Skip to content

pull from upstream#3

Open
fvalle1 wants to merge 1743 commits into
Elemento-Modular-Cloud:masterfrom
kubernetes:master
Open

pull from upstream#3
fvalle1 wants to merge 1743 commits into
Elemento-Modular-Cloud:masterfrom
kubernetes:master

Conversation

@fvalle1

@fvalle1 fvalle1 commented May 15, 2026

Copy link
Copy Markdown

No description provided.

k8s-ci-robot and others added 26 commits May 13, 2026 14:00
hetzner: upgrade CSI driver to v2.20.2
Disable kube-proxy when Calico runs in eBPF mode
VFS used to fall back to listing every EC2 region via DescribeRegions
and fanning out one GetBucketLocation per region whenever the initial
GetBucketLocation failed (cross-account buckets). That pulled the entire
EC2 SDK into every kops binary purely for one call.

HeadBucket can be called against any region in the partition: S3 returns
the bucket region in BucketRegion on success and in the x-amz-bucket-region
response header on cross-region 301 redirects. One call replaces the
fanout and drops the EC2 SDK from the channels binary (~28 MB smaller).
vfs: Use HeadBucket to resolve S3 bucket region
Used only for two well-known tag constants(kubernetes.io/role/elb and
kubernetes.io/role/internal-elb), shrinking the kops binary by ~4MB.
aws: Drop cloud-provider-aws dependency
Cloud.DeregisterInstance failures were logged but the node-groups loop continued,
leaving every worker tainted with kops.k8s.io/scheduled-for-update. Mark these
errors exitable so the roll bails out on the first failure.
Abort rolling update on load balancer deregister failure
Nodeup uses curl which does not support s3://, so accepting such
URLs silently leads to nodes failing to boot. Validate the scheme
upfront.
The kops create instancegroup command applies a kops.k8s.io/instancegroup
node label so workloads can target a specific instance group via affinity
or label selectors. Instance groups generated by kops create cluster did
not receive this label, leaving the two code paths inconsistent.
aws: allow disabling NTH enableScheduledEventDraining in Queue Processor mode
Signed-off-by: Ciprian Hacman <ciprian@hakman.dev>
Reject non-http(s) URLs for assets.fileRepository
Align instancegroup node label across create cluster/instancegroup
aws: apply onDemandAllocationStrategy to ASG mixed instances policy
aws: propagate taints without value to ASG tags
A single-file hostPath mount binds to the source inode at mount time,
so when something rewrites the kubeconfig the container keeps reading
the orphaned inode. Mounting the parent directory matches what
kube-scheduler and kube-controller-manager already do.
kube-proxy: bind-mount kubeconfig directory instead of the file
Signed-off-by: Ciprian Hacman <ciprian@hakman.dev>
kOps sets --cloud-provider=external for every cloud on all supported
Kubernetes versions (1.31+), so no in-tree cloud provider is ever
initialized. The --cloud-config flag and the in-tree-cloud.config file
it points to are read only by an in-tree provider, so they have been a
no-op on every supported version.
@fvalle1

fvalle1 commented May 15, 2026

Copy link
Copy Markdown
Author

@Paolo-Beci @iliy27 let's rebase this :)

k8s-ci-robot and others added 3 commits May 15, 2026 19:40
Remove the unused in-tree cloud config
The Azure cloud-controller-manager and CSI drivers will read their cloud
config from the azure-cloud-provider Secret instead of a file on disk.
This addon publishes that Secret, built from the cluster spec, as a shared
dependency applied before the components that consume it.
ameukam and others added 30 commits June 6, 2026 13:28
Signed-off-by: Arnaud Meukam <ameukam@gmail.com>
gVisor (runsc) was previously installable on any instance group role.
Restrict it to nodes with role Node: reject cluster/IG configs that
enable gVisor on control plane, apiserver, or bastion roles, strip the
gVisor config from non-worker nodeup configs, and only apply the
gVisor node label and RuntimeClass addon when a worker has it enabled.

Also harden nil handling for cluster.Spec.Containerd in nodeup config
and bootstrapchannelbuilder. Update release notes and add tests across
validation, nodeup config, gvisor builder, and instancegroup spec.
Signed-off-by: Arnaud Meukam <ameukam@gmail.com>
Bumps [actions/checkout](https://github.com/actions/checkout) from 6.0.2 to 6.0.3.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](actions/checkout@de0fac2...df4cb1c)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-version: 6.0.3
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
feat: add gVisor RuntimeClass support for containerd
…ctions/checkout-6.0.3

build(deps): bump actions/checkout from 6.0.2 to 6.0.3
Surface Calico's Felix NFTablesMode (Disabled, Enabled, Auto) as a
field on CalicoNetworkingSpec and propagate it to the calico-node
DaemonSet via FELIX_NFTABLESMODE. When left unset, the upstream
Calico chart default applies, preserving existing behavior.

On distributions where iptables is only present as a shim over
nftables (e.g. RHEL10+, Rocky10+), routing Felix's data plane
through iptables-nft / nft_compat has produced BGP session flapping
and broken pod networking on GCE. This field lets clusters opt
their Calico install into native nftables on those nodes.
Signed-off-by: Ciprian Hacman <ciprian@hakman.dev>
etcd-manager: switch to go-runner-based distroless image
kube-proxy: assert buildPod command in unit test
gVisor: add HasGVisor() helper function
…nect

dump: add --node-dump-timeout flag for per-node dump timeout
scaltest: Default node dump timeout to 5m in scalability run-test.sh
…ity scenario

Signed-off-by: Jefftree <jeffrey.ying86@live.com>
Signed-off-by: Jefftree <jeffrey.ying86@live.com>
Support etcd 3.7.0-rc.0 and allow overriding the etcd version in the scalability scenario
e2e: Filter random AZ selection by instance type availability
azure: Use IMDS attested metadata document for node identity
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.