Consequences of non-acking stateless bolt in stateful topology

[Alexander T <mi...@gmail.com>]

Hello!

Regarding https://issues.apache.org/jira/browse/STORM-1608, what would
happen if a single stateless bolt delayed acking in an otherwise stateful
topology? We were theorizing here that if such a stateless bolt delayed
acking for say an hour and then suddenly failed, that the last hour of
messages would be replayed against stateful bolts without the state being
reset correctly to a previous checkpoint. Would such a non-acking stateless
bolt cause this, or prevent the stateful bolts from checkpointing or
something entirely different? :)

Best regards,

Alexander

Re: Consequences of non-acking stateless bolt in stateful topology

[Alexander T <mi...@gmail.com>]

Hi Nathan,

If 1608 was to be implemented I understand it as if the stateless bolt will
be responsible for the acking itself. I was  simply wondering which
implications a delayed acking would have in this scenario.

Regards
Alexander
On Mar 9, 2016 2:19 PM, "Nathan Leung" <nc...@gmail.com> wrote:

> Unless you have abnormally high tuple timeout they will timeout and be
> replayed.  Anyways, if you bolt is stateless how would it delay acking by
> an hour?
>
> On Wed, Mar 9, 2016 at 8:14 AM, Alexander T <mi...@gmail.com>
> wrote:
>
>> Hello!
>>
>> Regarding https://issues.apache.org/jira/browse/STORM-1608, what would
>> happen if a single stateless bolt delayed acking in an otherwise stateful
>> topology? We were theorizing here that if such a stateless bolt delayed
>> acking for say an hour and then suddenly failed, that the last hour of
>> messages would be replayed against stateful bolts without the state being
>> reset correctly to a previous checkpoint. Would such a non-acking stateless
>> bolt cause this, or prevent the stateful bolts from checkpointing or
>> something entirely different? :)
>>
>> Best regards,
>>
>> Alexander
>>
>
>

Re: Consequences of non-acking stateless bolt in stateful topology

[Nathan Leung <nc...@gmail.com>]

Unless you have abnormally high tuple timeout they will timeout and be
replayed.  Anyways, if you bolt is stateless how would it delay acking by
an hour?

On Wed, Mar 9, 2016 at 8:14 AM, Alexander T <mi...@gmail.com> wrote:

> Hello!
>
> Regarding https://issues.apache.org/jira/browse/STORM-1608, what would
> happen if a single stateless bolt delayed acking in an otherwise stateful
> topology? We were theorizing here that if such a stateless bolt delayed
> acking for say an hour and then suddenly failed, that the last hour of
> messages would be replayed against stateful bolts without the state being
> reset correctly to a previous checkpoint. Would such a non-acking stateless
> bolt cause this, or prevent the stateful bolts from checkpointing or
> something entirely different? :)
>
> Best regards,
>
> Alexander
>