Sorry , here are the attachments .. :)-
Ophir wrote: > > Hi , > Im attaching an example process which invokes 4 times in Sequence an > invoke-receive activities > note that only one is actualy executed . ( because of the fast response .. > , if i delay the service response , then all activities are completed ). > Another failed scenario is the Split where there are more then one > RECEIVEs on the same service and the > response messages are handled at the same time. > (looks like a transactions problem ). > * sequence.zip -> Contains the fast asynch responses failure.. pkg and > logs > * split.zip -> Contains concurrent receives failure..pkg and logs > > I have tried to debug the split and found out that the instance of > org.apache.ode.axis2.ODEService > in onAxisMessageExchange , when the two thread are trying to commit , > only one succeeds.. > > Thanks , > Ophir. > > > > > > > > > > > > > Rafal Rusin wrote: >> >> Can you provide zipped logs with DEBUG level category on all classes? >> It may not happen when you turn on debug, so then could you give INFO >> category logs? >> >> Regards, >> -- >> Rafał Rusin >> www.mimuw.edu.pl/~rrusin >> >> >> 2009/1/25 Ophir <[email protected]>: >>> >>> Hi , >>> We work with ode1.2 on tomcat and mysql db. >>> I am attaching a sample process which does not work >>> for splitting - multi-receives on the same time.. >>> if the receives activities are no concurrent called , they work fine and >>> the >>> process >>> completes ( finishes the splitting ). >>> BTW , another wierd behaviour we have is when the invoke-recieve >>> (asynch) is >>> a fast >>> call , means that if the time between the invoke and its receive is less >>> then ~100ms , then >>> it will not work as well. >>> As i mentioned , these processes are quite simple , but they fail on >>> short >>> response and concurrent responses >>> on SAME service. >>> >>> Please, if you spot anything wrong . let me know.. >>> (For now, we limit our customers for only sequence flow - with no >>> parallel >>> actions ) >>> >>> Thanks, >>> Ophir. >>> >>> >>> mriou wrote: >>>> >>>> On Thu, Jan 22, 2009 at 5:08 AM, Ophir <[email protected]> wrote: >>>> >>>>> >>>>> Hi , >>>>> It looks like the ODE has a problem to handle a more then one >>>>> call on Receive on the same partner link. >>>>> In case there is a split activity (flow) that each branch receive on >>>>> the >>>>> same service , On concurrent >>>>> invokes , only one is handled. the other one is ignored and therefore >>>>> the >>>>> process is stuck. >>>>> >>>>> This issue happens 90% of the cases. >>>>> In order to simulte this , just have a flow-activity with invoke one >>>>> way >>>>> and >>>>> receive on both branches(paths). >>>>> ( when the receives are called on diffrenert times , there is no >>>>> problem). >>>>> -- this issue actually prevents using the Flow(split) activity in the >>>>> process . >>>>> >>>>> Can somebody check this scenario, >>>> >>>> >>>> On which branch/release do you have this problem? I'm pretty sure this >>>> works >>>> on 1.x, those type of patterns are pretty common. Also would you happen >>>> to >>>> have a test process that reproduces it? The devil is often in the >>>> details. >>>> >>>> Thanks, >>>> Matthieu >>>> >>>> >>>>> >>>>> Thanks, >>>>> Ophir. >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> View this message in context: >>>>> http://www.nabble.com/concurrent-Receives-problem-tp21603812p21603812.html >>>>> Sent from the Apache Ode User mailing list archive at Nabble.com. >>>>> >>>>> >>>> >>>> >>> http://www.nabble.com/file/p21654137/gaz.zip gaz.zip >>> -- >>> View this message in context: >>> http://www.nabble.com/concurrent-Receives-problem-tp21603812p21654137.html >>> Sent from the Apache Ode User mailing list archive at Nabble.com. >>> >>> >> >> > > http://www.nabble.com/file/p21664933/split.zip split.zip http://www.nabble.com/file/p21664933/sequence.zip sequence.zip -- View this message in context: http://www.nabble.com/concurrent-Receives-problem-tp21603812p21664933.html Sent from the Apache Ode User mailing list archive at Nabble.com.
