Requested cpu is too high
For oas.greenhost.net
(which follows openappstack
master, is that intended?) we're now at 96%:
Allocated resources:
(Total limits may be over 100 percent, i.e., overcommitted.)
Resource Requests Limits
-------- -------- ------
cpu 5812m (96%) 16100m (268%)
memory 8374Mi (69%) 19222Mi (160%)
Some requests that jump out to me in a glance:
-
loki-0
: 400m -
prometheus-kube-prometheus-stack-prometheus-0
: 600m -
wekan-mongodb
: 512m (weird number which makes it seems copy-pasted from a memory request)
Together with other components (kube-prometheus-stack-operator
, promtail
, alertmanager
, eventrouter
), that means that at least 1.7 cores are reserved for monitoring and logging!
So we should probably lower many of these numbers. Also, maybe there's a toggle to prevent kubernetes from using the resource requests at all during scheduling? As long as we're only using single-node clusters there's no advantage in doing that -- except maybe showing us that we're by now installing a large set of supporting software for a supposedly small server