[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: request vs. limit in OpenShift v3 1.0.6

Prior to the introduction of "requests", "limits" mapped to CPU CFS shares because that was all that was exposed in Docker at the time for any type of enforcement.

So in 1.0.6, if you set a limit of 300m CPU as a default in the project LimitRange, you should get the behavior you desire on the actual node.

When you move up to 1.1, the request for those containers will default to the limits value.  You should then update your LimitRange to specify a default request for all new pods, remove the default limits, and then run a new deployment so you end up with pods that only specify CPU requests and not limits.


On Wednesday, November 25, 2015, v <vekt0r7 gmx net> wrote:

I know that limits were changed with the newer OpenShift versions (they are much better in the newer versions) but our production cluster still uses OpenShift v3 1.0.6.

I'd like to create a project-wide limit which makes every pod "request" 300m CPU and allows them to burst for as much as there are CPUs (no "limit").

I can configure all of this in the DC but this is not a good solution. Is it possible to set this in the project wide limits? How would that look like for 1.0.6?

Setting "default" in project wide limit creates a "limit" directive in the pod which is not what I want.


users mailing list
users lists openshift redhat com

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]