Hi, We have been having cases recently were people cannnot connect to our ActiveMQ (4.1) when using proxies. After lots of digging, i have found that the issue may be in this class: org.apache.activemq.transport.tcp.TcpTransport and especially in the connect() method. There we call remoteAddress.getAddress() which will cause a remote address to be resolved. This means a trip to the dns server to get the address.
In restrictive environments, where only a proxy can do this, this dns lookup will fail and we see something on the DEBUG level that looks like this: DEBUG [28.08.07 15:38:33.690] Connect fail to: [someaddress:someport], reason: java.lang.NullPointerException (btw, we are using failover transport) The code "works on my machine" because i am allowed to resolve external ip addresses. But some people's computers are not allowed and they have to connect to the proxy to do this. A final proof that this issue is at play is that we have http transport configured (with SSL) and if they open the Http transport URL on their browser, they get notified about a certificate. but via activeMQ, the connection fails. My only workaround for this is to put the host's ip in the user's /etc/hosts file or in C:\WINDOWS\system32\drivers\etc To reproduce the problem, use the hostname that cannot be resolved to an IP in your connection URI. Is this a bug? or a feature? ;) Regards, Saqib -- View this message in context: http://www.nabble.com/ActiveMQ-Tries-to-resolve-ip-adresses-to-names-which-failes-when-only-a-proxy-should-do-that-tf4354047s2354.html#a12406430 Sent from the ActiveMQ - User mailing list archive at Nabble.com.
