[lfx-mentorship-2023-Sep-Nov] supports promote dependent resources automatically

[XiShanYongYe-Chang]

CNCF LFX mentorship
https://github.com/cncf/mentoring/tree/main/programs/lfx-mentorship/2023/03-Sep-Nov

Mentor: @jwcesign
Backup: @RainbowMango

What would you like to be added:
Provide an automatic promotion mechanism for dependent resources in karmadactl. When promoting a resource, all the resources that it depends on will be automatically promoted as well. For example, promoting the Secret that is dependent by a Deployment.

Why is this needed:

Related issue: #1862

Tasks:

TDB

[zhy76]

I am interested in this and would like to apply for lfx.

[XiShanYongYe-Chang]

Now we are still in the project proposal stage. According to the timeline, mentees can start applying at the time Wed Aug 16 - Tues Aug 29, 5:00 PM PDT.

[AdiAkhileshSingh15]

Hey,
I'm interested in this project and I plan to apply as a Mentee once LFX Fall applications are open.
Also, I'm curious to know if there are any beginner-friendly issues available for me to start contributing and get to know more about the project.

Any help would be highly appreciated.
Thanks!

[YashPimple]

Hello @jwcesign @RainbowMango I am interested in learning about this project and want to work on this project under LFX Mentorship Karmada supports promote dependent resources automatically
also, this issue seems like a great starting point for getting started with a contribution to the Karmada. Landed here from the CNCF mentorship repository

I liked how everything is very well documented. I look forward to applying to this project for this term.

[tamil07]

Hi, I am interested to contribute to this project but not as a Mentee but just as a part time as I don't wish to get paid.
Would it be possible to so?

[XiShanYongYe-Chang]

Hi, I am interested to contribute to this project but not as a Mentee but just as a part time as I don't wish to get paid.
Would it be possible to so?

Hi @tamil07, thank you very much for participating in the open-source activities of the Karmada community. I believe that you can contribute to the entire project, such as solution design, code review, test design and execution, documentation writing and promotion, and so on.

[AvineshTripathi]

Hey @XiShanYongYe-Chang wrt to this issue I looked after the normal promote feature PR with -d=true flag and based on discussion I can see the main task is to give support for CRD and for not only promote command with dep but also advance scheduling that cannot be done as finding dependency is a problem(https://karmada.io/docs/v1.3/userguide/globalview/customizing-resource-interpreter/#interpretdependency)

I tried looking for ways to get dependecies for CRDS but still have no success finding one! wanted to ask if there are any discussions wrt to it where any rough solutions was discussed so that I can deep dive those and work in similar direction for possible solutions

[XiShanYongYe-Chang]

cc the owner @jwcesign to help take an answer.

[jwcesign]

To automatically promote the dependency resource, we need to identify the specific resources on which it depends. There are two methods for parsing these dependencies: through a webhook or using Lua. Here are the steps involved in each approach:

1. Using a webhook:

   • karmadactl sends a request to the webhook and parses the dependencies.
   • Based on this information, it creates the corresponding resource template and PropagationPolicy in the karmada control plane.
   • Note that requesting the webhook may pose difficulties, such as requiring port-forwarding.

2. Using Lua configuration:

   • karmadactl retrieves the Lua configuration and parses the dependencies.
     

     karmada/pkg/karmadactl/interpret/execute.go

     Line 83 in 9f7215a

     configurableInterpreter := declarative.NewConfigurableInterpreter(nil)

   • It then generates the relevant resource template and pp in the Karmada control plane.

/cc @AvineshTripathi

[AvineshTripathi]

Thanks @jwcesign for the description I'll look into the lua interpretor today. Does that also find dependency or does it only generates the template?

[jwcesign]

Does that also find dependency or does it only generates the template?

Hi, @AdiAkhileshSingh15 , It only finds the dependencies. The related code:

karmada/pkg/util/interpreter/rule.go

Line 327 in 9f7215a

func (h *healthInterpretationRule) Run(interpreter *declarative.ConfigurableInterpreter, args RuleArgs) *RuleResult {

[AvineshTripathi]

Oh thanks let me try that as well! would check out today

[AvineshTripathi]

Hey mentors I have submitted my application for this issue and with my cover later I have added a bit about my understanding of the issue and problems we are facing and finally a little bit about solutions. Please have a look into it and do let me know if you have some more question or you need me to add more details. Thank you!

[zhy76]

I have a question after studying this problem. Karmada already supports two methods of interpreting crd:

1. webhook interpreter.
2. configuration interpreter.

Both methods can query the dependencies of resources. It seems that we only need to get the obj of the resources that need to be promoted in the member cluster during the promotion process, call ResourceInterpreter.GetDependencies(obj) to get the dependency information of the resources, create resource template and corresponding PP for each dependency.
Is my understanding correct? What points need to be paid attention to in this process? Thanks~

[jwcesign]

Hi @zhy76, the main issue is how to implement it with karmadactl. From the perspective of karmada components, this is not difficult:

1. webhook interpreter - call the webhook service from the Karmada components.
2. configuration interpreter - call it with Luva engine. (The functionality could be integrated into karmadactl.)

But with karmadactl, How can we call the service in the clusters?

[zhy76]

Task to be done:

• support promote native dependent resources automatically.
• support promote thirdparty dependent resources automatically.
• support promote CRD dependent resources automatically.
  • implement webhook interpreter adapter to get dependencies.
  • use configuration interpreter to get dependencies.
• E2E tests.
• Doc update.

[zhy76]

/assign

[jwcesign]

Hi, we need a proposal first. Can we add it to tasks?

[XiShanYongYe-Chang]

Hi @jwcesign @zhy76 The current issue is used for the release of LFX tasks. A new issue can be submitted to manage tasks.

[zhy76]

Hi @jwcesign @zhy76 The current issue is used for the release of LFX tasks. A new issue can be submitted to manage tasks.

OK, I will submit a new issue.

[XiShanYongYe-Chang]

Thank you to everyone interested in Karmada. Next year, the lfx Karmada community will also apply for more projects. Please continue to stay tuned.

It will be tracked by #4036.
/close

[karmada-bot]

@XiShanYongYe-Chang: Closing this issue.

In response to this:

Thank you to everyone interested in Karmada. Next year, the lfx Karmada community will also apply for more projects. Please continue to stay tuned.

It will be tracked by #4036.
/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.