not fixed on version: 4.8.0-0.nightly-2021-03-08-184701

for parameter "overlaySize", the ctrcfg CR can't recognize non-digital value(such as 9asadG) as invalid value, neither prompt error messages, nor trigger mcp roll out. I think this should refer to the way of patameter "pidsLimit".(when you input non-digital value, you can't save the ctrcfg cr file successfully):
spec:
  containerRuntimeConfig:
    overlaySize: 9asadG
  machineConfigPoolSelector:
    matchLabels:
      custom-crio-overlay: overlay-size
status:
  conditions:
  - lastTransitionTime: "2021-03-09T08:42:15Z"
    message: 'Error: invalid overlaySize "-3G", cannot be less than 0'
    status: "False"
    type: Failure
  observedGeneration: 1



And another issue is the historical lastTransitionTime should keep historical data, not latest time:
spec:
  containerRuntimeConfig:
    overlaySize: 0G
  machineConfigPoolSelector:
    matchLabels:
      custom-crio-overlay: overlay-size
status:
  conditions:
  - lastTransitionTime: "2021-03-09T08:44:47Z" // this time should keep historical data, but it sync to latest time as the next item.
    message: 'Error: invalid overlaySize "-3G", cannot be less than 0'
    status: "False"
    type: Failure
  - lastTransitionTime: "2021-03-09T08:44:47Z"
    message: Success
    status: "True"
    type: Success
  observedGeneration: 3



for parameter "pidsLimit", ctrcfg description won't show error messages:
$ oc get ctrcfg set-pids-limit-master -o yaml 
apiVersion: machineconfiguration.openshift.io/v1
kind: ContainerRuntimeConfig
metadata:
  creationTimestamp: "2021-03-09T09:36:32Z"
  generation: 1
  managedFields:
...
    manager: oc
    operation: Update
    time: "2021-03-09T09:36:32Z"
  name: set-pids-limit-master
  resourceVersion: "75129"
  uid: 7416f64e-a60b-42a7-8d38-dca3d72a1d86
spec:
  containerRuntimeConfig:
    pidsLimit: -90
  machineConfigPoolSelector:
    matchLabels:
      custom-crio: pidLimit
[lyman@localhost env]$ oc get mcp 
NAME     CONFIG                                             UPDATED   UPDATING   DEGRADED   MACHINECOUNT   READYMACHINECOUNT   UPDATEDMACHINECOUNT   DEGRADEDMACHINECOUNT   AGE
master   rendered-master-ea560dfc69f54e1eb2b30821f68e56cc   True      False      False      3              3                   3                     0                      157m
worker   rendered-worker-c14e3bbca7d1434ad8024ce7587120c4   True      False      False      3              3                   3                     0                      157m


for parameter "log_size_max", whatever value I set, 0、positive or negative, the ctrcfg description will reset null for this item:
spec:
  containerRuntimeConfig: {} // this line show empty
  machineConfigPoolSelector:
    matchLabels:
      custom-crio-max: log_size_max

@minmli Could you double-check the pidsLimit, I tested the release payload I built from the MCO master, the error message showed up. (it has a print verb issue though, I will fix it).

$ oc describe containerruntimeconfig.machineconfiguration.openshift.io/set-pids

...
...
Spec:
  Container Runtime Config:
    Pids Limit:  -10
  Machine Config Pool Selector:
    Match Labels:
      Custom - Crio - Pids:  pids
Status:
  Conditions:
    Last Transition Time:  2021-03-18T18:05:27Z
    Message:               Error: invalid PidsLimit %!q(int64=-10)
    Status:                False
    Type:                  Failure
  Observed Generation:     1
Events:                    <none>

Could you provide the "log_size_max" specification? That parameter is logSizeMax, I checked the status can update after I applied the crd.

apiVersion: machineconfiguration.openshift.io/v1
kind: ContainerRuntimeConfig
...
spec:
  containerRuntimeConfig:
    logSizeMax: 10K
  machineConfigPoolSelector:
    matchLabels:
      custom-crio-log: log

Hi, Qi Wang

I checked the pidsLimit and got the same results as you pasted. So my previous issue no longer exists.

As to "log_size_max", I refer to https://docs.openshift.com/container-platform/4.7/post_installation_configuration/machine-configuration-tasks.html#create-a-containerruntimeconfig_post-install-machine-configuration-tasks. This page doesn't introduce the correct paramter "logSizeMax", I will file a doc bug to fix it.

@minmli @qiwang given MinLi's last comment, can we set this bug to "ON_QA" again or perhaps verified?

The `Last Transition Time` seems to work fine, could you also double-check this?

