On Fri, 2007-11-16 at 16:29 +0000, Brad wrote: > Oleg, > > once again thanks for taking the time to respond. > > What I don't understand is that if it reuses sockets by default and > I'm not calling anything to close them, why is it running out? I'm > calling recycle (to reuse the same http connection) and yet it > consistently runs out of sockets around the 3200 iteration mark. To my > untrained eye, it looks like its using up every available port between > 1024 and 4999 which seems to be the highest port number windows is > willing to assign to it. > > I do agree that it would be impossible for it to re-use closed > sockets. What I don't understand is why they are being closed in the > first place. If commons-http is using pooling by default and I'm > calling recycle(), theoretically shouldn't it be able to execute http > methods indefinitely? > > I expect I'm missing something here and am sure I'll have an "oh > yeah!" moment any time soon! >
Brad, It takes two to tango. Both endpoints, client and server, need to be (1) able and (2) willing to keep the connection alive for it to be persistent between requests. It can well be the target server. You can examine HttpClient wire/context logs to see exactly why connections get closed in the first place. http://jakarta.apache.org/httpcomponents/httpclient-3.x/logging.html Hope this helps Oleg > Thanks, > Brad. > > On Nov 16, 2007 12:47 PM, Oleg Kalnichevski <[EMAIL PROTECTED]> wrote: > > > > On Fri, 2007-11-16 at 11:25 +0000, Brad wrote: > > > Oleg, > > > > > > thanks for your reply. > > > > > > So if I understand correctly, what that is implying in this situation > > > is that there is a limit to how many sockets can be opened and closed > > > in a given time period? > > > > > > > Brad, > > > > (1) As far as JVM is concerned sockets in the TIME_WAIT state are > > considered closed. Those sockets are kept active for a while by the > > underlying OS in order to catch stray packets after the connection has > > been closed by the JVM process. > > > > You need to tweak TCP/IP settings of the OS you are using if you want to > > reduce the period of time sockets are kept in the TIME_WAIT state. > > > > > > > If that's the case, then fair enough but I was hoping there would be > > > some way to make commons-http re-use a socket. That seemed to be the > > > whole point of the recycle() method. The documentation seemed to be > > > suggesting that it would use keep-alives to avoid opening a new socket > > > every time. Is there any way around this? > > > > > > > HttpClient can re-use connections and does that per default [1], but it > > obviously cannot re-use closed sockets. > > > > Hope that helps > > > > Cheers, > > > > Oleg > > > > [1] > > http://jakarta.apache.org/httpcomponents/httpclient-3.x/performance.html#Connection_persistence > > > > > > > > > Brad. > > > > > > On Nov 16, 2007 11:03 AM, Oleg Kalnichevski <[EMAIL PROTECTED]> wrote: > > > > > > > > > > > > On Fri, 2007-11-16 at 10:54 +0000, Brad wrote: > > > > > Hi, > > > > > > > > > > I'm using http-commons to make some calls to a http server. There are > > > > > situations where I may have to make a large volumen of calls in a > > > > > short period of time. > > > > > > > > > > My problem is that http-commons seems to be eating connections, > > > > > resulting in eventually getting this error > > > > > > > > > > java.net.BindException: Address already in use: connect > > > > > > > > > > A quick netstat from the command prompt shows thousands of sockets > > > > > with status TIME_WAIT. > > > > > > > > > > Here's my simple test code: > > > > > > > > > > public void testHttpGet() { > > > > > > > > > > HttpClient client = new HttpClient(); > > > > > GetMethod get = new GetMethod(); > > > > > > > > > > try { > > > > > > > > > > for (int i = 0; i < 10000; i++) { > > > > > > > > > > get.setPath("http://localhost:8081/";); > > > > > int code = client.executeMethod(get); > > > > > byte[] responseBody = > > > > > get.getResponseBody(); > > > > > get.recycle(); > > > > > System.out.println(i + ", response: " > > > > > + code); > > > > > } > > > > > > > > > > } catch (Exception e) { > > > > > e.printStackTrace(); > > > > > } finally { > > > > > get.releaseConnection(); > > > > > > > > > > } > > > > > } > > > > > > > > > > I checked the tutorial and I've made sure to read the response body. I > > > > > call recycle() in the loop and yet it still uses a new connection > > > > > every time. > > > > > > > > > > All help very much appreciated. > > > > > > > > > > > > > Please see item 4 in the micro FAQ below > > > > > > > > http://tangentsoft.net/wskfaq/articles/debugging-tcp.html > > > > > > > > Oleg > > > > > > > > > > > > > Cheers, > > > > > Brad. > > > > > > > > > > --------------------------------------------------------------------- > > > > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > > > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > > > > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > > > > > > > > > --------------------------------------------------------------------- > > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > For additional commands, e-mail: [EMAIL PROTECTED] > > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
