Re: Trouble with Shootist TLS example
"Jeroen van Bemmel" <[email protected]> Mon, 12 Feb 2007 22:08:26 +0100
| Newsgroups | gmane.comp.voip.nist-sip |
|---|---|
| Message-ID | <00c201c74ee9$f04022b0$0601a8c0@BEMBUSTER> |
I've added a TLS test based on the reinvite case, under /src/test/tck/msgflow/callflows/tls Note that there is a hardcoded relative path to address the keystore Somehow we should make this case optional, since TLS is not mandatory for the TCK, right? Regards, Jeroen M. Rangnathan wrote: > Hello, Jeroen et al, > > Thanks to Yanick Belanger ( who has also been trying some performance > tweaking), this problem is resolved. It had nothing to do with the TLS > accelerator code. Rather it was a problem caused by another change > that had been made earlier. Can we write a regression test for TLS > transport so we will not run into this again? Any takers? > > Thanks Yanick for a quick turnaround on this one. > > Regards, > > > Ranga > > > Jeroen van Bemmel wrote: > >> Andrew, >> >> I'm having the same problem. You can see what is going on when you >> add '-Djavax.net.debug=all': something goes wrong with the TLS >> handshake, looks like some kind of version conflict (TLS1.0 versus >> SSL2.0). Not sure how/where to fix this, Ranga? >> >> Regards, >> Jeroen >> >> Andrew T Gin wrote: >> >>> I've tried changing each instance of "tls" to "tcp" in shootist.java >>> and shootme.java, and it works fine (it exchanges a sequence of >>> INVITES, ACKS, BYES, etc). >>> My understanding is that you can make it use TCP by simply >>> subsituting "TCP" wherever it has TLS, is that correct? >>> >>> I used my own keystore, as testPass is not the password of the >>> keystore, testKeyStore, given in the example. >>> >>> I did a wireshark trace, and found that after Shootist and Shootme >>> have exchanged what look to be encrypted packets, Shootme (listening >>> on port 5071) closes the TCP connection (FIN ACK). Shootist sends >>> another encrypted packet, but since the Shootme has closed its end >>> of the TCP connection, Shootist receives an RST. >>> >>> When I change everything back to TCP, everything works fine. Does >>> anyone have any idea what I could be doing wrong? >>> Is there more to using TLS other than changing the protocol used >>> from TCP to TLS and specifying the keystores? >>> Is is possible there is something wrong in JAIN SIP, which is >>> causing it to close the tcp connection prematurely when TLS is used? >>> >>> Thanks >>> Andrew T Gin wrote: >>> >>>> Hi >>>> Im having trouble getting the Shootist TLS example to work; It all >>>> compiles fine, and I run Shootme: >>>> >>>> $ java -Djavax.net.ssl.trustStore=testKeyStore >>>> -Djavax.net.ssl.trustStorePassword=testPass examples/tls/Shootme >>>> sipStack = gov.nist.javax.sip.SipStackImpl@ed0338 >>>> tls provider gov.nist.javax.sip.SipProviderImpl@16de49c >>>> >>>> Then I run Shootist: >>>> >>>> $ java -Djavax.net.ssl.trustStore=testKeyStore >>>> -Djavax.net.ssl.trustStorePassword=testPass examples/tls/Shootist >>>> createSipStack gov.nist.javax.sip.SipStackImpl@6e70c7 >>>> IO Error sending request >>>> javax.sip.SipException: IO Error sending request >>>> at >>>> gov.nist.javax.sip.stack.SIPClientTransaction.sendRequest(SIPClientTransaction.java:916) >>>> >>>> >>>> at examples.tls.Shootist.init(Shootist.java:390) >>>> at examples.tls.Shootist.main(Shootist.java:400) >>>> java examples.shootistTLS.Shootist >>>> >>>>>>>> is your class path set to the root? >>>>>>> >>>> $ >>>> >>>> What could be causing the IO error? >>>> _______________________________________________ >>>> nist-sip mailing list >>>> [email protected] >>>> http://www-x.antd.nist.gov/mailman/listinfo/nist-sip >>> >>> _______________________________________________ >>> nist-sip mailing list >>> [email protected] >>> http://www-x.antd.nist.gov/mailman/listinfo/nist-sip >> >> >> _______________________________________________ >> nist-sip mailing list >> [email protected] >> http://www-x.antd.nist.gov/mailman/listinfo/nist-sip