If everything you ever want to do is a web service call, use BPEL. Otherwise,
JPDL.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4171515#4171515
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4171515
You probably have a race condition between the beginning of the async node
finishing up, and the end of the async firing. I wasn't clear on what's
causing the wait node to complete - can it be very quick? If so, try putting a
delay in there and see if it goes away.
This isn't a fix,
Petia,
If you have async=false on everything, all the work is performed sequentially
on one thread - there's no opportunity for a stale object (or race condition)
to occur.
-Ed
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4171123#4171123
Reply to the post :
Agreed - READ_COMMITTED seems required.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4161559#4161559
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4161559
___
jboss-user mailing
I'd think about using an event handler before/after the nodes that you are
interested in monitoring. It would write to a DB table of your own design.
This would give you a lot more flexibility - you can add whatever fields you
want, etc.
View the original post :
Is READ_COMMITTED sufficient isolation level for Oracle?
Yes. -Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4158739#4158739
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4158739
Can I do something except change of db?
Change from which database?
From Oracle? Why do you need to?
From HSQLDB? If you need concurrency, you must move off HSQLDB - JBPM relies
on the database for synchronization, so that it can be clustered. I'm sure
some other db's work too - you
Sorry, I haven't worked with the JBoss app-server... hopefully someone else can
help. Sorry! -Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4158345#4158345
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4158345
Pavel, look at hibernate config values for the following.
The ones with question-marks I can't help you with - I'm using local wrappers,
and can't take time to look up generic solutions. Check the Hibernate docs,
etc for these properties. I'm (obviously) on WebLogic - you'll need different
I would not use hsqldb with any application that has forking and asynchronous
activity, such as a node that blocks waiting for external input, or a node that
explicitly has async=true set.
You need a database that supports transactions and has isolation. I thought
Tom put something in the
Probably CMT. The question to ask is whether there is an outer contextual
transaction that you want the engine to participate in.
Usually, there's an MDB driving at least some activity, and usually you want
the MDB to roll back and retry on an engine exception. But your situation may
be
I don't do tasks, so I can't help much with specifics.
But the engine is supposed to set up the ExecutionContext so that everything is
correct for the given ActionHandler context - so this is expected.
As for HOW - I don't know - I'd have to research it. You're probably just
missing something.
ExecutionContext is used to provide context for ActionHandlers. It is best not
to think of it as a long-lived object - more as a face that JBPM exposes to
ActionHandlers and other bits of client code that it calls into.
It is at least token-specific - I don't THINK more than one can exist
See http://wiki.jboss.org/wiki/ForEachForkActionHandler.
-Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4153770#4153770
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4153770
Adding async=true to the node on each leg will make it execute independently.
I don't know how/if this works if you are not persisting; someone else may
know. If you are persisting, you'll need to use a database with good isolation
and transaction characteristics - in particular, not HSQL.
jaydub,
Can you say a little more about how/where you plugged in your own persistence
model?
Do you have multiple tokens, i.e., do you do any forking into nodes that are
either explicitly async='true', or that asynchronously block, waiting for an
external event to signal?
Thanks,
-Ed Staub
As well, you did not comment on the task instance example of the previous
post - how do you interpret that code?
Sorry, I don't remember much about the task support - I don't have the time to
research it.
View the original post :
Tina,
Why are you looking for alternatives?
Do you just not like Eclipse in general, or is there something specific about
the JPDL designer, or...
Thanks,
-Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4140983#4140983
Reply to the post :
In the specific case you talk about, a loop is probably a good design fit - all
chapters are similar, presumably.
In the similar case of a dynamic number of similar parallel activities, a fork
can be told to generate as many similar child tokens (like threads) as
necessary.
View the original
Ah, ok, so it's parallel, not sequential. Then use the dynamic fork. See
http://wiki.jboss.org/wiki/Wiki.jsp?page=ForEachForkActionHandler.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4141104#4141104
Reply to the post :
What database is in use?
If it's HSQLDB, I'm with ffernandez - don't do it.
You need a database that supports transaction isolation.
I think there's now a note to this effect in the user guide, but I can't find
it now.
View the original post :
Are you using at least READ_COMMITTED isolation in the DB?
If not, I think that before you finish, you'll find that you need to.
