| |
| |
| |
|
Page: 1 2 3 4 5
Comments:
<0> are you closing each connection ?
<1> server has been running fine for years ...
<1> i think the problem started when we upgraded to java 1.5
<0> I mean are you closing the connection each time a connection is over ?
<0> in your source.. what does your exception part say ?
<1> and yes we are closing sockets ...even set linger time to 0 to see if it solves the issue
<0> you have any exceptions ?
<1> yes
<0> put it
<0> here.
<2> is recursion and "ok" way to do file searchs on a given directory ?
<0> vsDude: could be.
<1> }catch (SocketTimeoutException sto){
<1> retry++; // this is to count how long client has been idle
<1> } finally {
<1> removeClient();
<1> }
<2> are there other ways ? i cant seem to find any
<0> if an exception is thrown,, why not close the connection ?
<1> removeclient has the closing socket part
<1> mySocket.shutdownInput();
<0> ok
<1> mySocket.shutdownOutput();
<1> mySocket.close();
<0> try sto.printStackTrace() after retry++
<0> and paste the exception here.
<1> client is supposed to be closed after retry reaches a certain limit
<0> Orpheus: you making any progress with Gnu/Linux installation reading ? ;-)
<1> btw ...
<1> when server is working fine ... there are no CLOSE_WAIT sockets in netstat
<0> CLOSE_WAIT is just another fancy name for waiting for the sockets to be closed.
<2> the thing is, i wrote a recursive method to do file search, works fine. Now, i want to put it in a runnable but i seem to be stuck. I need to somehow notify the caller that the search is over and that results may be retrieved.
<0> vsDude: avoid threads if you can.
<1> I know ... the output from netstat is the same for last hour or so after the server stopped accepting responses
<2> really ? i just dont want to slow down the rest of the app...
<0> threads are good if you want to fork multiple processes and there is always a danger of locking up if you code it wrongly.
<1> what might cause serversocket to now accept new connections? besides closing or throwing an exception?
<0> it's hard to know without looking at source.
<1> now = not
<1> while(listening){
<1> try{
<1>
<1> Socket temp = server.accept();
<1>
<1> SocketHandler client = new SocketHandler(this,temp);
<1> client.start();
<1>
<1> }catch(Exception t){
<1> System.out.println("LoungeServer: problem with ServerSocket or creating a socket handler");
<1> t.printStackTrace(System.out);
<1> }
<1> }
<1> System.out.println("LoungeServer: not listening anymore");
<1> this is server code
<0> where is your listen() ?
<1> beleive it ot not, exception was never thrown
<2> Is there another way to do the search without blocking the main thread ?
<1> server = new ServerSocket();
<1> server.setReuseAddress(true);
<1> server.bind(new InetSocketAddress(listenPort),100);
<1> u mean this part?
<0> Where are you running this code ?
<0> http://java.sun.com/j2se/1.4.2/docs/api/java/net/InetSocketAddress.html#InetSocketAddress(java.lang.String,%20int)
<1> ?
<0> new InetSocketAddress("IP_ADD", int port);
<1> I have this server.bind(new InetSocketAddress(listenPort),100);
<0> each time you are creating new InetSocketAddress, you are not giving it an IP address!
<1> no no
<1> my mistake ..
<1> the last 3 lines are called once .. prior to the loop
<0> what IP address are you trying to bind it to ?
<1> but when u asked about the listen() part ... i thought u wanted me to post the part where I bind the server socket
<1> localhost
<0> try this server.bind(new InetSocketAddress("127.0.0.1",listenPort),100);
<1> the server works perfectly... it only stopps acceptign connections for reasonse that are unknown after running for a few days
<1> doesnt seem to be related to memory ...
<0> the system must have rebooted..lol
<1> nor is it a deadlock because the server keeps servecing those that are still connected ..
<1> doesnt seem like the system reached the FD limit
<0> what about maximum number of connections ?
<1> there is nothing on the java community website or bug website
<1> it reached 300 concurrent conenctions with no problems ...
<1> but when problem occurs ... it doesnt seem to be related to the number of connected people ... there is no specific number
<0> why not move your removeClient() part inside finally { } instead ?
<1> however, we notice it when the number of connections is around 300
<1> it is inside finally
<0> as i said, move the removeClient() method's contents to finally { //here } and then write a for loop in your client to simulate 1000 connections.
<0> ???
<1> will this make a difference?
<0> Trying will not harm..
<0> when you run out of ideas, exhaustive coding will help.
<1> I understand the simulaiton ..
<0> writing your client is easy (will take 10-15 minutes max)
<1> but moving content of a method doesnt seem to make any difference ..
<0> what's your removeClient() ?? post it here
<1> cause eitherway ..the same code will be called ... there are no scopr issues here to make me think that this will make a difference
<0> since you don't want the source to be seen, I can't help you.
<0> put the whole source at rafb.net
<0> the server part.
<0> or Server cl*** if you prefer
<0> ?
<1> http://rafb.net/paste/results/W0H3a929.html
<1> im afraid its too big :)
<1> been developing the system for the past 5 years
<0> mySocket.shutdownOutput(); what's that doing ?
<0> and mySocket.shutdownInput(); as well.
<1> closes this
<1> outStream = new DataOutputStream(mySocket.getOutputStream());
<1> it used to be outStream.close()
<0> why did you remove that then ?
<1> but thought I'd use this function that was introduced in 1.4 to see if it helps with the CLOSE_WAIT sockets staying there
<1> its the same
<1> problem was there in either case :)
<0> all streams are supposed to be closed( ) no matter waht.
<0> that's your problem there.
<1> yes ... this functoin closes it
<0> you have obvioulsy not closed it. that's why it's left in hanging connection.
<1> ok
<0> put the source.
<0> put the mySocket.shutdownOutput(); source here
<1> will try it and see hoe things go
<1> one sec ..
<1> http://java.sun.com/j2se/1.5.0/docs/api/java/net/Socket.html
<1> shutdownoutput will be there
<0> As I said, you are suppose to move all the contents of close() to finally { //here }. every connection will close no matter what.
<1> thank u
<0> yeah.. but why not use close() instead? I don't understand.. you are serving someone something and then when it is done, just use close ( )
<1> I will
<0> ok
<0> goodluck
<1> the function might be doing something other than what I was expecting
<1> thank u .. and thank u for your time
<0> no prob.
<1> but just incase someone else comes with a similar problem ...
<1> this might be a bug
<1> I'll let u know if the change fixes the problem
<0> maybe. ok.
<0> I don't go to using newer versions as soon as they are released..
<0> i just wait. ;~P
<3> Hello
<3> No Java expert here ?
<1> I'll keep that in mind :D
<3> I didn't understand you C[_]
<0> xNokia: don't worry abt it. ;-)
<2> starsky, would you have any suggestions to give me ? I mean, as it is right now, the recusive search method is not in a thread anymore, works ok, but im thinking of what would happen if there are many files to search through
<0> vsDude: use loop.
<3> starsky thanks
<3> anyone can help here with Java Secure Socket Extension ?!
<0> vsDude: start at your parent directory and check the file according to how you want it to be checked (e.g., if it is a directory => f.isDirectory() ).. etc..
<0> xNokia: you could go to other server.. no one here seems to be interested in helping you. ;)
<4> !ping
Return to
#java
or
Go to some related
logs:
Xzistent
#mirc
#php
#linux
dvd writer problems mandriva
#php
how to enable dvi output linux
teh 3r2k
fsoap
Issa Dante
|
|