Hi

the developers in my organisation have made a web service call
the problem i am facing is that this webservice call just stops functioning in a while. we have to just restart the jboss instance and it starts functioning again.

everytime restart is not the solution for this.
can anyone suggest me a work around for this or a way to find the cause why it suddenly stops.

the entire code is written in java and is hosted on centos 5.3 server.

any help will be appreciated

If you're not a Java developer and if the WAR deployed according to specs (if any) then if the WAR doesn't already log debug nfo somewhere to look at or Jboss doesn't throw any exceptions you can fling back at them then I would allow the responsible developers in your organization to troubleshoot the issue themselves. (Maybe my opinion is colored by past experiences but IMO the only way to make developers understand is to "teach" them by letting them feel the pain of having to debug their own mess.) BTW Centos is at 5.5 now with 5.6 hitting repos RSN.

I checked with my developers , they want say that the web call is made , but we dont know whether the call reached the remote server. is there a way where in I can check if the call is made to the remote server and trace where it is being dropped.

I am not a developer , I am a linux system admin, ya i know this setup was done long back , i am in process of upgrading the hardware as well as the OS.

if you can help me trace where the issue lies using tcpdump or some other command would really be good

If you say "this webservice call just stops functioning in a while" to me this means it does work (right?) but for an undisclosed period of time before crashing. That would mean that you could elicit information from it in terms of debugging Java performance (garbage collection etc) and maybe Jboss returns a nice stack trace to send to your developers? In any case you'll have to find out how your "web service call" makes its calls. If it does this all by itself then you'll have to find out how to get (or your developers to supply a version that generates) debug output. If it relies on an application that makes SIP calls then you'd check that (logs?). Before you're going to compare a packet capture of a successful SIP call with one that isn't (http://wiki.snom.com/SIP_Traces"]exaple[/url]) you'll probably also want to look at several SIP-level trace tools like WIST, a Web Interface for SIP Trace (includes "siptraced"), sipsak, the SIP swiss army knife or if the SIPtrace and Statistics Monitor from OpenSIPS-CP works for you. That's about all you're going to get from me as I don't do SIP. And I still maintain your developers should share the burden of solving this problem with you: don't leave stuff that b0rks on deployment unpunished.