You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@ode.apache.org by kodeninja <ko...@gmail.com> on 2009/07/10 09:14:56 UTC
Re: How does ODE's BPEL Engine recover in the event of system
failure?
Thanks for the reply, Matthieu.
I have a BPEL that performs a CRUD operation on an entity and responds back
to the calling client with a success/failure message accordingly. The client
starts the BPEL by sending a CRUD request. It provides a callback handler to
the BPEL to intimate it back when it's done. Now, if the BPEL engine crashes
in the middle of the process, what I see is that the client is never updated
with the actual status of the request (CRUD failed), whenever the engine is
re-started. So, in this case, where there's a callback to the client
involved, is it possible for the BPEL process to just proceed from the point
where it was, when the engine crashed, and complete the request?
Thanks,
-Kodeninja
mriou wrote:
>
> On Wed, Jul 8, 2009 at 12:17 AM, kodeninja <ko...@gmail.com>
> wrote:
>
>>
>> Hi all,
>>
>> I want to find out about the recovery capabilities built into ODE. What
>> kind
>> of recovery does ODE provide in case if ODE itself crashes, or if a
>> particular BPEL process instance within ODE crashes? Does it restart the
>> failed process automatically? And does it do so from the beginning or
>> from
>> the point that the instance crashed?
>
>
> If you stick with one-ways, yes. For two-way messages, the client
> connection
> is usually lost so it's a bit more messy.
>
>
>>
>>
>> Also, in case the process engine itself crashes and is restarted, does it
>> restart all the failed process instances automatically? Or is there a way
>> to
>> do so manually?
>
>
> All transactions are going to be discarded and the process will be back in
> the state it was before. Instances aren't failed in most cases, though.
> There's also the special case of a crash just during a two-way invoke, for
> which we'll force the invoke to enter activity recovery, which you can
> then
> retry manually.
>
> Matthieu
>
>
>>
>>
>> -Thanks
>> Kodeninja
>> --
>> View this message in context:
>> http://www.nabble.com/How-does-ODE%27s-BPEL-Engine-recover-in-the-event-of-system-failure--tp24386617p24386617.html
>> Sent from the Apache Ode User mailing list archive at Nabble.com.
>>
>>
>
>
--
View this message in context: http://www.nabble.com/How-does-ODE%27s-BPEL-Engine-recover-in-the-event-of-system-failure--tp24386617p24422793.html
Sent from the Apache Ode User mailing list archive at Nabble.com.
Re: How does ODE's BPEL Engine recover in the event of system failure?
Posted by Lars Ericsson <la...@yahoo.com>.
>is it possible for the BPEL process to just proceed from the point
>where it was, when the engine crashed, and complete the request?
No it is not possible.
________________________________
From: kodeninja <ko...@gmail.com>
To: user@ode.apache.org
Sent: Friday, July 10, 2009 9:14:56 AM
Subject: Re: How does ODE's BPEL Engine recover in the event of system failure?
Thanks for the reply, Matthieu.
I have a BPEL that performs a CRUD operation on an entity and responds back
to the calling client with a success/failure message accordingly. The client
starts the BPEL by sending a CRUD request. It provides a callback handler to
the BPEL to intimate it back when it's done. Now, if the BPEL engine crashes
in the middle of the process, what I see is that the client is never updated
with the actual status of the request (CRUD failed), whenever the engine is
re-started. So, in this case, where there's a callback to the client
involved, is it possible for the BPEL process to just proceed from the point
where it was, when the engine crashed, and complete the request?
Thanks,
-Kodeninja
mriou wrote:
>
> On Wed, Jul 8, 2009 at 12:17 AM, kodeninja <ko...@gmail.com>
> wrote:
>
>>
>> Hi all,
>>
>> I want to find out about the recovery capabilities built into ODE. What
>> kind
>> of recovery does ODE provide in case if ODE itself crashes, or if a
>> particular BPEL process instance within ODE crashes? Does it restart the
>> failed process automatically? And does it do so from the beginning or
>> from
>> the point that the instance crashed?
>
>
> If you stick with one-ways, yes. For two-way messages, the client
> connection
> is usually lost so it's a bit more messy.
>
>
>>
>>
>> Also, in case the process engine itself crashes and is restarted, does it
>> restart all the failed process instances automatically? Or is there a way
>> to
>> do so manually?
>
>
> All transactions are going to be discarded and the process will be back in
> the state it was before. Instances aren't failed in most cases, though.
> There's also the special case of a crash just during a two-way invoke, for
> which we'll force the invoke to enter activity recovery, which you can
> then
> retry manually.
>
> Matthieu
>
>
>>
>>
>> -Thanks
>> Kodeninja
>> --
>> View this message in context:
>> http://www.nabble.com/How-does-ODE%27s-BPEL-Engine-recover-in-the-event-of-system-failure--tp24386617p24386617.html
>> Sent from the Apache Ode User mailing list archive at Nabble.com.
>>
>>
>
>
--
View this message in context: http://www.nabble.com/How-does-ODE%27s-BPEL-Engine-recover-in-the-event-of-system-failure--tp24386617p24422793.html
Sent from the Apache Ode User mailing list archive at Nabble.com.