You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@tomee.apache.org by ka...@gmail.com on 2015/03/28 13:36:24 UTC

Question about transactions and asynchronous

Client transaction context does not propagate with an asynchronous method invocation. From the Bean Developer’s view, there is never a transaction context flowing in from the client. This means, for example, that the semantics of the REQUIRED transaction attribute on an asynchronous method are exactly the same as REQUIRES_NEW.

Does this mean that when stateless calls async behavior is requries new but when async calls others behavior is normal?

I have a long running batch step and when it fails and rollback it also rolls back failure traces we write in the db so I want it in a separate transaction so I figured async seems like nice fit but I am not sure if I need requires new or not? 

Cheers

Re: Question about transactions and asynchronous

Posted by Romain Manni-Bucau <rm...@gmail.com>.
In jbatch you can configure chunks, often helps
Le 28 mars 2015 14:44, <ka...@gmail.com> a écrit :

> Hello, thanks for your reply, so I understood async correctly then?
>
> It's not really a workaround but more about handling many trace rows yet
> keep the batch fast... I might skip async if I it does not help performance
> but I still want to be sure I understand how it works :)
>
> I am using batchee and I am gonna refactor batch to be one single big
> transaction (now its one per thread) because entities getting detatched is
> killing performance.
>
> Not sure why we ended up with different transactions - did not write the
> code but I guess something in jsr352 will explain how to manage this.
> Actually a bit unsure about transactions and batches currently...
>
>
> Skickat från min iPhone
>
> > 28 mar 2015 kl. 13:40 skrev Romain Manni-Bucau <rm...@gmail.com>:
> >
> > Hi Karl
> >
> > You want requires new so use it. Would avoid issue if you refactor later
> > your code.
> >
> > What is the origin of the question? Why wanting to use a workaround?
> > Le 28 mars 2015 13:37, <ka...@gmail.com> a écrit :
> >
> >> Client transaction context does not propagate with an asynchronous
> method
> >> invocation. From the Bean Developer’s view, there is never a transaction
> >> context flowing in from the client. This means, for example, that the
> >> semantics of the REQUIRED transaction attribute on an asynchronous
> method
> >> are exactly the same as REQUIRES_NEW.
> >>
> >> Does this mean that when stateless calls async behavior is requries new
> >> but when async calls others behavior is normal?
> >>
> >> I have a long running batch step and when it fails and rollback it also
> >> rolls back failure traces we write in the db so I want it in a separate
> >> transaction so I figured async seems like nice fit but I am not sure if
> I
> >> need requires new or not?
> >>
> >> Cheers
>

Re: Question about transactions and asynchronous

Posted by ka...@gmail.com.
Hello, thanks for your reply, so I understood async correctly then?

It's not really a workaround but more about handling many trace rows yet keep the batch fast... I might skip async if I it does not help performance but I still want to be sure I understand how it works :)

I am using batchee and I am gonna refactor batch to be one single big transaction (now its one per thread) because entities getting detatched is killing performance. 

Not sure why we ended up with different transactions - did not write the code but I guess something in jsr352 will explain how to manage this. Actually a bit unsure about transactions and batches currently...


Skickat från min iPhone

> 28 mar 2015 kl. 13:40 skrev Romain Manni-Bucau <rm...@gmail.com>:
> 
> Hi Karl
> 
> You want requires new so use it. Would avoid issue if you refactor later
> your code.
> 
> What is the origin of the question? Why wanting to use a workaround?
> Le 28 mars 2015 13:37, <ka...@gmail.com> a écrit :
> 
>> Client transaction context does not propagate with an asynchronous method
>> invocation. From the Bean Developer’s view, there is never a transaction
>> context flowing in from the client. This means, for example, that the
>> semantics of the REQUIRED transaction attribute on an asynchronous method
>> are exactly the same as REQUIRES_NEW.
>> 
>> Does this mean that when stateless calls async behavior is requries new
>> but when async calls others behavior is normal?
>> 
>> I have a long running batch step and when it fails and rollback it also
>> rolls back failure traces we write in the db so I want it in a separate
>> transaction so I figured async seems like nice fit but I am not sure if I
>> need requires new or not?
>> 
>> Cheers

Re: Question about transactions and asynchronous

Posted by Romain Manni-Bucau <rm...@gmail.com>.
Hi Karl

You want requires new so use it. Would avoid issue if you refactor later
your code.

What is the origin of the question? Why wanting to use a workaround?
Le 28 mars 2015 13:37, <ka...@gmail.com> a écrit :

> Client transaction context does not propagate with an asynchronous method
> invocation. From the Bean Developer’s view, there is never a transaction
> context flowing in from the client. This means, for example, that the
> semantics of the REQUIRED transaction attribute on an asynchronous method
> are exactly the same as REQUIRES_NEW.
>
> Does this mean that when stateless calls async behavior is requries new
> but when async calls others behavior is normal?
>
> I have a long running batch step and when it fails and rollback it also
> rolls back failure traces we write in the db so I want it in a separate
> transaction so I figured async seems like nice fit but I am not sure if I
> need requires new or not?
>
> Cheers