See http://jira.jboss.org/jira/browse/JBPM-983
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4131532#4131532
Reply to the post :
See
http://fisheye5.cenqua.com/browse/javaserverfaces-sources/jsf-ri/src/com/sun/faces/renderkit/RenderKitUtils.java?r1=1.34r2=1.35
I don't know anything about jsf, but the problem was a race condition between
threads, and was fixed in this diff. If it's easy to config down to a single
worker
Olivier,
You might consider a script.
The custom node seems hard to do while retaining the parameter evaluation
semantics for action nodes. Don't let me stop you, though!
I think Ronald assumed you needed the actions in parallel... I assumed you need
them sequential, or don't care.
I'm
anonymous wrote : I solved the problem by migrating to 3.2.2 version.
| It would be a good idea to replace the download link to 3.2.1 version on
the JBpm home page of labs.jboss.org ;)
I don't know if that's the final 3.2.2 - watch out.
See my note at
Adrian,
Very cool! Groovy, even ;-)
I've been noodling over how to make commands more extensible - this is
definitely the ticket.
-Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4085762#4085762
Reply to the post :
anonymous wrote : 1. I cannot understand is how can I tie a workflow to a
document. Would that be done in each process instance as a context variable
that I pass to the process instance.
Yes
anonymous wrote : 2. If I want my webapp to be generic enough to accommodate
any workflow definition
anonymous wrote : @petia... ... I'm not that 'hands-on' in 3.1.x anymore but it
might have been the new/current syntax did not work then.
Yes, that sounds right... I think I upgraded and the problem went away.
-Ed Staub
View the original post :
com.it.model.User must be serializable. Is it?
The code for it must be accessible in either the process deployment or in the
JBPM classpath. Is it?
-Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4084389#4084389
Reply to the post :
JobExecutor.start() creates a new instance of LockMonitorThread. But I can't
find where that thread is ever started.
Ditto.
-Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4084396#4084396
Reply to the post :
Are there any tests, examples, documentation, wiki entries, forum entries, code
comments, stickynotes, bathroom graffitti, or other sources of information on
how to use CompositeCommand? If not, can someone illuminate me?
I can't find a darned thing.
My best guess from reading the code is
See Oct 11, 2006 at 11:23 in thread
http://www.jboss.com/index.html?module=bbop=viewtopict=90146postdays=0postorder=ascstart=30.
He thought it was buggy and thus hard to reverse-engineer!
I've JIRA'd this; see http://jira.jboss.com/jira/browse/JBPM-1056
View the original post :
Ignore petia's example since that also uses a processvariable and just
another (imo less wanted) syntax.
That is, of course, unless it works! ;-)
I have a vague memory of running into the same problem and using Petia's
workaround early this year., when I was just starting with JBPM.
It's trying to instantiate a scheduler in order to kill off all the timers in
use by the process.
(I know, there probably aren't any timers in your case.)
It looks like the underlying problem is this:
anonymous wrote : java.lang.ClassNotFoundException: No ClassLoaders found for:
I said earlier that it was failing trying to instantiate the ServiceFactory.
That's not exactly right. It's failing in casting it to a ServiceFactory.
org.jbpm.scheduler.ejbtimer.EjbSchedulerServiceFactory is indeed a
ServiceFactory, the only thing I can think of is that ServiceFactory and
How are you observing that the old jbpmConfig is staying in memory?
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4083405#4083405
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4083405
In normal execution, when the subprocess completes, the parent process needs to
be signaled to continue.
You may want to use variables and a parent-process decision node after the
ProcessState.
-Ed Staub
View the original post :
Petia,
This shouldn't be necessary.
I'm not saying it ISN'T necessary in some context - but I can't imagine what it
is. It'll be interesting to see if it helps Arutha.
This is the kind of problem I'd debug into JBPM to figure out. However, be
warned - the variable-handling code is probably a
I'm not sure, but I don't think the servlet is required for JobExecutor - it
only provides a way to monitor, I think.
The only thing I'm missing is that I don't see where JobExecutor.start() is
called if the servlet isn't included.
View the original post :
A wild guess might somehow calling JbpmConfiguration.close() help? That
would free up the JobExecutor threads, servicefactories, etc.
I don't know enough about WebLogic redeployment to know whether it would make a
difference.
-Ed Staub
View the original post :
You can't leave a node from the node-enter event.
