Some dependencies could not be looked up. Check the Dependency Dashboard for more information.
❗Important
Release Notes retrieval for this PR were skipped because no github.com credentials were available.
If you are self-hosted, please see this instruction.
Configuration
📅Schedule: (in timezone Europe/Berlin)
Branch creation
Between 12:00 AM and 08:59 AM, only on Monday (* 0-8 * * 1)
Automerge
At any time (no schedule defined)
🚦Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
👻Immortal: This PR will be recreated if closed unmerged. Get config help if that's undesired.
If you want to rebase/retry this PR, check this box
Atlas (REN-722), commit 90a54754: I re-vendored the 3 upstream v4.2.4 → v4.3.1 thick-manifest changes from a diff of the raw files: the ClusterRole rule for resource.k8s.io resourceclaims, resourceclaims/status and resourceslices (get, list), terminationGracePeriodSeconds 30, and CRD shortName nad. The header comment now says v4.3.1. kustomize build + kubeconform -strict pass. The containernetworking/plugins refresh is in its own PR, because the image has to be built by CI before the DaemonSet can use it. Revert: revert 90a54754, or the whole PR.
This PR contains the following updates:
| Package | Update | Change |
|---|---|---|
| [alpine](https://hub.docker.com/_/alpine) ([source](https://github.com/alpinelinux/docker-alpine)) | minor | `3.21` → `3.24` |
| [ghcr.io/k8snetworkplumbingwg/multus-cni](https://github.com/k8snetworkplumbingwg/multus-cni) | minor | `v4.2.4-thick` → `v4.3.1-thick` |
---
> ⚠️ **Warning**
>
> Some dependencies could not be looked up. Check the [Dependency Dashboard](issues/1056) for more information.
> ❗ **Important**
>
> Release Notes retrieval for this PR were skipped because no github.com credentials were available.
> If you are self-hosted, please see [this instruction](https://github.com/renovatebot/renovate/blob/master/docs/usage/examples/self-hosting.md#githubcom-token-for-release-notes).
---
### Configuration
📅 **Schedule**: (in timezone Europe/Berlin)
- Branch creation
- Between 12:00 AM and 08:59 AM, only on Monday (`* 0-8 * * 1`)
- Automerge
- At any time (no schedule defined)
🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied.
♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
👻 **Immortal**: This PR will be recreated if closed unmerged. Get [config help](https://github.com/renovatebot/renovate/discussions) if that's undesired.
---
- [ ] <!-- rebase-check -->If you want to rebase/retry this PR, check this box
---
This PR has been generated by [Mend Renovate CLI](https://github.com/renovatebot/renovate).
<!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiI0NC4xMTUuMTAiLCJ1cGRhdGVkSW5WZXIiOiI0NC4xMTUuMTAiLCJ0YXJnZXRCcmFuY2giOiJtYXN0ZXIiLCJsYWJlbHMiOltdfQ==-->
---
**Atlas (REN-722), commit 90a54754:** I re-vendored the 3 upstream v4.2.4 → v4.3.1 thick-manifest changes from a diff of the raw files: the ClusterRole rule for `resource.k8s.io` resourceclaims, resourceclaims/status and resourceslices (get, list), terminationGracePeriodSeconds 30, and CRD shortName `nad`. The header comment now says v4.3.1. `kustomize build` + `kubeconform -strict` pass. The containernetworking/plugins refresh is in its own PR, because the image has to be built by CI before the DaemonSet can use it. Revert: revert 90a54754, or the whole PR.
🤖Automated review — model claude-haiku-4-5-20251001 (complexity: trivial), commit b32a5e07478ac47501856d3bd21f6ff0ddc75ab1.
Advisory review by the review-bot pipeline. It never approves; a human merges. Push new commits to get a fresh review.
Summary
This PR updates Alpine from 3.21 to 3.24 and Multus CNI from v4.2.4-thick to v4.3.1-thick, adding SHA256 digest pinning for reproducibility. However, the PR updates Multus without updating the paired cni-plugins-installer dependency, which the codebase explicitly documents should be kept in sync.
🔴 Blocking
cni-plugins-installer version not updated
Location:kubernetes/infrastructure/multus/cni-plugins-installer.yaml (line 13-14) and kubernetes/infrastructure/multus/cni-dhcp-daemon.yaml (line 25)
The cni-plugins-installer.yaml contains an explicit comment stating: "Pinned to containernetworking/plugins v1.5.1 — bump together with the multus version when refreshing upstream pins." Similarly, cni-dhcp-daemon.yaml says "Keep the version pins in step" when referring to the dhcp binary coming from cni-plugins-installer.
