Hi all,
I am trying to deploy latest ONAP (Beijing) using OOM (master branch) and I found some challenges using suggested HW resources for all-in-one (rancher server + k8s host on single node) deployment of ONAP:
HW resources:
- 24 VCPU
- 128G RAM
- 500G disc space
- rhel7.4 os
- rancher 1.6.14
- kubernetes 1.8.10
- helm 2.8.2
- kubectl 1.8.12
- docker 17.03-ce
Problem:
some random pods are unable to spawn and are on "Pending" state with error "No nodes are available that match all of the predicates: Insufficient pods (1).“
Analyzis:
it seems that mine server can allocate max. 110 pods, which I found from "kubectl describe nodes" output
...
Allocatable:
cpu: 24
memory: 102861656Ki
pods: 110
...
full ONAP deployment with all components enabled might be 120+ pods but I did not get even that 110 running,
maybe some race-condition for latest pods. If I disable 5+ components in onap/values.yaml, it will fit into that server but then it seems that "Minimum Hardware Configuration" described in
http://onap.readthedocs.io/en/latest/submodules/oom.git/docs/oom_cloud_setup_guide.html
is wrong (120GB RAM, 160G disc, 16vCPU)
or is there any hint how to increase that maximum number of allocatable pods ?
thanks,
Michal
|
|
_______________________________________________ onap-discuss mailing list onap-discuss@lists.onap.org https://lists.onap.org/mailman/listinfo/onap-discuss