Hi asankha, the initial problem is definitely the NIO transports - I switched the transport sender and receiver back to the ones that were used before Synapse support NIO and the connection timeout problem went away.
I'm now having a problem where the client is sent a 202 accepted response immediately from Synapse, even though I have a clone mediator followed by an aggregator in my Synapse configuration. Once the cloned tasks come back and are aggregated, the client has long gone due to getting an immediate 202 response. The operating system is: SunOS 5.10 The hardware is: Sun V240 If you have some simple test Synapse configs/jars you want me to run on this hardware/OS that can help diagnose this problem or prove that it actually works (and my setup was wrong) then I'm happy to try them out. Note that we've also got boxes there running SunOS 5.8 which I can try this out on too. cheers, Ants. On Tue, Dec 2, 2008 at 6:10 PM, Asankha C. Perera <[EMAIL PROTECTED]> wrote: > Anthony > > We have seen an issue with the NIO transport on the older Solaris machines > about an year ago (http://markmail.org/message/sc2ilnlfqlswoh7k) and thats > fixed now... Could you let us know the Solaris version, and hardware used in > your environment? > > asankha >> >> I made some progress with this - It seemed to be something in the >> axis2.xml for Synapse 1.2 that didn't work on Solaris - so I grabbed >> an older Synapse 1.1.1 axis2.xml and used that and it worked. I will >> do a diff later to see what the difference between these two is. >> However, I'm now having another issue that may be to do with my >> synapse.xml: >> >> The messages get sent back and aggregated, however the client gets an >> HTTP 202 response from Synapse immediately, apparently ignoring the >> cloning and subsequent aggregation sequences I have set up. Is this >> an issue with the clone mediator? I have tried both continueParent >> set to true and false, but it still returns early in both cases. >> >> Anyone had experience in this situation? >> >> ============================== >> Hi, I'm using Synapse 1.2 to take an input message, do some mediation >> on it, then clone it so it can be sent to two different web services >> and then I'm aggregating the results with a custom aggregate mediator >> I have written (as the aggregation has business rules in it, so I >> can't use the default Synapse aggregator). >> >> This works perfectly on Windows using Synapse 1.2 stand-alone. When >> using the same setup on a Solaris box, the cloned messages time out >> when trying to on-send the message to the end points. These >> web-services are running in a Tomcat on the same machine as axis 2 web >> services. Using localhost and the actual server domain names both >> fail. Moving these web services to a different machine on the network >> fail also, as does going via an external www URL to those services. >> Every single message times out. >> >> Initially I thought that this would be a DNS problem or something - >> however setting up a plain Synapse proxy in the same Synapse instance >> runs fine, and it makes contact to the axis 2 web services. >> >> I have debugged the Synapse 1.2 source and it goes all the way through >> and sends off the message fine - its just it never gets a response. >> >> Has anyone had issues on Solaris (this is an older version) or similar >> behaviour where messages timeout every time? >> >> My synapse.xml is as follows: >> >> <definitions xmlns="http://ws.apache.org/ns/synapse";> >> >> <!-- datahub.in.sequence takes a request, sends it to VIS and VBATS and >> aggregates the results. --> >> <sequence name="main" trace="enable"> >> <log level="full"/> >> <in> >> <sequence key="topicregister.getvehicleidentifier.sequence"/> >> </in> >> <out> >> <class >> name="nz.co.toyota.mediators.datahub.GetVehicleResponseAggregator"/> >> <send/> >> </out> >> </sequence> >> >> <!-- datahub.vis.sequence takes a request, sends it to the VIS >> datahub vehicle >> service, and returns the result, as-is. --> >> <sequence name="datahub.vis.getvehicle.sequence"> >> <in> >> <!-- Send all incoming traffic to the VIS datahub vehicle >> service --> >> <property name="source-system" value="vis"/> >> <send> >> <endpoint> >> <address >> uri="http://tnz017:8071/tnz/services/DataHubVehicleService-1.2"/> >> </endpoint> >> </send> >> </in> >> </sequence> >> >> <!-- datahub.vbats.getvehicle.sequence takes a request, sends it to the >> VBATS datahub vehicle service, and returns the result, as-is. --> >> <sequence name="datahub.vbats.getvehicle.sequence"> >> <in> >> <!-- Send all incoming traffic to the VBATS datahub >> vehicle service --> >> <property name="source-system" value="vbats"/> >> <send> >> <endpoint> >> <address >> uri="http://tnz017:8071/tnz/services/DataHubUsedVehicleService-1.0"/> >> </endpoint> >> </send> >> </in> >> </sequence> >> >> <!-- topicregister.getvehicleidentifier.sequence takes a vehicle >> request and looks up its >> vin in the topic-register. --> >> <sequence name="topicregister.getvehicleidentifier.sequence"> >> <in> >> <!-- parameters for the vehicle identifier lookup call to >> the topic register --> >> <property name="axis2-repository-path" >> value="C:/Java/Projects/data-hub/data-hub-vehicle-service-test-client"/> >> <property name="topic-register-service-url" >> value="http://tnz017:8071/tnz/services/TopicRegisterService-1.1"/> >> <property name="username" value="*****"/> >> <property name="password" value="*****"/> >> <class >> >> name="nz.co.toyota.mediators.datahub.TopicRegisterGetVehicleIdentifierMediator"> >> </class> >> <clone continueParent="false"> >> <target sequence="datahub.vbats.getvehicle.sequence"/> >> <target sequence="datahub.vis.getvehicle.sequence"/> >> </clone> >> </in> >> </sequence> >> >> </definitions> >> >> > > > -- > Asankha C. Perera > http://adroitlogic.org > > http://esbmagic.blogspot.com > >
