[BUG] guaranteed-engine-manager-cpu and guaranteed-replica-manager-cpu settings are no longer supported

[bk201]

Describe the bug

Longhorn's guaranteed-engine-manager-cpu and guaranteed-replica-manager-cpu settings are no longer supported and replaced by guaranteed-instance-manager-cpu.

Note this applies to the upcoming v1.2.2 release (Longhorn v1.5.5) and v1.3.0 (Longhorn v1.6.0). v1.2.1 still uses Longhorn v1.4.3 and doesn't have this issue.

To Reproduce
Steps to reproduce the behavior:

1. Install the harvester with the two configurations:

• https://docs.harvesterhci.io/v1.3/install/harvester-configuration#installharvesterlonghorndefault_settingsguaranteedenginemanagercpu
• https://docs.harvesterhci.io/v1.3/install/harvester-configuration#installharvesterlonghorndefault_settingsguaranteedreplicamanagercpu

2. These settings don't take effect.

Expected behavior

• We should support configuring the CPU on instance-manager pod with the guaranteed-instance-manager-cpu setting.
• We should update the document.

Support bundle

Environment

• Harvester ISO version:
• Underlying Infrastructure (e.g. Baremetal with Dell PowerEdge R630):

Additional context
Add any other context about the problem here.

[harvesterhci-io-github-bot]

added backport-needed/1.2.2 issue: #5570.

[harvesterhci-io-github-bot]

added backport-needed/1.2.2 issue: #5569.

[harvesterhci-io-github-bot]

added backport-needed/1.3.1 issue: #5571.

[harvesterhci-io-github-bot]

added backport-needed/1.3.1 issue: #5572.

[harvesterhci-io-github-bot]

Pre Ready-For-Testing Checklist

• If labeled: require/HEP Has the Harvester Enhancement Proposal PR submitted?
  The HEP PR is at:

• Where is the reproduce steps/test steps documented?
  The reproduce steps/test steps are at:

• Is there a workaround for the issue? If so, where is it documented?
  The workaround is at:

• Have the backend code been merged (harvester, harvester-installer, etc) (including backport-needed/*)?
  The PR is at: Add installation configure of LH GuaranteedInstanceManagerCPU harvester-installer#706

  • Does the PR include the explanation for the fix or the feature?

  • Does the PR include deployment change (YAML/Chart)? If so, where are the PRs for both YAML file and Chart?
    The PR for the YAML change is at:
    The PR for the chart change is at:

• If labeled: area/ui Has the UI issue filed or ready to be merged?
  The UI issue/PR is at:

• If labeled: require/doc, require/knowledge-base Has the necessary document PR submitted or merged?
  The documentation/KB PR is at: Add installation configure of LH GuaranteedInstanceManagerCPU docs#549

• If NOT labeled: not-require/test-plan Has the e2e test plan been merged? Have QAs agreed on the automation test case? If only test case skeleton w/o implementation, have you created an implementation issue?

  • The automation skeleton PR is at:
  • The automation test case PR is at:

• If the fix introduces the code for backward compatibility Has a separate issue been filed with the label release/obsolete-compatibility?
  The compatibility issue is filed at:

[harvesterhci-io-github-bot]

Automation e2e test issue: harvester/tests#1224

[bk201]

@albinsun It looks like all v1.3/v1.2 backport issues are all verified. Can you also help check this main issue?

[albinsun]

Test Result

Test FAIL as behavior is not identical to v1.3/v1.2.

Environment

• harvester-v1.4.0-dev-20240918
  • Profile: 2 nodes QEMU/KVM (8C/16G/500G per node)
  • ui-source: Auto

Prerequisite

1. VM => ⋮ => Migrate

   • Check 1
     SCREENSHOT

Scenarios

w/o guaranteedInstanceManagerCPU

1. 🟡 No any configs -> cpu should be 960m (8 x 1000m x 0.12 (default value))

   842 / 0.12 / 1000 ~ 7.02
   Behavior not the same as v1.3.1

   • v1.4
     
   • v1.3.1
     
     Ref. [backport v1.3] [BUG] guaranteed-engine-manager-cpu and guaranteed-replica-manager-cpu settings are no longer supported #5571 (comment)
2. ⚪ Even with guaranteedEngineManagerCPU & guaranteedReplicaManagerCPU -> cpu should still 960m

   • Check 1
     SCREENSHOT
   • Check 2 ❌

     ⚠️: Fail due to reason
     SCREENSHOT

w/ guaranteedInstanceManagerCPU

1. ⚪ guaranteedInstanceManagerCPU: 10 -> cpu should be 800m (8 x 1000m x 0.10)

