[MOBY-dev] Updates to MOBY Central and IMPORTANT NOTES forregistry hosts

Tue Feb 6 15:03:31 UTC 2007

Hmmmm... Well that's a bit of a disaster!  

Could you check what happens when you call one of my services (eg MOBYSHoundGetGenbankff)?  I wonder if SOAP::Lite versions don't talk to each other properly (which would be a nightmare, but given our experiences yesterday, not surprising!)

M

--
Mark Wilkinson
...on the road!

-----Original Message-----
From: Romina Royo <rroyo at lsi.upc.edu>
Date: Tue, 06 Feb 2007 15:25:15 
To:Core developer announcements <moby-dev at lists.open-bio.org>
Cc:mobydev <moby-dev at biomoby.org>
Subject: Re: [MOBY-dev] Updates to MOBY Central and IMPORTANT NOTES for
 registry hosts

Hello,

I updated the moby-live/Perl libraries from cvs and I installed 
SOAP::Lite 0.69.

Here are some of the results I got:

- When executing a synchronous Perl service (function execute in 
Client/Service.pm) I got this error:

<SOAP-ENV:Fault><faultcode>SOAP-ENV:VersionMismatch</faultcode><faultstring>Wrong 
SOAP version specified. Supported versions:
  1.1 (http://schemas.xmlsoap.org/soap/envelope/)
  1.2 (http://www.w3.org/2001/06/soap-envelope)
</faultstring></SOAP-ENV:Fault>

mm.. I guess it is OK because the service provider might be using some 
other SOAP version?

- When doing the same test on one of our services (SOAP::Lite 0.69 
installed on the service side). I got this error:

<soap:Fault><faultcode>soap:Client</faultcode><faultstring>Application 
failed during request deserialization:
xml declaration not at start of external entity at line 1, column 941, 
byte 941 at 
/usr/lib/perl5/site_perl/5.8.3/x86_64-linux-thread-multi/XML/Parser.pm 
line 187
</faultstring></soap:Fault>

- When executing a Java service:
  <soapenv:Fault>
   <faultcode>soapenv:Client</faultcode>
   <faultstring>org.jboss.axis.AxisFault: Version Mismatch</faultstring>
   <detail/>
  </soapenv:Fault>

and again, it might be ok?

- When trying to retrieve the wsdl of a 'moby-async' service (function 
retrieveService) I got this error:

Connection to MOBY Central at 
'http://mobycentral.icapture.ubc.ca/MOBY/Central'
died because:
        Can't use string ("<?xml version="1.0"?>
        <definiti") as a symbol ref while "strict refs" in use at 
/usr/lib/perl5
/site_perl/5.8.5/MOBY/Central.pm line 3290, <IN> line 371.

 ERROR ERROR ERROR

The function retrieveService does work when the service is not 
'moby-async' though.

- When executing an asynchronous service (function 
Async/Service/execute) everything seems ok.

Has anyone had similar problems? Any ideas?

Thank you!
Romina

Mark Wilkinson wrote:
> Hi all,
>
> First off, apologies to anyone who was trying to get through to MOBY  
> Central this afternoon - Eddie and I were making the changes necessary to  
> get it ready for SOAP::Lite 0.69 and the changes that worked perfectly on  
> the test server didn't work on the real MOBY Central... so it took a  
> couple of hours of troubleshooting before we had it back up again.
>
> The good news is that it seems to work correctly with both Perl and Java  
> clients.  What we don't know is if it is backwards compatible with earlier  
> versions of the MOBY::Client::* libraries, in particular,  
> MOBY::Client::Service.  I *think* it is, but we didn't have the  
> opportunity to test it.  In any case, the API did not change, so if you  
> just cvs update your client libraries everything should come back to life  
> if it isn't working.
>
> Upgrading to SOAP::Lite 0.69 was a nightmare!  The new SOAP::Lite behaves  
> subtly differently than the old one in several ways that we were not  
> expecting.  Thankfully, Pieter and Eddie had done most of the  
> trouble-shoting already, but it still threw us for a loop in some places  
> with incompatible SOAP version errors and a new auto-encoding of strings  
> by SOAP::Lite that we used to do in-code by ourselves (resulting in  
> double-coded messages).
>
> Anyway, it's done now.  Please scream loudly if it doesn't work for you!!   
> I have backups of everything so if necessary we can roll-back quickly.
>
> All of this was done primarily to get us to be compatible with the new  
> Asynchronous services API that use the Manchester WSRF::Lite modules which  
> depend on SOAP::Lite 0.69.  Can I ask the good folks at INB to test the  
> code to see if it works for them v.v. Async services?  The database can  
> now hold Category=moby-async (which reminds me, I need to update the  
> database templates in the CVS...), and Eddie assures me that the code will  
> now provide correct WSDL for an Asych interface.  The myGrid folks have  
> added moby-async as a valid service category into their ontology as well,  
> so in principle... it should all work... (touch wood!)
>
> *********************
> IMPORTANTLY this means that the new MOBY Central code will likely NOT work  
> on versions of SOAP::Lite prior to 0.69!!!  Keep this in mind if you are  
> running your own registry...
> *********************
>
> Tomorrow (if I have regained my sanity after todays hell ;-) ) I'll try to  
> make the changes to gbrowse_moby such that it can also invoke asynchronous  
> services.  I think the Async modules also need to be added to the MANIFEST  
> before they are installed by default, so I'll try to remember to do that  
> as well.
>
> Otherwise there were just some small changes to the test suite to better  
> clean-up the registry if something goes wrong during testing, and I  
> cleaned-up the test database so that it also doesn't cause the test suite  
> to fail.  I will restart the test registry using the new codebase first  
> thing in the morning so that the two are identical.
>
> We still haven't figured out why those namespace changes happened a couple  
> of days ago.  It coincided with the Gene Ontology curators making some  
> additions to the Namespace list, so I suspect that there is a bug in the  
> interface I have given to them (which talks directly to the database  
> without using the MOBY API), but exactly WHY this happened is still a  
> mystery.  I've alerted Midori to the problem and she's going to hold-off  
> on making any more changes until we can figure out what is going wrong.
>
> That's all the news from MOBY Central!
>
> Let us know ASAP if you notice anything wrong.  It's been a rough ride  
> today...
>
> Mark
>
>
>
> _______________________________________________
> MOBY-dev mailing list
> MOBY-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/moby-dev
>   

_______________________________________________
MOBY-dev mailing list
MOBY-dev at lists.open-bio.org
http://lists.open-bio.org/mailman/listinfo/moby-dev