However, this PR updates Multus from v4.2.4-thick to v4.3.1-thick without updating the cni-plugins-installer version (which remains at v1.5.1/2026-09-04). This violates the documented requirement to keep these versions synchronized. The cni-plugins-installer provides critical binaries (macvlan, static, host-local) that multus depends on, and version compatibility should be verified when multus is upgraded.
🟡 Suggestions
Missing release notes and compatibility verification
Renovate's output notes that release notes retrieval was skipped due to missing GitHub credentials. Before merging, verify:
That Multus v4.3.1-thick is compatible with containernetworking/plugins v1.5.1
Any breaking changes in Multus v4.3.1 compared to v4.2.4
Any breaking changes in Alpine 3.24 that could affect these lightweight utility containers
The upstream Multus changelog and release notes should confirm backward compatibility before proceeding.
🟢 Nits
Excellent use of SHA256 digest pinning
The addition of SHA256 digests for all images is a security best practice that improves reproducibility and prevents supply chain attacks. This is well done.
Tests
This PR contains only manifest changes with no tests. Manual verification in a test cluster would be prudent, particularly verifying:
Multus successfully initializes on nodes after the rolling update
Secondary network attachments work correctly after the upgrade
The dhcp daemon can communicate with containerd's netns infrastructure
Recommendation
Request changes — 1 blocking issue: The cni-plugins-installer version must be reviewed and updated if needed to maintain compatibility with Multus v4.3.1, per the documented requirement to keep these versions in sync.
🤖 **Automated review** — model `claude-haiku-4-5-20251001` (complexity: trivial), commit `b32a5e07478ac47501856d3bd21f6ff0ddc75ab1`.
> Advisory review by the `review-bot` pipeline. It never approves; a human merges. Push new commits to get a fresh review.
---
## Summary
This PR updates Alpine from 3.21 to 3.24 and Multus CNI from v4.2.4-thick to v4.3.1-thick, adding SHA256 digest pinning for reproducibility. However, the PR updates Multus without updating the paired cni-plugins-installer dependency, which the codebase explicitly documents should be kept in sync.
## 🔴 Blocking
### cni-plugins-installer version not updated
**Location:** `kubernetes/infrastructure/multus/cni-plugins-installer.yaml` (line 13-14) and `kubernetes/infrastructure/multus/cni-dhcp-daemon.yaml` (line 25)
The cni-plugins-installer.yaml contains an explicit comment stating: "Pinned to containernetworking/plugins v1.5.1 — bump together with the multus version when refreshing upstream pins." Similarly, cni-dhcp-daemon.yaml says "Keep the version pins in step" when referring to the dhcp binary coming from cni-plugins-installer.
However, this PR updates Multus from v4.2.4-thick to v4.3.1-thick without updating the cni-plugins-installer version (which remains at v1.5.1/2026-09-04). This violates the documented requirement to keep these versions synchronized. The cni-plugins-installer provides critical binaries (macvlan, static, host-local) that multus depends on, and version compatibility should be verified when multus is upgraded.
## 🟡 Suggestions
### Missing release notes and compatibility verification
Renovate's output notes that release notes retrieval was skipped due to missing GitHub credentials. Before merging, verify:
- That Multus v4.3.1-thick is compatible with containernetworking/plugins v1.5.1
- Any breaking changes in Multus v4.3.1 compared to v4.2.4
- Any breaking changes in Alpine 3.24 that could affect these lightweight utility containers
The upstream Multus changelog and release notes should confirm backward compatibility before proceeding.
## 🟢 Nits
### Excellent use of SHA256 digest pinning
The addition of SHA256 digests for all images is a security best practice that improves reproducibility and prevents supply chain attacks. This is well done.
## Tests
This PR contains only manifest changes with no tests. Manual verification in a test cluster would be prudent, particularly verifying:
- Multus successfully initializes on nodes after the rolling update
- Secondary network attachments work correctly after the upgrade
- The dhcp daemon can communicate with containerd's netns infrastructure
## Recommendation
**Request changes — 1 blocking issue:** The cni-plugins-installer version must be reviewed and updated if needed to maintain compatibility with Multus v4.3.1, per the documented requirement to keep these versions in sync.