   • Check 1
     SCREENSHOT
   • Check 2 ❌

     ⚠️: Fail due to reason
     SCREENSHOT

2. ⚪ guaranteedInstanceManagerCPU: 13 -> cpu should be 960m (8 x 1000m x 0.12)

   • Check 1
     SCREENSHOT
   • Check 2 ❌

     ⚠️: Fail due to reason
     SCREENSHOT

3. ⚪ guaranteedInstanceManagerCPU: 12 -> cpu should be 960m (8 x 1000m x 12)

   • Check 1
     SCREENSHOT
   • Check 2 ❌

     ⚠️: Fail due to reason
     SCREENSHOT

4. ⚪ guaranteedInstanceManagerCPU: 0 -> Not to explicit requests CPU limit

   • Check 1
     SCREENSHOT
   • Check 2 ❌

     ⚠️: Fail due to reason
     SCREENSHOT

5. ⚪ guaranteedInstanceManagerCPU: q -> Should error due to non-integer

   • Check 1
     SCREENSHOT
   • Check 2 ❌

     ⚠️: Fail due to reason
     SCREENSHOT

Supportbundle

Note

1. ...

QA Check List

• Labels: not-require/test-plan, not-require/release-note, require-ui, regression...
• Update corresponding tests ticket lables

[albinsun]

Hi @w13915984028 ,
We found the behavior is not indentical to v1.3/v1.2.
Using the same 8 cpus, I get 842m (7.02 x 1000m x 0.12) on v1.4 instead of 960m (8 x 1000m x 0.12) on v1.3/v1.2.
Is this expected?

cc. @bk201

[w13915984028]

@albinsun The differences should be related to node usable CPU, in each version (with different rke2&k8s) the node usable CPU may vary as k8s reserves some CPU for itself at first, when LH tries to get a percentage CPU from the node, it is based on the kubeNode.Status.Allocatable.Cpu, not the physical available CPU on the node.

For the validation, we can compare: the LH CPU with [guaranteed-instance-manager-cpu] and without (use the default value) in the same version, with bigger guaranteed-instance-manager-cpu LH get more CPU.

Comparing the absolute CPU value between different releases can be an reference. LH is not requesting an absolute value, but a percentage, there could be differences on the final computed value.

thanks.

LH CPU resources calculation code, there looks no change.

allocatableMilliCPU := float64(kubeNode.Status.Allocatable.Cpu().MilliValue())

https://github.com/longhorn/longhorn-manager/blob/a3768493385042b65da343d569daa9e977ebf928/controller/controller_manager.go#L273

https://github.com/longhorn/longhorn-manager/blob/3dde300284b4a7ca8dfa1c2757ee2c2613c766be/controller/controller_manager.go#L161

[albinsun]

Test PASS, close as fixed.

Environment

• harvester-v1.4.0-dev-20240918
  • Profile: 2 nodes QEMU/KVM (8C/16G/500G per node)
  • ui-source: Auto
  • Allocatable CPU: 7.02

Scenarios

w/o guaranteedInstanceManagerCPU

1. No any configs -> cpu should be 842m ✔️

   

7.02 x 1000m x 0.12 (default)

1. Even with guaranteedEngineManagerCPU & guaranteedReplicaManagerCPU -> cpu should still 842m ✔️

   

   7.02 x 1000m x 0.12 (default)

w/ guaranteedInstanceManagerCPU

1. guaranteedInstanceManagerCPU: 10 -> cpu should be 702m ✔️

   

   7.02 x 1000m x 0.10

2. guaranteedInstanceManagerCPU: 13 -> cpu should be 842m ✔️

   • PXE Server
     
   • Harvester
     

   7.02 x 1000m x 0.12 (default)

3. guaranteedInstanceManagerCPU: 0 -> Not to explicit requests CPU limit ✔️

   • PXE
     
   • Harvester
     
4. ⚪ guaranteedInstanceManagerCPU: q -> Should error due to non-integer ✔️

   

Reference

1. https://docs.harvesterhci.io/v1.4/install/harvester-configuration#installharvesterlonghorndefault_settingsguaranteedinstancemanagercpu
2. https://longhorn.io/docs/1.7.1/references/settings/#guaranteed-instance-manager-cpu

QA Check List

• Labels: not-require/test-plan, not-require/release-note, require-ui, regression...
• Update corresponding tests ticket lables