Hello!
I would like to know if the set up for TOPOLOGY_MAX_SPOUT_PENDING will
be applied only when the spout was set up for emitting anchored tuples?
I've made some testes and I could see that the spout was slow down only
when the tuples were anchored. When the tuples were not anchored then, with
this set up, the spout was NOT working as expected (meaning slowing
down).
If this is the case, that this set up to work properly only when the spout
emits anchored tuples, then I advice the Storm team to update the
documentation accordingly in order do not create confusion
(see
https://storm.incubator.apache.org/documentation/Running-topologies-on-a-production-cluster.html
).
Also, see it explicitly that Config.TOPOLOGY_MESSAGE_TIMEOUT_SECS will be
taken into consideration only when you are anchoring the tuples (as it is
stated in the book of P. Taylor Goetz
storm-blueprints-patterns-distributed-real-time-computation
https://www.packtpub.com/big-data-and-business-intelligence/storm-blueprints-patterns-distributed-real-time-computation
).
Hope that these remarks help other to save some time.
Best regards,
Florin