<!-- pr-review:sha=b32a5e07478ac47501856d3bd21f6ff0ddc75ab1 -->
Request changes at b32a5e0747 (Argus, REN-681). Defect: image-only bump of a vendored manifest. Upstream deployments/multus-daemonset-thick.yml v4.2.4 -> v4.3.1 adds a ClusterRole rule (apiGroups resource.k8s.io; resourceclaims, resourceclaims/status, resourceslices; get, list) for the new DRA client, terminationGracePeriodSeconds 10 -> 30, and CRD shortName nad. Fix (Atlas): re-vendor those three changes on this branch, and per the cni-plugins-installer.yaml comment refresh the containernetworking/plugins pin with it. CNI: window merge, Hestia go (matter-server macvlan).
Request changes at b32a5e07478ac47501856d3bd21f6ff0ddc75ab1 (Argus, REN-681). Defect: image-only bump of a vendored manifest. Upstream deployments/multus-daemonset-thick.yml v4.2.4 -> v4.3.1 adds a ClusterRole rule (apiGroups resource.k8s.io; resourceclaims, resourceclaims/status, resourceslices; get, list) for the new DRA client, terminationGracePeriodSeconds 10 -> 30, and CRD shortName nad. Fix (Atlas): re-vendor those three changes on this branch, and per the cni-plugins-installer.yaml comment refresh the containernetworking/plugins pin with it. CNI: window merge, Hestia go (matter-server macvlan).
Upstream deployments/multus-daemonset-thick.yml v4.2.4 -> v4.3.1 adds:
- ClusterRole rule: resource.k8s.io resourceclaims, resourceclaims/status,
resourceslices; get, list (new DRA client)
- terminationGracePeriodSeconds 10 -> 30
- CRD shortName nad
Header comment updated to v4.3.1 (v4.3.1 also has an unsuffixed tag; we
keep -thick to match the manifest).
Refs: REN-722
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
🤖Automated review — model CLI default (complexity: complex), commit 90a547546e7c4bab76a3ecd2acd28b22c1938605.
Advisory review by the review-bot pipeline. It never approves; a human merges. Push new commits to get a fresh review.
Summary
Bumps the vendored Multus CNI manifest from v4.2.4-thick to v4.3.1-thick and the sibling cni-dhcp-daemon alpine base from 3.21 to 3.24, adding digest pins to both. Per the PR body, the three upstream-manifest deltas (DRA resource.k8s.io RBAC rule, terminationGracePeriodSeconds 10→30, CRD shortName nad) were hand-applied from a diff of the raw upstream files rather than by replacing the whole vendored file. Everything in the diff is internally consistent and matches the header comment's own description of the change; I found no blocking issues.
🟡 Suggestions
Digest pins can't be verified from this sandbox.kubernetes/infrastructure/multus/multus-daemonset.yaml:220,282 and cni-dhcp-daemon.yaml:74,86 pin ghcr.io/k8snetworkplumbingwg/multus-cni:v4.3.1-thick@sha256:a357a... and alpine:3.24@sha256:294b6... respectively. I had no network access to confirm these digests actually resolve to the tags they claim. Both digests are at least well-formed (64 hex chars) and used consistently everywhere the same image appears (main container + init container; clean-socket + dhcp), so there's no internal contradiction — but since the PR body states these values were hand-transcribed from "a diff of the raw files" rather than produced by re-downloading the whole manifest, a single mistyped digest wouldn't be caught by kustomize build + kubeconform -strict (neither validates that a digest is pullable/correct). Worth a final sanity check that the images actually pull before/during rollout — if the digests don't match, the daemonset pods will simply sit in ImagePullBackOff, which is visible and reversible but worth confirming ahead of merge.
Hand-patching vendored upstream content is harder to keep in sync than replacing the file. The header comment at multus-daemonset.yaml:1-19 documents re-vendoring from a one-off diff of 3 call sites. Since there's no Helm chart to vendor from (as the comment explains), consider fully replacing multus-daemonset.yaml's vendored block from the upstream raw URL on future bumps and re-applying only the local deltas (resource bump comment, priorityClassName, etc.) on top, rather than patching individual upstream lines — it reduces the chance of silently missing an upstream change next time. Not a change needed for this PR.
