Yeah, I got a bit confused because you didn't say anything about nodes
crashing or how the question actually related to Infinispan. So I assumed
it had something to do with
https://issues.jboss.org/browse/ISPN-3063 :)
On Thu, May 16, 2013 at 6:50 PM, Mircea Markus <mmarkus(a)redhat.com> wrote:
I think we're discussing about two different things here.
My question was a general one: what if an XAResouce ack the commit to the
TransactionManager and then the Recovery Process (TM-side process
independent of Infinispan) determines that the given transaction is in
doubt. As per Jonathan's email the TM doesn't handle this well.
In our case with async commit the scenario above is possible when the node
crashes after ack the commit to the TM and before broadcasting the
CommitCommand. So we simply shouldn't support async commit when the users
want recovery.
HTH
On 16 May 2013, at 10:32, Dan Berindei <dan.berindei(a)gmail.com> wrote:
> Mircea, I think I'm missing something here. Why would the originator
send a TxCompletionNotificationCommand at all if the commit command was
asynchronous?
>
> I don't think recovery should require the originator to send a
TxCompletionNotificationCommand.
> Our commit commands can't fail anyway, so the only way for a transaction
to become in-doubt would be if the cache crashed before sending the
command. (Or maybe if another resource's commit phase failed.)
Cheers,
--
Mircea Markus
Infinispan lead (
www.infinispan.org)
_______________________________________________
infinispan-dev mailing list
infinispan-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev