Add Kustomize dependency

[fulviodenza]

Is this a BUG FIX or a FEATURE ?:

/kind dependencies

What this PR does / why we need it:

This PR adds the dependency for Kustomize as requested in #1307 in the python script to fetch dependencies, the controller-gen was not required anymore.
We can use this PR also to update the tasks.py file in case there are some dependencies to update. Further we should verify the kustomize version.

Release note:

NONE

[fulviodenza]

We should verify that this is actually the version needed, I could not find which exact version metallb uses

[fedepaol]

I guess the latest should work. What's important is pinning so we'll have reproducible behaviour.

[fedepaol]

I guess the latest should work. What's important is pinning so we'll have reproducible behaviour.

[fedepaol]

can you also use this in place where we currently call kustomize straight? IIUC should only be here

metallb/tasks.py

Line 324 in 28b13e3

fetch_controller_gen()

(also, call fetch_kustomize on top of the function)

[fulviodenza]

Hi, I've added the fetch_kustomize on top of the function as requested,
For what concerns the comment

can you also use this in place where we currently call kustomize straight

I may be misunderstanding, but as it is now I should be already using that kustomize_path when we call kustomize. Did you find any place where I'm missing it?

[fedepaol]

ah, apologies. I did a quick search and found

metallb/tasks.py

Line 335 in 28b13e3

res = run("{} kustomize config/{} > {}".format(kubectl_path, layer, output))

where we use kubectl kustomize. Ignore my comment

[fedepaol]

And checking again, I can't find a place where kustomize_path is used, so it maybe makes sense just to drop it?

[fulviodenza]

Actually I'm using it in my fetch_kustomize function to run the get version command

[fedepaol]

yep but the point of fetch_kustomize is to fetch it to use it :-)
If we use kubectl kustomize there is no point in even fetching it

[fedepaol]

so, let's use kustomize in

metallb/tasks.py

Line 335 in 28b13e3

res = run("{} kustomize config/{} > {}".format(kubectl_path, layer, output))

instead of kubectl kustomize.

Does it make sense?

[fulviodenza]

yeah it makes sense, so I keep the variable kustomize_path and will modify this line of code.

[fedepaol]

Sorry for the delay. Left a couple of comments. Can you also sign the commit?

[fulviodenza]

/kind dependencies

[fulviodenza]

I will fix the commit sign at the end so that I don't have to do it at any rebase

[fedepaol]

Tried locally, getting the following error:

inv generatemanifests
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0Warning: Failed to open the file
Warning: /home/fpaoline/devel/upstream/metallb/build/bin/kustomize/kustomize_v5
Warning: .4.0_linux_amd64.tar.gz: No such file or directory
100     9  100     9    0     0     31      0 --:--:-- --:--:-- --:--:--    31
curl: (23) Failure writing output to destination
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
  0     0    0     0    0     0      0      0 --:--:--  0:00:01 --:--:--     0curl: (6) Could not resolve host: xzf
curl: (3) URL rejected: No host part in the URL

[fulviodenza]

Can i ask you how you are testing it locally? @fedepaol

[fedepaol]

Can i ask you how you are testing it locally? @fedepaol

I just run inv generatemanifests.
CI failed with the same error: https://github.com/metallb/metallb/actions/runs/8815367797/job/24363517861?pr=2366

[fulviodenza]

I was looking at the error from the pipeline and I actually wanted to test it locally. Will fix this and back to you