You can manage this all by hand, as you are doing, or you can use
resources to manage it. For example, if you have 32 core machines,
and mongo needs lots of beef, set the mongo CPU requirement to 31
cores. Nothing else should land there except best-effort work, which
is lower priority anyway.
Th
Hi,
i have a mongo cluster that uses the i2.xlarge instances and did not want
the api worker instances to be deployed there as these don't need such big
machines. also as i am using coreos with k8s and did not want any pods
landing on the etcd machines.
On 29 August 2016 at 01:21, 'Tim Hockin' via