feat: Added new option enableProvidedByTopology

[lukasmetzner]

We are reintroducing a feature originally present in v2.10.0 to prevent pods from getting stuck in the pending state in clusters with non-cloud nodes. This feature is now optional and can be enabled via the Helm Chart. By default, it remains disabled to avoid compatibility issues with Nomad clusters, which have a different CSI spec implementation.

Learn more about it in #400.

[codecov]

Codecov Report

Attention: Patch coverage is 53.65854% with 19 lines in your changes missing coverage. Please review.

Project coverage is 35.85%. Comparing base (a9d8505) to head (3f62d81).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
internal/app/app.go	0.00%	6 Missing ⚠️
cmd/aio/main.go	0.00%	4 Missing ⚠️
cmd/controller/main.go	0.00%	3 Missing ⚠️
cmd/node/main.go	0.00%	3 Missing ⚠️
internal/driver/node.go	72.72%	2 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #780      +/-   ##
==========================================
- Coverage   35.95%   35.85%   -0.10%     
==========================================
  Files          20       20              
  Lines        1847     1874      +27     
==========================================
+ Hits          664      672       +8     
- Misses       1150     1168      +18     
- Partials       33       34       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

---

🚨 Try these New Features:

• Flaky Tests Detection - Detect and resolve failed and flaky tests

[resmo]

sorry for the question here but as a nomad user, we probably also want make use of this feature.

So I briefly looked through some docs and this is what I found:

On nomad client nodes, we can define meta info:

client {
  enabled       = true
  node_class    = "prod"

  meta {
    owner           = "ops"
    rack            = "rack-12-1"
  }
}

Which seems to be used in ebs csi plugin

topology_request {
  required {
    topology { segments { rack = "R1", zone = "us-east-1a" } }
    topology { segments { rack = "R2", zone = "us-east-1a" } }
  }
  preferred {
    topology { segments { rack = "R1", zone = "us-east-1a"} }
  }
}

that is why I tried this to check if this would work with v2.10.0, but it didn't but, but shouldn't it? Is there maybe just a small mistake somewhere?

client {
  enabled = true

  meta {
    "instance.hetzner.cloud/provided-by" = "cloud"
  }

[lukasmetzner]

@resmo I don't have much Nomad knowledge yet, but it might be enough to specify an anti-affinity towards the non-cloud nodes on the CSI Node Job.

This should also suffice in Kubernetes, but there is an upstream issue in the scheduler, which we built this workaround for.