Add configmap in openshift-config namespace and watch it #57
Conversation
openshift-ci-robot
added
size/L
|
/retest |
|
@deads2k - Is it ok to revert to old revision of a config when the current revision fails for x minutes or something similar to that? Do we have logic for it currently? |
ravisantoshgudimetla
force-pushed
the
add-configmap
branch
from
da8af4c
to
6a8a632
Compare
|
/retest |
6 similar comments
|
/retest |
|
/retest |
|
/retest |
|
/retest |
|
/retest |
|
/retest |
|
/lgtm |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: ravisantoshgudimetla, sjenning The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
| kind: KubeSchedulerConfiguration | ||
| clientConnection: | ||
| kubeconfig: /etc/kubernetes/static-pod-resources/secrets/scheduler-kubeconfig/kubeconfig | ||
| algorithmSource: |
There was a problem hiding this comment.
Shouldn't we installing default policies in some environments? For example - it is pointless to enable Cinder, Azure, GCE-PD predicates in Openshift-4.0 environment on top of AWS?
There was a problem hiding this comment.
it is pointless but doesn't hurt anything and, if we can use the same default config for all platforms, then it is worth the reduced complexity in my mind.
There was a problem hiding this comment.
There is a slight performance improvement for sure if you disable the unneeded predicates.
I am looking to enable a Cinder specific predicate which is disabled by default in Openshift enrivonments.
I guess in this case - we will have to create a configmap with cinder predicate in openshift-kube-scheduler namespace by default via this operator if deployed on Openstack and operator will merge user specified configmap with our own configmap? Or we could install Cinder specific predicate in all environments.
There was a problem hiding this comment.
- The configmap needs to be created to in
openshift-confignamespace not inopenshift-kube-schedulernamespace. - If you don't create in configmap in
openshift-confignamespace, we would start usingDefault providersimilar to our old behaviour.(without scheduler-policy.json in older releases) - If you want to enable cinder specific predicate, you need to create a configmap with cinder predicate enabled in OpenShift config namespace. I will create a docs PR which talks about how to use the policy-configmap in 4.0 once I am back from vacation.
There was a problem hiding this comment.
I understand how it works but I thought creating configmap in openshift-config was cluster-admin action? We want to ship Cinder predicate enabled by default in Openshift-4.0 installation on Openstack. This is not something admin should have to configure.
So lets say - if we did create the configmap policy-configmap in openshift-config namespace - so I guess admin will have to edit the configmap to make any changes to scheduler policy?
There was a problem hiding this comment.
I'd say creating configmap in openshift-config is the only way at this point of time. Usually it's done through UI as post-install operations. Let me check if there is a generic way for defining set of post-install operations in 4.0. If it exists, you can perhaps add the policy configmap there.
There was a problem hiding this comment.
If you don't create in configmap in
openshift-confignamespace, we would start usingDefault providersimilar to our old behaviour.(without scheduler-policy.json in older releases)
What is the 'Default provider` bahavior?
There was a problem hiding this comment.
It is similar to what we have currently in 3.11, https://docs.openshift.com/container-platform/3.11/admin_guide/scheduling/scheduler.html#default-predicates, https://docs.openshift.com/container-platform/3.11/admin_guide/scheduling/scheduler.html#default-priorities. In 4.0, we have additional default priority available imagelocality..
At a high level, following is what this PR does:
/cc @sjenning