Created attachment 1337567 [details] free-int_edit_resource_limits_page Description of problem: The configuration about cluster resource override does not take effect. This can be shown on web console: both Memory/CPU, Limit/Request can be shown in the Edit Resource Limits page, and the scope is wrong, e.g.Memory is shown "100 MiB min to ..." See screenshot free-stg does not have the problem Version-Release number of selected component (if applicable): v3.7.0-0.147.0 (online version 3.6.0.38) How reproducible: Always Steps to Reproduce: 1. Already clear in above description Actual results: Expected results: Additional info:
(In reply to Xingxing Xia from comment #0) > Description of problem: > free-stg does not have the problem Now upgraded free-stg [v3.7.0-0.178.2 (online version 3.6.0.35)] also has the same issue. Please check
What are the expected results? Can you please clarify what is expected to be different? Notes: 1. The resource limits are no longer being hidden in the UI 2. The min pod/container memory "limit" has been lowered to 200Mi 3. The memory request->limit ratio is now 0.5 (50%) 4. The new request->limit ratio results in min memory "request" of 100Mi
Sorry, didn't get to know the changes happen as your "Notes" :) It was thought well known issue since https://bugzilla.redhat.com/show_bug.cgi?id=1324825#c12 and later reoccurred in other bugs https://url.corp.redhat.com/1522957 (RH internal), the attached case OSE/workitem?id=OCP-10075 from online sub-team was for tracking it, which checks Online UI should only show the input box of Memory Limit. Per your "Notes", I checked points 2 ~ 4 are working in free-stg. However, in the DC Edit/Set Resource Limits page, when input and save: CPU request 50m, limit 200m Memory request 120Mi, limit 200Mi The DC's pod yaml is: resources: limits: cpu: 390m memory: 200Mi requests: cpu: 19m memory: 100Mi Per it, the input for CPU request/limit and Memory request are ignored. But UI allows user inputs for them. This is bug. And your point 1 seems not the actual result.
Created attachment 1347169 [details] CPU request and limit and memory request can be set but not applied in pod (video)
`limitRequestOverrides` are not being set in the config.js that comes back to the web console, which usually means that `ClusterResourceOverrides` are not set in master-config.yaml. https://docs.openshift.org/latest/admin_guide/overcommit.html#configuring-masters-for-overcommitment
Same issue on online-int: OpenShift Master: v3.7.9 (online version 3.6.0.90) Kubernetes Master: v1.7.6+a08f5eeb62
The config is not getting plumbed through to the console by the API server in 3.7 as discovered in Bug 1533564. 3.7 fix: https://github.com/openshift/origin/pull/18076 3.8 fix: https://github.com/openshift/origin/pull/18077
*** Bug 1533564 has been marked as a duplicate of this bug. ***
Now the ResourceOverride is enabled on free-int: only memory limit is shown on resource limit page. But the scope is still wrong: "Can't be less than 100 MiB". It should be "200Mi" instead of "100Mi".
bingli -- you need to test 3.7.z to verify the fix. We've changed how it works in 3.9, and free-int is now on 3.9.
Hi, Samuel, there are two problems: A. one is comment 9 about the min value, min value should min memory limit that can be input and make pod running. B. Comment 7's 3.7 fix PR is merged in origin release-3.7, but not picked up in OCP enterprise-3.7: $ cd /local/github.com/ose_repo; git pull $ git checkout -b enterprise-3.7 upstream/enterprise-3.7 $ git log --pretty="%h %cd - %s (%an)" upstream/enterprise-3.7 | grep "plumb through the clusterresourceoverride" # get empty output # While 3.8 picked: git log --pretty="%h %cd - %s (%an)" upstream/enterprise-3.8 | grep "plumb through the clusterresourceoverride" dff8f87855 Thu Jan 11 12:59:05 2018 -0500 - plumb through the clusterresourceoverride settings (David Eads) Also double confirmed the test by launching latest 3.7.35 env, web console still shows CPU, steps are: 1. Add in master-config.yaml admissionConfig: pluginConfig: ClusterResourceOverride: ... Restart master. (In OCP 3.9, need extra `oc edit cm/webconsole-config -n openshift-web-console`, add below `true` and re-deploy web console pod: data: webconsole-config.yaml: | features: clusterResourceOverridesEnabled: true **** In OCP 3.7, web is not seprated, does it need what extra setting? Is above master-config.yaml modification and master restart enough? **** ) 2. In the test project, set limitrange that is `oc export`ed from free-int 3. Re-login to web console, check DC Edit Resources Limit page
> Comment 7's 3.7 fix PR is merged in origin release-3.7, but not picked up in OCP enterprise-3.7 David, can you check? > In OCP 3.7, web is not seprated, does it need what extra setting? Is above master-config.yaml modification and master restart enough? Correct. You don't need that setting in 3.7.
(In reply to Xingxing Xia from comment #12) > Hi, Samuel, there are two problems: > A. one is comment 9 about the min value, min value should min memory limit > that can be input and make pod running. Samuel, this display issue can be seen in free-int 3.9, is there PR for it? If need separate web bug for tracking it, I'll open one, thanks
100MiB is now allowed, so the min is correct. See https://github.com/openshift/origin/pull/18553
Will verify when above new 3.7 PR merged
Created attachment 1412972 [details] 3.7.40 fixed result Tracked puddle update, 3.7 PR is merged into 3.7 OCP since v3.7.40 (latest version). Tested per comment 12 steps in v3.7.40 env, bug is fixed, web console now only shows Memory limit input box and the input range is right. Clusters in bug title are already 3.9 or to be 3.9 in which version the issue is already fixed. Thus move bug to VERIFIED
For online-stg, it's still version v3.7.23 (online version 3.6.0.90). So this issue still exists. Will verify this bug once online-stg is upgraded.
Move this bug to verified. I've filed a bug especially for online environments: bug 1562286
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:0636