Instead, use a instead of , and put the actionhandler in the body of the
node.
You probably want to figure out how to use a Decision node without a
ClassCastException, though - you don't want to write custom actionhandlers for
every decision!
What database is in use?
What transaction isolation is in place?
What version of JBPM are you using?
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4080056#4080056
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4080056
The state of a process is not saved until there is nothing left to do, and the
JbpmContext is closed. This typically happens under 3 conditions:
- encountering a node marked async
- completion of the process
- the process is completely blocked, waiting for external stimuli to resume.
These
harpritt,
Couldn't you copy the value into a process variable, using a task controller?
See http://docs.jboss.com/jbpm/v3/userguide_single/#taskcontrollers
-Ed Staub[/url]
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4079627#4079627
Reply to the post :
More log info please... a stack trace, and a real exception message.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4076851#4076851
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4076851
I don't have any experience in this area, but I see very little useful relevant
information outside of the user guide and the source code. Maybe somebody with
more experience will respond...
I think the one tricky thing is that you need to immediately terminate the
process when anyone DENIES.
By session scoped, do you mean token-scoped by adding variables to the
process definition, or are you referring to something outside of JBPM?
If you mean token-scoped, AFAIK this is the standard way to pass data from
node to node. I'm curious - why don't you like it?
-Ed Staub
View the
why do you need, or do you think you need this low level access to the
processdefinition? The api makes it possible to retrieve this all.
I think Naveen was just trying to understand how it works. I remember being
equally mystified at the same thing - it works, so the actionhandler class
They're in the database.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4074867#4074867
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4074867
___
jboss-user mailing list
See my comment on the JIRA. -Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4074974#4074974
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4074974
___
jboss-user mailing
stana,
If it's a JBPM issue, file a JIRA issue, after first making sure there isn't
one already. If there already is one, you'll want to vote for it.
If you understand what's going on, and can describe it clearly, it's likely to
get quicker attention. If possible, you want to aim for someone
token.setNode();
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4073317#4073317
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4073317
___
jboss-user mailing list
You need to say more about what data, if any, will be shared across servers.
In general, if you use read-committed isolation with transactions, I think
you'll be ok.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4073318#4073318
Reply to the post :
galvino,
In case it isn't obvious ... you should be reading up on Hibernate. The online
documentation is good, and so is the book.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4073320#4073320
Reply to the post :
How did the subprocess end?
You probably want to read ProcessState.java (for the parent's perspective) and
ProcessInstance.end() (for the child's perspective).
I'm pretty sure you've made a mistake somewhere... maybe grabbed the wrong
token at some point.
View the original post :
I don't have time to dig into it, but I'd guess that you're using too low an
object to do the delete. I'd try to find a process instance delete method at a
higher level that deletes swimlanes first.
This is just a guess!
-Ed Staub
View the original post :
After digging more, I'm starting to wonder if you might have a bug.
If you're not running on HSQL, I'd wonder if it's db-dependent. Also, what
version of JBPM?
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4073325#4073325
Reply to the post :
All child tokens, when they're done, should/could transition to a join. The
join will count down the outstanding tokens for the parent token, and allow it
to resume when the last child completes.
To avoid other potential problems, I suggest that you make your
token-generating node a fork.
1) I sure steps executed are //--1, //--2, //--3, //--4 because I executed
them in debug mode.
Sorry, I misread part of your abstract.
It looks to me like the intent is to transfer ownership (responsibility for
closing) of the Connection to the Session. So the Service is responsible for
I don't have much.
I think you misread DbPersistenceService.getSession().
Reading the first branch in DbPersistenceService.getSession(), it looks like an
implementation of session-sharing - the first DbPersistenceService to get a
session is responsible for closing it, and others just reuse it.
O.M.,
You're now into appserver-dependent stuff.
What appserver are you on?
You mentioned WLI - WebLogic, perhaps?
My wondering is about where that ejb is defined and/or deployed during the
processus of deploying the jbpm-enterprise.ear ?
C'mon, I know you know this stuff!
web.xml?
O.M.,
You're going to need to start looking at the source code to resolve this stuff.
Take a look at where this is coming from:
org.jbpm.scheduler.ejbtimer.EjbSchedulerService.(EjbSchedulerService.java:34)
and you'll get a really good idea of what's going on now.
