I need help
We have cluster ignite with some caches and updated data in this caches from kafka, usual use StreamSingleTupleExtractor.
But we have one "Multi cache" for one Object, it's mean 3 cache for storage one object
1 cache for storage full object
2 cache for storage keys for 1 cache
3 cache for storage part of full object from 1 cache
It's dificult structure give me fastest access to data by key
And i use classic Kafka listener (@KafkaListener) for this topic, because i want get all values from all 3 cache for update use getAllAsync, not for every single msg
In test it's look perfect, but on production sometimes i take heavy load on network when read from kafka
for example
You can see this, for 4k records needed 1 hours, but it's can process 5к per 5s easily (limit kafka prop)
And my question is - "Why? What cluster doing and how fix it" ?
My opinion, its sync caches between nodes, but its looks crazy, my ALL caches size about 3gb
I started with 15k record limit in first time, down to 10k, and now down to 5k and this problem doesn't go away...
Ignite version is 2.12.0