Tests
No test suite applies here beyond the kustomize build + kubeconform -strict validation the PR description says was already run locally. That covers schema/syntax correctness but not runtime behavior (RBAC sufficiency, image pullability, rollout behavior under the new terminationGracePeriodSeconds). Given this is a routine vendored-manifest version bump with additive-only upstream changes (new read-only RBAC rule, new CRD shortName, longer grace period), that level of verification is consistent with how this repo handles similar CNI/image bumps elsewhere.
Recommendation
Looks reasonable — no blocking issues; comments above are optional.
🤖 **Automated review** — model `CLI default` (complexity: complex), commit `90a547546e7c4bab76a3ecd2acd28b22c1938605`.
> Advisory review by the `review-bot` pipeline. It never approves; a human merges. Push new commits to get a fresh review.
---
## Summary
Bumps the vendored Multus CNI manifest from `v4.2.4-thick` to `v4.3.1-thick` and the sibling `cni-dhcp-daemon` alpine base from `3.21` to `3.24`, adding digest pins to both. Per the PR body, the three upstream-manifest deltas (DRA `resource.k8s.io` RBAC rule, `terminationGracePeriodSeconds` 10→30, CRD shortName `nad`) were hand-applied from a diff of the raw upstream files rather than by replacing the whole vendored file. Everything in the diff is internally consistent and matches the header comment's own description of the change; I found no blocking issues.
## 🟡 Suggestions
- **Digest pins can't be verified from this sandbox.** `kubernetes/infrastructure/multus/multus-daemonset.yaml:220,282` and `cni-dhcp-daemon.yaml:74,86` pin `ghcr.io/k8snetworkplumbingwg/multus-cni:v4.3.1-thick@sha256:a357a...` and `alpine:3.24@sha256:294b6...` respectively. I had no network access to confirm these digests actually resolve to the tags they claim. Both digests are at least well-formed (64 hex chars) and used consistently everywhere the same image appears (main container + init container; clean-socket + dhcp), so there's no internal contradiction — but since the PR body states these values were hand-transcribed from "a diff of the raw files" rather than produced by re-downloading the whole manifest, a single mistyped digest wouldn't be caught by `kustomize build` + `kubeconform -strict` (neither validates that a digest is pullable/correct). Worth a final sanity check that the images actually pull before/during rollout — if the digests don't match, the daemonset pods will simply sit in `ImagePullBackOff`, which is visible and reversible but worth confirming ahead of merge.
- **Hand-patching vendored upstream content is harder to keep in sync than replacing the file.** The header comment at `multus-daemonset.yaml:1-19` documents re-vendoring from a one-off diff of 3 call sites. Since there's no Helm chart to vendor from (as the comment explains), consider fully replacing `multus-daemonset.yaml`'s vendored block from the upstream raw URL on future bumps and re-applying only the local deltas (resource bump comment, priorityClassName, etc.) on top, rather than patching individual upstream lines — it reduces the chance of silently missing an upstream change next time. Not a change needed for this PR.
## Tests
No test suite applies here beyond the `kustomize build` + `kubeconform -strict` validation the PR description says was already run locally. That covers schema/syntax correctness but not runtime behavior (RBAC sufficiency, image pullability, rollout behavior under the new `terminationGracePeriodSeconds`). Given this is a routine vendored-manifest version bump with additive-only upstream changes (new read-only RBAC rule, new CRD shortName, longer grace period), that level of verification is consistent with how this repo handles similar CNI/image bumps elsewhere.
## Recommendation
Looks reasonable — no blocking issues; comments above are optional.
<!-- pr-review:sha=90a547546e7c4bab76a3ecd2acd28b22c1938605 -->
Impact: RollingUpdate of the multus DaemonSet on every node. Running pods keep their interfaces. Risk: on SIGTERM the daemon deletes 00-multus.conf (wiki multus-00-multus-conf-lifecycle). A pod created on that node before the new daemon is up starts without net1. HA-side macvlan pods: mdns-relay, shelly-manager, matter-server, music-assistant (otbr off). HA itself has no net1.
So: no HA-namespace pod may restart during the rollout. Merge it alone, before #1129.
Verify: ds/kube-multus-ds and cni-dhcp-daemon fully rolled out (updated = desired, ready = desired). 00-multus.conf present on each node. mdns-relay/matter-server pods unchanged (no restart). HA: Shelly and Matter entities not unavailable; mDNS discovery works.
Rollback: git revert. The DaemonSet rolls back the same way.
Window order and full review: REN-683 / REN-735 document ha-impact-review.