Since you're now
O.M.,
Some of us read EVERY POST to try to help as many folks as we can.
Unfortunately, there aren't very many folks who try to answer queries.
So you should try to make their (unpaid, volunteer) job easier, not harder by
cluttering up the forums with redundant queries.
you don't have to
Well, looking at the exception,
anonymous wrote : java.lang.ClassCastException:
org.jbpm.scheduler.ejbtimer.EjbSchedulerServiceFactory
| org.jbpm.svc.Services.getServiceFactory(Services.java:122)
|
then looking at the source it points at...
| public ServiceFactory
JBPM deployed more than once, e.g.:
- in the WAR
and also in the EAR (if there is one)
and also in the appserver classpath.
I'm really guessing here - if I were in your situation, I'd drop some logging
into getServiceFactory to try to find out what's going on.
-Ed Staub
View the original post
The only reasons the hibernate.cfg.xml file wouldn't be found is if there's
another one on your classpath, or if jbpm.cfg.xml is pointing at another file.
How do you know it's not being found?
Go to your local bookstore and read the new Hibernate book chapter on
configuration. Or browse
It looks like you're trying to use JTA transactions (specified in
hibernate.cfg.xml) with a JDBC persistenceServiceFactory (specified in
jbpm.cfg.xml).
Once again, read the Hibernate book on configuration.
View the original post :
See jbpm.cfg.xml. It points at the JBPM hibernate config file. Rename the
file, and change the pointer.
-Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4069156#4069156
Reply to the post :
I would guess that you can't access ANYTHING in the database via JBPM. Is that
possibly true?
If so, it looks like some kind of transaction configuration problem, probably
in hibernate.cfg.xml.
I'm guessing from the error messages that you're using JTA. Just as a test,
you might want to
Ronald,
Thanks, stay cool, have fun!
-Ed
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4067826#4067826
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4067826
___
jboss-user
This is a short question with a really LOOONG answer - several pages long.
The answer depends on the database you're using, what transaction isolation
you've specified, and, most importantly, the specific concurrency scenario
you're thinking of.
To see what I mean about the scenario, consider
efip10,
Sorry, I used too broad a brush.
At least in some cases, the behavior I described is dependent on the isolation
level selected. If at least level 2 (read-committed) is selected, then I
think it works as I described, if the database supports it. I believe that
row-locking is used
efip10,
On thinking about your problem some more...
Are you using JTA transactions?
AFAIK, an MDB message sent as part of a JTA transaction will not actually be
sent until the transaction is committed. In this case, if the MDB were created
as part of the transaction that's saving the
Ricardo,
I'm no expert, but I'd be looking at your database and transaction
configurations. What database are you using? It looks like you're running in
Tomcat - yes?
At the point that you're failing, you're probably opening a Job session for the
first time. Jobs are interesting because
Koen,
See also
http://www.jboss.com/index.html?module=bbop=viewtopicp=4008324#4008324
It's the same thing, I think. Unfortunately, no JIRA or unit test there,
either.
-Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4064497#4064497
Reply to the post
anonymous wrote : Bit a pitty it is so easily overlooked if you don't define an
exception handler in your process.
The log message certainly doesn't jump out and bite you!
Do you perhaps want to JIRA it?
-Ed Staub
View the original post :
kukeltje wrote : Ed,
|
| Will putting an async=true on the node realy force a commit? That has to be
documented well then (rollback
| 's further on in the process a rollback will not be possible then) This is
not a problem if the process has to wait for the return value of the async
What's the question? Is it
Where is the rollback?
If so, all I can offer is the usual advice - log everything.
If that doesn't lead anywhere, you could start inserting getRollbackOnly()
calls to narrow it down.
If you're wondering how to keep an application rollback from rolling back
I don't know anything about the Sybase driver.
But I think it's very possible that setRollbackOnly() was set much earlier, and
the log message is only reflecting it much later at commit time.
The 4,033ms executing PreparedStatement: SELECT... log at the end looks
strange in the context of
I'm not sure, but I think that if you mark your message-firing node as
'asynch=true', this will force a commit and take care of the problem.
If you happen to be firing the message from the node-entry event, this won't
work; you'll need to mark a prior node, perhaps introducing a dummy one just
[EMAIL PROTECTED] wrote : Performing a signal on the current token from an
action handler is not a good practice.