Spec:
  Container Runtime Config:
    Pids Limit:  10
  Machine Config Pool Selector:
    Match Labels:
      Custom - Crio - Log:  log
Status:
  Conditions:
    Last Transition Time:  2021-03-22T22:21:01Z
    Message:               Success
    Status:                True
    Type:                  Success
    Last Transition Time:  2021-03-22T22:33:03Z
    Message:               Error: invalid PidsLimit '\n'
    Status:                False
    Type:                  Failure
  Observed Generation:     2
Events:                    <none>

@Tom Sweeney, this bug still has some flakes to fix, can not move it to on_qa

@Qi Wang, oh, I get the point, if the current parameter's value is invalid, then the kubelet will probe every minutes to check if the error value not exist any more, so the 'Last Transition Time' keep up to date, yet if the check is successful, then the 'Last Transition Time' will not update. 
So the 'Last Transition Time' really works fine.

(In reply to MinLi from comment #9)
> @Qi Wang, oh, I get the point, if the current parameter's value is invalid,
> then the kubelet will probe every minutes to check if the error value not
> exist any more, so the 'Last Transition Time' keep up to date, yet if the
> check is successful, then the 'Last Transition Time' will not update. 
> So the 'Last Transition Time' really works fine.

Thanks for the explanation!
Discussed the non-digital overlaySize bug, this is a WONTFIX. The crd decode validation has been taken out from the MCO code.
I could only set up a PR as a follow up of this BZ to fix the print verb issue.

The reason we dropped the support for validation of kubeletconfig https://github.com/openshift/machine-config-operator/issues/2357

@ Qi Wang,  according to  Comment 12, does it mean we can handle the non-digital overlaySize bug?

(In reply to MinLi from comment #13)
> @ Qi Wang,  according to  Comment 12, does it mean we can handle the
> non-digital overlaySize bug?

No. ContainerRuntimeConfig is defined locally, but the overlaySize field is defined as resource.Quantity that is from kubernetes/apimachinery/pkg/api/resource. So it can not be validated inside MCO.  @harpatil  Is my understanding correct?


If it is fixable, at least I saw the pod log has the error 

W0316 16:45:13.344669       1 reflector.go:436] github.com/openshift/machine-config-operator/pkg/generated/informers/externalversions/factory.go:101: watch of *v1.ContainerRuntimeConfig ended with: an error on the server ("unable to decode an event from the watch stream: unable to decode watch event: v1.ContainerRuntimeConfig.Spec: v1.ContainerRuntimeConfigSpec.MachineConfigPoolSelector: ContainerRuntimeConfig: v1.ContainerRuntimeConfiguration.OverlaySize: unmarshalerDecoder: quantities must match the regular expression '^([+-]?[0-9.]+)([eEinumkKMGTP]*[-+]?[0-9]*)$', error found in #10 byte of ...|\":\"9asadG\"},\"machine|..., bigger context ...|:

Maybe MCO can capture this error somewhere? but I got lost in the code for now to have this fix.

@Harshal Patil 
thanks for your explanation, to my understanding, we can fix the validation of overlaySize. cc:@Qi Wang

@ Harshal Patil, ok, got you, it's acceptable if we add specification about the validation of ContainerRuntimeConfig in doc.

Btw, will you fix the print verb issue which @Qi Wang mentioned in  Comment 10? If yes, we can reopen this bug to track it.

Spec:
  Container Runtime Config:
    Pids Limit:  -10
  Machine Config Pool Selector:
    Match Labels:
      Custom - Crio - Pids:  pids
Status:
  Conditions:
    Last Transition Time:  2021-03-18T18:05:27Z
    Message:               Error: invalid PidsLimit %!q(int64=-10)
    Status:                False
    Type:                  Failure
  Observed Generation:     1
Events:                    <none>

(In reply to MinLi from comment #20)
> @ Harshal Patil, ok, got you, it's acceptable if we add specification about
> the validation of ContainerRuntimeConfig in doc.
> 
> Btw, will you fix the print verb issue which @Qi Wang mentioned in  Comment
> 10? If yes, we can reopen this bug to track it.
> 
> Spec:
>   Container Runtime Config:
>     Pids Limit:  -10
>   Machine Config Pool Selector:
>     Match Labels:
>       Custom - Crio - Pids:  pids
> Status:
>   Conditions:
>     Last Transition Time:  2021-03-18T18:05:27Z
>     Message:               Error: invalid PidsLimit %!q(int64=-10)
>     Status:                False
>     Type:                  Failure
>   Observed Generation:     1
> Events:                    <none>

Hi, there's the fix for the print issue https://github.com/openshift/machine-config-operator/pull/2485. It's merged already.

Thanks Qi for the link.

The needinfo request[s] on this closed bug have been removed as they have been unresolved for 500 days