The host node would crash if the memory not enough.
For example,those services which running on the host node need 8G mem,but the host node has only 4G mem
Maybe it povide all memory for services running,as a result,the os slowly until crashed
Another case:A host node has many services with Inactive status and some services with ‘Active’ status,and the memory is enough for Active services,what would be happened if i actived those Inactive services?
I think rancher host maybe has Resource Reserve Protocol to abvoid this situation ,but i can’t find it
Currently, when services are scheduled, it is scheduled based purely on the number of containers on a host. As of now, we expect users to monitor the memory usage of their containers on their hosts.
There is a Github issue to allow us to schedule services based on resources.
We also would like to add a Dashboard to Rancher to be able to monitor the host resources.