Koen,
You mean that to apply only to event-handling ActionHandlers, correct?
Node-execution ActionHandlers can signal, yes?
Thanks,
-Ed Staub
View the original post :
kukeltje wrote : multiple end states are not supported afaik
Ronald,
Are you sure? Why?
fewagewasd,
event type=before-signal
| action name=check foo bar class=TestActionHandler/action
| /event
|
You shouldn't call signal or leave-node from an event-handler - do
The Graphical Process Designer (GPD) is an eclipse plugin and is in the starter
kit. If the docs you found didn't make this clear, you aren't reading the
right stuff (or you aren't reading the stuff right).
-Ed Staub
View the original post :
You need to help yourself some before you start asking other folks for help.
Search the forum - there's lots of info.
I do not see your earlier post - I guess it must have been on a different
forum, or under a different alias.
-Ed Staub
View the original post :
Oh, I see now... you hijacked an earlier thread...
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4061251#4061251
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4061251
___
See http://jira.jboss.com/jira/browse/JBPM-983, and vote for it if relevant -
it probably is.
Are you using the JobExecutor?
More specifics on your workflow would be helpful, if it's at all more
complicated than you've stated.
It may help your particular case to mark a node on each fork with
See http://jira.jboss.com/jira/browse/JBPM-983 , and vote for it if relevant -
it probably is.
Are you using the JobExecutor?
It may help your particular case to mark a node on each fork with
asynchronous=exclusive. This helps in some simple cases. But it really just
changes the odds in a
Ronald,
Examples often do 'wrong' things..
I think you're suggesting that this isn't a real use case. I think I see your
point -- without any asynchronous behavior, the two branches are serialized and
so might as well be modeled as a straight line.
But imagine an actionhandler that may
By sibling do you mean fire and forget - that is, the first process starts
the second and then doesn't wait for it or communicate with it again?
If so, then:
- it's not modeled at all, but,
- it's really easy to write an ActionHandler that will do this. If you want,
the name of the
kukeltje wrote : weblogic8 does not officially support the ejb3 spec (afaik)
so that might be the reaseon if you use the enterprise version.
|
| Look at WLS 10 that one is compliant
Ronald,
I'm confused... is the JBPM enterprise support dependent on ejb3?
That's surprising, especially
In case anyone has this problem again and is searching the forum for clues...
see http://jira.jboss.com/jira/browse/JBPM-983
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4058884#4058884
Reply to the post :
See (lengthy) update on http://jira.jboss.com/jira/browse/JBPM-983
-Ed Staub
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4058347#4058347
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4058347
fady.matar wrote:
anonymous wrote : It is not recommended to use two JNDI for the same physical
database.
I'm no expert here, but I don't think this is exactly correct. Accessing
different databases (schemas) in the same physical database using different
JNDI names is perfectly fine, except
Ronald,
Re JMS... can you explain how you think it will solve these problems? If you
mean that there would be only one bean in the pool, that's not solving it at
all - that's single-threading all requests.
As I see it, the problem looks like one of token synchronization, and I don't
see
Alex,
Apart from Ronald's request for scriptlessness, it doesn't look like anyone's
dug into your http://jira.jboss.com/jira/browse/JBPM-983. Do you still need a
fix? If so, I'll try to go after it in the next few days.
-Ed Staub
View the original post :
Fork (as in ForEachForkActionhandler) is what you want.
There's nothing preventing multiple tokens for the same way of execution.
The fact that they CAN be different doesn't mean they MUST be different.
You wait for all of them to complete by having them transition to a join node
(in the
The one-hour difference from UTC/GMT/Zulu could be due to a daylight savings
time confusion somewhere.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4053966#4053966
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4053966
anonymous wrote : From what i have get until now, please correct me if I am
wrong, at process design level, we can't declare variable types, only string
variables, so... in some cases those strings must be cast\converter to other
formats on the java code (actions).
|
No. Please read the
... to be more explicit - you can't declare variable types, but the only
string variables is way off.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4053571#4053571
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4053571
Jon,
Threads don't share hibernate caches on the same process instance. If another
thread looks at the token, it's going to have fetched it independently from the
database.
The database is updated when the token reaches a wait state.
I'm still trying to understand what it means that the
1 - 100 of 262 matches
Mail list logo