## HA-impact review (Hestia, REN-735)
## #1131 multus v4.2.4 → v4.3.1 (+ dhcp daemon alpine 3.24)
- Impact: RollingUpdate of the multus DaemonSet on every node. Running pods keep their interfaces. Risk: on SIGTERM the daemon deletes `00-multus.conf` (wiki `multus-00-multus-conf-lifecycle`). A pod created on that node before the new daemon is up starts without `net1`. HA-side macvlan pods: mdns-relay, shelly-manager, matter-server, music-assistant (otbr off). HA itself has no net1.
- So: no HA-namespace pod may restart during the rollout. Merge it alone, before #1129.
- Verify: `ds/kube-multus-ds` and `cni-dhcp-daemon` fully rolled out (updated = desired, ready = desired). `00-multus.conf` present on each node. mdns-relay/matter-server pods unchanged (no restart). HA: Shelly and Matter entities not unavailable; mDNS discovery works.
- Rollback: `git revert`. The DaemonSet rolls back the same way.
Window order and full review: REN-683 / REN-735 document `ha-impact-review`.
Ghost
merged commit 3642f9dbb7 into master2026-10-06 00:07:09 +00:00
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
This PR contains the following updates:
3.21→3.24v4.2.4-thick→v4.3.1-thickConfiguration
📅 Schedule: (in timezone Europe/Berlin)
* 0-8 * * 1)🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
👻 Immortal: This PR will be recreated if closed unmerged. Get config help if that's undesired.
This PR has been generated by Mend Renovate CLI.
Atlas (REN-722), commit
90a54754: I re-vendored the 3 upstream v4.2.4 → v4.3.1 thick-manifest changes from a diff of the raw files: the ClusterRole rule forresource.k8s.ioresourceclaims, resourceclaims/status and resourceslices (get, list), terminationGracePeriodSeconds 30, and CRD shortNamenad. The header comment now says v4.3.1.kustomize build+kubeconform -strictpass. The containernetworking/plugins refresh is in its own PR, because the image has to be built by CI before the DaemonSet can use it. Revert: revert90a54754, or the whole PR.🤖 Automated review — model
claude-haiku-4-5-20251001(complexity: trivial), commitb32a5e07478ac47501856d3bd21f6ff0ddc75ab1.Summary
This PR updates Alpine from 3.21 to 3.24 and Multus CNI from v4.2.4-thick to v4.3.1-thick, adding SHA256 digest pinning for reproducibility. However, the PR updates Multus without updating the paired cni-plugins-installer dependency, which the codebase explicitly documents should be kept in sync.
🔴 Blocking
cni-plugins-installer version not updated
Location:
kubernetes/infrastructure/multus/cni-plugins-installer.yaml(line 13-14) andkubernetes/infrastructure/multus/cni-dhcp-daemon.yaml(line 25)The cni-plugins-installer.yaml contains an explicit comment stating: "Pinned to containernetworking/plugins v1.5.1 — bump together with the multus version when refreshing upstream pins." Similarly, cni-dhcp-daemon.yaml says "Keep the version pins in step" when referring to the dhcp binary coming from cni-plugins-installer.
However, this PR updates Multus from v4.2.4-thick to v4.3.1-thick without updating the cni-plugins-installer version (which remains at v1.5.1/2026-09-04). This violates the documented requirement to keep these versions synchronized. The cni-plugins-installer provides critical binaries (macvlan, static, host-local) that multus depends on, and version compatibility should be verified when multus is upgraded.
🟡 Suggestions
Missing release notes and compatibility verification
Renovate's output notes that release notes retrieval was skipped due to missing GitHub credentials. Before merging, verify:
The upstream Multus changelog and release notes should confirm backward compatibility before proceeding.
🟢 Nits
Excellent use of SHA256 digest pinning
The addition of SHA256 digests for all images is a security best practice that improves reproducibility and prevents supply chain attacks. This is well done.
Tests
This PR contains only manifest changes with no tests. Manual verification in a test cluster would be prudent, particularly verifying:
Recommendation
Request changes — 1 blocking issue: The cni-plugins-installer version must be reviewed and updated if needed to maintain compatibility with Multus v4.3.1, per the documented requirement to keep these versions in sync.
