:00,"Jeff MAURY" wrote:
>Even if you running on a single processor, as you are writing in
Le 3/26/13 9:44 AM, Jeff MAURY a écrit :
> Even if you running on a single processor, as you are writing into file, I
> would recommand to use a pool of threads to write onto files.
> You current setting prevents MINA from reading the socket while writing the
> previous packet to the file.
I forgo
Le 3/26/13 2:59 AM, pongjy123 a écrit :
> with no packet loss,Mina UDP server receives packet can reach 10/s? Only
> receive packets and sequential write files, do not do anything else.
> 1/s cannot be achieved in my test environment,
>
>
> DatagramSessionConfig dcfg = acceptor.getSessionC
Even if you running on a single processor, as you are writing into file, I
would recommand to use a pool of threads to write onto files.
You current setting prevents MINA from reading the socket while writing the
previous packet to the file.
Jeff
On Tue, Mar 26, 2013 at 2:59 AM, pongjy123 wrote
with no packet loss,Mina UDP server receives packet can reach 10/s? Only
receive packets and sequential write files, do not do anything else.
1/s cannot be achieved in my test environment,
DatagramSessionConfig dcfg = acceptor.getSessionConfig();
dcfg.setReceiveBufferSize(64*1024);
sing