[news.eclipse.technology.ohf] Re: Query Document Failure using the OHF

[Matthew Davis <mattadav@xxxxxxxxxx>]

Ok, are you working with the source code?  If so, please try this:

Comment out lines 194-196 of 
org.eclipse.ohf.ihe.common.ws.AbstractOHFSOAPSender.java and try again 
with your proxy settings set as System properties... e.g.

System.setProperty("http.proxyHost", proxyHost);
System.setProperty("http.proxyPort", proxyPort);

It should then obey those proxy settings for non-TLS requests.

If this works, I'm going to have you try a couple more things and then 
we'll install a permanent fix for both non-TLS and TLS connections 
(although, again, I'm not sure you would ever proxy a TLS connection).

Thanks,
-Matt



Matthew Davis wrote:
> Jason,
>
> Yep, what you found in AtnaAgent.getSocket() is indeed how we control 
> the socket construction.  This allows us to manage, at runtime, the 
> certificates used in TLS authentication on a per-domain basis.  However, 
> it's not a requirement for non-TLS connections, it's only there for 
> consistency.
>
> I'm going to experiment with a couple things and let you know the results.
>
> Thanks!
> -Matt
>
>
> Jason wrote:
> > Hi Matt
> >
> > First, thanks for your answer!
> >
> > Actually, i did some more tests on this problem. After hours of 
> > research it seems that that the problem seems to be on a lower level. 
> > Normally you create an instance of the B_Consumer or Consumer object 
> > which is then able to create a valid query request. The creation of 
> > this message seems to be ok, except of the sending part. The 
> > B_Consumer::invokeStoredQuery contains a send call which fails. So 
> > after going deeper you can see that the lower level 
> > AbstractXDSSOAPClient::send method fails on the part where it tries to 
> > send the message. So after going deeper it seems that the 
> > AbstractOHFSOAPSender fails on a specific method call. The problem 
> > sections seems to be the "configureTransportProtocol" method on the 
> > part where the getSocket method of the AtnaAgent is invoked. Thats the 
> > point where an exception is thrown. ... Infact, the endpoit adress 
> > seems to be valid and the chunked parameter is configured. Actually, 
> > it seems to be mysterious problem.
> >
> > A possible workaround...
> > It seems that a working workaround would be to write transaction like 
> > the iti-14 (where we set the proxy in the same way) to straight 
> > forward the well formated query to the registry. So the best way is to 
> > replace only the sending call of the B_Consumer which has then the 
> > effect that this transaction works properly.
> > By the way, it seems that the proxy never gets reached by the typical 
> > sending call of the B_Consumer class.
> > The whole test id done without mutual TLS
> >
> > Hope this helps to find the problem
> > Greetings jason