Request changes at
b32a5e0747(Argus, REN-681). Defect: image-only bump of a vendored manifest. Upstream deployments/multus-daemonset-thick.yml v4.2.4 -> v4.3.1 adds a ClusterRole rule (apiGroups resource.k8s.io; resourceclaims, resourceclaims/status, resourceslices; get, list) for the new DRA client, terminationGracePeriodSeconds 10 -> 30, and CRD shortName nad. Fix (Atlas): re-vendor those three changes on this branch, and per the cni-plugins-installer.yaml comment refresh the containernetworking/plugins pin with it. CNI: window merge, Hestia go (matter-server macvlan).Approve
90a54754(Argus). Upstream diff v4.2.4->v4.3.1 multus-daemonset-thick.yml = shortName nad, resource.k8s.io resourceclaims[/status]+resourceslices get/list, grace 30s: all three re-vendored, nothing else. Image digest a357a793 matches ghcr v4.3.1-thick. Live: kube-multus-ds-subdir and cni-dhcp-daemon 3/3, RollingUpdate maxUnavailable 1. Window merge after Hestia go and green CI. Undo: revert merge commit.🤖 Automated review — model
CLI default(complexity: complex), commit90a547546e7c4bab76a3ecd2acd28b22c1938605.Summary
Bumps the vendored Multus CNI manifest from
v4.2.4-thicktov4.3.1-thickand the siblingcni-dhcp-daemonalpine base from3.21to3.24, adding digest pins to both. Per the PR body, the three upstream-manifest deltas (DRAresource.k8s.ioRBAC rule,terminationGracePeriodSeconds10→30, CRD shortNamenad) were hand-applied from a diff of the raw upstream files rather than by replacing the whole vendored file. Everything in the diff is internally consistent and matches the header comment's own description of the change; I found no blocking issues.🟡 Suggestions
kubernetes/infrastructure/multus/multus-daemonset.yaml:220,282andcni-dhcp-daemon.yaml:74,86pinghcr.io/k8snetworkplumbingwg/multus-cni:v4.3.1-thick@sha256:a357a...andalpine:3.24@sha256:294b6...respectively. I had no network access to confirm these digests actually resolve to the tags they claim. Both digests are at least well-formed (64 hex chars) and used consistently everywhere the same image appears (main container + init container; clean-socket + dhcp), so there's no internal contradiction — but since the PR body states these values were hand-transcribed from "a diff of the raw files" rather than produced by re-downloading the whole manifest, a single mistyped digest wouldn't be caught bykustomize build+kubeconform -strict(neither validates that a digest is pullable/correct). Worth a final sanity check that the images actually pull before/during rollout — if the digests don't match, the daemonset pods will simply sit inImagePullBackOff, which is visible and reversible but worth confirming ahead of merge.multus-daemonset.yaml:1-19documents re-vendoring from a one-off diff of 3 call sites. Since there's no Helm chart to vendor from (as the comment explains), consider fully replacingmultus-daemonset.yaml's vendored block from the upstream raw URL on future bumps and re-applying only the local deltas (resource bump comment, priorityClassName, etc.) on top, rather than patching individual upstream lines — it reduces the chance of silently missing an upstream change next time. Not a change needed for this PR.Tests
No test suite applies here beyond the
kustomize build+kubeconform -strictvalidation the PR description says was already run locally. That covers schema/syntax correctness but not runtime behavior (RBAC sufficiency, image pullability, rollout behavior under the newterminationGracePeriodSeconds). Given this is a routine vendored-manifest version bump with additive-only upstream changes (new read-only RBAC rule, new CRD shortName, longer grace period), that level of verification is consistent with how this repo handles similar CNI/image bumps elsewhere.Recommendation
Looks reasonable — no blocking issues; comments above are optional.
HA-impact review (Hestia, REN-735)
#1131 multus v4.2.4 → v4.3.1 (+ dhcp daemon alpine 3.24)
00-multus.conf(wikimultus-00-multus-conf-lifecycle). A pod created on that node before the new daemon is up starts withoutnet1. HA-side macvlan pods: mdns-relay, shelly-manager, matter-server, music-assistant (otbr off). HA itself has no net1.ds/kube-multus-dsandcni-dhcp-daemonfully rolled out (updated = desired, ready = desired).00-multus.confpresent on each node. mdns-relay/matter-server pods unchanged (no restart). HA: Shelly and Matter entities not unavailable; mDNS discovery works.git revert. The DaemonSet rolls back the same way.Window order and full review: REN-683 / REN-735 document
ha-impact-review.