Hey Walid,

I have an explanation of tuple trees (and anchoring) in my PyCon 2015
presentation on streamparse + Storm, starting at minute 11. You can find
that here:

https://youtu.be/ja4Qj9-l6WQ?t=11m

As for your second request, any topology that does the following will
implement "guaranteed processing":

1. assigns tuple IDs to tuples in the spout
2. "anchors" tuples in the bolts at every point where a tuple is emitted,
simply by passing it to the emit method
3. ack's and/or fail's tuples at every bolt stage of processing

This is explained in (good) detail in this blog post on fault tolerant
processing with Storm
<https://bryantsai.com/fault-tolerant-message-processing-in-storm/>.

Most topologies that you see in the wild will do this, since it's such a
big part of Storm's benefit.

Hope that helps,

--
Andrew Montalenti | CTO
<http://www.parsely.com/?utm_source=Signature&utm_medium=Andrew&utm_campaign=Signature>


On Sat, Jul 16, 2016 at 6:37 AM, Walid Aljoby <[email protected]>
wrote:

> Hi Everyone,
>
> I have fundamental questions about processing guarantee in Apache Storm.
> So Anyone please can help for the following:
>
> 1- Animation or clear tutorial with example show the anchoring mechanism..
> 2- Any topology code already implemented the guarantee processing function.
>
>
> Many thanks,
>
> --
> Regards
>
> WA
>

Reply via email to