I just want to ask you if you experience issues running your graphite-node inside a docker container with multiple carbon-cache on a multi CPU host.
I've got a setup close to yours but when I try to run two carbon-cache inside a container hosted on a two CPUs machine, I see that like when there is just one cache, only my first CPU has a big cpu-wait amount. I know that I probably have disk IO bottleneck but, at first, I thought that if I launch as many carbon-cache as I have CPUs the load will be shared between each one.
I see also that the relay in front of those carbon-caches (configured in consistent-hashing repartition mode) don't send metrics equali between all my carbon-caches.
Have you already have this kind of issues with your dockerized setup ? Have you some tips to be sure that inside my container each carbon-caches running on their own CPU ?
I just want to ask you if you experience issues running your graphite-node inside a docker container with multiple carbon-cache on a multi CPU host. I've got a setup close to yours but when I try to run two carbon-cache inside a container hosted on a two CPUs machine, I see that like when there is just one cache, only my first CPU has a big cpu-wait amount. I know that I probably have disk IO bottleneck but, at first, I thought that if I launch as many carbon-cache as I have CPUs the load will be shared between each one. I see also that the relay in front of those carbon-caches (configured in consistent-hashing repartition mode) don't send metrics equali between all my carbon-caches.
Have you already have this kind of issues with your dockerized setup ? Have you some tips to be sure that inside my container each carbon-caches running on their own CPU ?
Thank you