[
https://issues.apache.org/jira/browse/ODE-327?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12612419#action_12612419
]
rrusin edited comment on ODE-327 at 7/10/08 2:46 AM:
----------------------------------------------------------
OK, so I suggest a following solution:
Create a new Job (eg. MessageMatcherJob), wchich will check if there is stored
Message and Receive for a given correlation. And if there are both of them in
DB, it will continue process with them matched.
MessageMatcherJob will be queued to Scheduler on storing in DB (message in
transport layer or receive in bpel engine). So one of those MessageMatcher jobs
will be executed after the two parallel transactions from previous error case
will be committed. So it will see in DB this situation and will continue
process.
The other cases will be handled as currently and won't be broken.
What do you think about it?
Regards
was (Author: rrusin):
OK, so I suggest a following solution:
Create a new Job (eg. MessageMatcherJob), wchich will check if there is stored
Message and Receive for a given correlation. And if there are both of them in
DB, it will continue process with them matched.
MessageMatcherJob will be queued to Scheduler on storing message in DB both in
communication layer and bpel engine. So one of those jobs will be executed
after the two parallel transactions from previous error case will be committed.
So it will see in DB this situation and will continue process.
What do you think about it?
Regards
> Process instance hangs in state ACTIVE after processing concurrently RECEIVE
> in BPEL Engine and incomming message in JBI transport layer
> ----------------------------------------------------------------------------------------------------------------------------------------
>
> Key: ODE-327
> URL: https://issues.apache.org/jira/browse/ODE-327
> Project: ODE
> Issue Type: Bug
> Components: BPEL Runtime
> Affects Versions: 1.2
> Environment: ServiceMix, Oracle EXTERNAL, Commons DBCP Managed
> Connection Pool
> Reporter: Rafal Rusin
>
> The scenario is following:
> Process:
> 1. Declare correlation C with ID
> 2. Set some ID to message M
> 3. INVOKE Partner (initiate correlation C on sending)
> 4. RECEIVE PartnerCallback M2 (with correlation C set to initiate = NO)
> Partner Process (external):
> 1. RECEIVE M3
> 2. INVOKE PartnerCallback M3
> The problem is when the message from Partner Process arrives concurrently
> while processing RECEIVE in 4.
> The two transactions seem to read state in DB (there's no RECEIVE registered
> for incoming message; and no message stored in DB for entered RECEIVE). So
> those two register in DB and there's no more action.
> Here's a detailed log (cut some lines) of this situation:
> 11:50:28,984 | DEBUG | pool-19-thread-13 | BpelProcess |
> l.engine.PartnerLinkMyRoleImpl 127 | INPUTMSG: 42.deliveryReportEvent: MSG
> RCVD keys=[{CorrelationKey setId=236, values=[168244]}] mySessionId=null
> partnerSessionId=null
> 11:50:28,987 | DEBUG | pool-19-thread-13 | BpelProcess |
> l.engine.PartnerLinkMyRoleImpl 236 | INPUTMSG: 42.deliveryReportEvent:
> SAVING to DB (no match)
>
> 11:50:29,104 | DEBUG | pool-19-thread-20 | ScopeFrame |
> he.ode.bpel.runtime.ScopeFrame 263 | Write variable: name={Variable
> __PROCESS_SCOPE:PROCESS:OXsdTyp
> eVarType#35} value=<?xml version="1.0" encoding="UTF-8"?>
> <xsd-complex-type-wrapper>
> <status
> xmlns="http://touk.pl/schema/monitoringVariables";
> xmlns:monitoringVariables="http://touk.pl/schema/monitoringVariables";>SENT_CORRECTLY</status>
> <monitoringVariables:changeDate
> xmlns:monitoringVariables="http://touk.pl/schema/monitoringVariables";>2008-07-09T11:50:28+01:00</monitoringVariables:chang
> eDate>
> </xsd-complex-type-wrapper>
> 11:50:29,110 | DEBUG | pool-19-thread-20 | BpelProcess |
> .engine.BpelRuntimeContextImpl 309 | SELECT: PickResponseChannel#95: USING
> CORRELATOR 42.deliveryReportEvent
> 11:50:29,110 | INFO | pool-19-thread-20 | Query |
> log.Log4JLogFactory$LogAdapter 80 | This query on type "class
> org.apache.ode.dao.jpa.CorrelatorDAOImpl" must l
> oad the entire candidate class extent and evaluate the query in-memory. This
> may be very slow. The query must be executed in memory because OpenJPA is
> configured with IgnoreCache=f
> alse and FlushBeforeQueries=false and there are dirty instances that may
> affect the query's outcome in the cache.
> 11:50:29,117 | DEBUG | pool-19-thread-20 | BpelProcess |
> .engine.BpelRuntimeContextImpl 388 | SELECT: PickResponseChannel#95: ADDED
> ROUTE 42.deliveryReportEvent: {Corre
> lationKey setId=236, values=[168244]} --> 168244
> 11:50:29,118 | DEBUG | pool-19-thread-20 | BpelRuntimeContextImpl |
> .engine.BpelRuntimeContextImpl 839 | Setting execution state on instance
> 168244
> 11:50:29,151 | DEBUG | pool-19-thread-25 | InstanceLockManager |
> pel.engine.InstanceLockManager 50 | Thread[pool-19-thread-25,5,main]:
> lock(iid=168244, time=1MICROSECONDS)
> 11:50:29,152 | DEBUG | pool-19-thread-25 | InstanceLockManager |
> pel.engine.InstanceLockManager 63 | Thread[pool-19-thread-25,5,main]:
> lock(iid=168244, time=1MICROSECONDS)-->G
> RANTED
> 11:50:29,155 | DEBUG | pool-19-thread-25 | BpelProcess |
> he.ode.bpel.engine.BpelProcess 360 | >>
> handleWorkEvent(jobData={type=MATCHER, inmem=false, ckey=236~168244, co
> rrelatorId=42.deliveryReportEvent, iid=168244})
> 11:50:29,162 | DEBUG | pool-19-thread-25 | BpelRuntimeContextImpl |
> .engine.BpelRuntimeContextImpl 161 | BpelRuntimeContextImpl created for
> instance 168244. INDEXED STATE={{OScope
> 'SendPROCESSScope' id=229}::42=[SCOPE(...)], {OScope 'SendPROCESSScope'
> id=229}::41=[ACTIVITYGUARD(...)],
> OFlow#98-ProcessMessages::15=[ACTIVITYGUARD(...)],
> OSequence#100-SendPROCESS::19=[ACTIV
> ITYGUARD(...)], {OScope '__PROCESS_SCOPE:PROCESS' id=20}::0=[ACTIVE(...)],
> {OScope 'ProcessScope' id=99}::18=[SCOPE(...)], {OScope 'ProcessScope'
> id=229}::42=[ACTIVE(...)], OSequen
> ce#47-PROCESSAlert::1=[ACTIVITYGUARD(...)],
> OPickReceive#290-receive-activity-line-378::64=[PICK(...)],
> OSequence#237-SendPROCESS::44=[SEQUENCE(self=(OSequence#237-SendPROCESS,TerminationChannel
> #61,ParentScopeChannel#63), [EMAIL PROTECTED],
> remaining=[OPickReceive#290-receive-activity-line-378,
> OSwitch#293-if-activity-line-383])], OSequen
> ce#47-PROCESSAlert::2=[SEQUENCE(self=(OSequence#47-PROCESSAlert,TerminationChannel#3,ParentScopeChannel#5),
> [EMAIL PROTECTED], remaining=[OFlow#98-Proces
> sMessages])],
> OPickReceive#290-receive-activity-line-378::63=[ACTIVITYGUARD(...)], {OScope
> 'SendPROCESSScope' id=99}::18=[ACTIVE(...)], {OScope 'SendPROCESSScope'
> id=99}::17=[ACTIVITYGUARD(.
> ..)],
> OSequence#100-SendPROCESS::20=[SEQUENCE(self=(OSequence#100-SendPROCESS,TerminationChannel#29,ParentScopeChannel#31),
> [EMAIL PROTECTED], remaining=[
> OSwitch#227-PROCESSNotEmpty, OThrow#333-Completed])],
> OSequence#237-SendPROCESS::43=[ACTIVITYGUARD(...)],
> OSwitch#227-PROCESSNotEmpty::39=[ACTIVITYGUARD(...)],
> OFlow#98-ProcessMessages::16=[<T:A
> ct:Flow:ProcessMessages>], {OScope '__PROCESS_SCOPE:PROCESS_Alert'
> id=20}::0=[SCOPE(...)]}
> 11:50:29,163 | DEBUG | pool-19-thread-25 | BpelProcess |
> he.ode.bpel.engine.BpelProcess 406 | Matcher event for iid 168244
> 11:50:29,164 | DEBUG | pool-19-thread-25 | BpelRuntimeContextImpl |
> .engine.BpelRuntimeContextImpl 1316 | MatcherEvent handling:
> correlatorId=42.deliveryReportEvent, ckey={Correlat
> ionKey setId=236, values=[168244]}
> 11:50:29,152 | DEBUG | pool-19-thread-24 | rRoleMessageExchangeImpl |
> PartnerRoleMessageExchangeImpl 63 | replyAsync mex=hqejbhcnphr3f2nurln3kn
> 11:50:29,168 | DEBUG | pool-19-thread-24 | BpelRuntimeContextImpl |
> .engine.BpelRuntimeContextImpl 839 | Setting execution state on instance
> 168246
> 11:50:28,993 | DEBUG | pool-19-thread-16 | ScopeFrame |
> he.ode.bpel.runtime.ScopeFrame 237 | Initialize variable: name={Variable
> __PROCESS_SCOPE:UnifyFMS.dispatcherPlk
> VarResponse:OMessageVarType#22} value=<?xml version="1.0" encoding="UTF-8"?>
> <message><parameters><sendResponse
> xmlns="http://.../queue"/></parameters></message>
> 11:50:29,460 | DEBUG | pool-19-thread-25 | BpelRuntimeContextImpl |
> .engine.BpelRuntimeContextImpl 1348 | MatcherEvent handling: nothing to do,
> no matching message in DB
> 11:50:29,461 | DEBUG | pool-19-thread-25 | InstanceLockManager |
> pel.engine.InstanceLockManager 88 | Thread[pool-19-thread-25,5,main]:
> unlock(iid=168244)
>
> And entry from ODE_MESSAGE_EXCHANGE table (indicating that MEX was committed
> to DB, however line 11:50:29,460 says it's not found):
> hqejbhcnphr3f2nurln3kj {http://.../DeliveryReport2 (null)
> ID:127.0.1.1-11b074d2328-2:3461 236~168244 QUEUED (null) 77
> (CLOB) (null) (null) deliveryReportEvent 42 REQUEST_ONLY (null)
> (null) 0 ASYNC 174 (null) 104 (null) 177888 (null)
> Regards
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
