fix(helm): update chart ceph-csi-drivers ( 1.0.1 → 1.0.4 )

[renovate]

This PR contains the following updates:

Package	Update	Change
ceph-csi-drivers	patch	1.0.1 → 1.0.4

---

Release Notes

ceph/ceph-csi-operator (ceph-csi-drivers)

v1.0.4: Release notes

Compare Source

What's Changed

• remove the default replicas from validation (backport #​538) by @​mergify[bot] in #​539
• ci: prepare for 1.0.4 release by @​Madhu-1 in #​540

Full Changelog: ceph/ceph-csi-operator@v1.0.3...v1.0.4

v1.0.3: Release notes

Compare Source

What's Changed

• controller: set replicas if not set (backport #​517) by @​mergify[bot] in #​528
• ci: use tag name instead of GIT_REF (backport #​529) by @​mergify[bot] in #​530
• ci: prepare for 1.0.3 release by @​Madhu-1 in #​531

Full Changelog: ceph/ceph-csi-operator@v1.0.2...v1.0.3

---

Configuration

📅 Schedule: (in timezone America/Chicago)

• Branch creation
  • At any time (no schedule defined)
• 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.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[claude]

ceph-csi-drivers: v1.0.1 → v1.0.3

Verdict: Safe to merge

Pure patch release. Both intermediate versions (1.0.2 and 1.0.3) consist entirely of backported bug fixes with no breaking changes, no deprecations, and no required configuration changes.

v1.0.2 fixes (backports):

• fix(helm): controllerPlugin and nodePlugin container resources — corrects Helm rendering of resource requests/limits for controller and node plugin containers
• helm: add volumeattributesclasses RBAC to ctrlplugin ClusterRoles — adds RBAC for the VolumeAttributesClasses API; the repo does not use this feature but the RBAC addition is non-breaking
• fix(helm): only render CephConnection readAffinity when crushLocationLabels is set — prevents rendering an empty readAffinity stanza when crushLocationLabels is absent
• change maxLogSize units to base 10 (remove the 'i') — corrects log size unit representation; no config change needed
• helm: fix bug with topology section in ceph-csi-drivers helm chart — fixes conditional topology block rendering (topology is not configured in this repo)
• helm: fix the fsGroupPolicy — corrects Helm rendering of fsGroupPolicy; the repo does not set this value explicitly so there is no behavior change

v1.0.3 fixes (backports):

• controller: set replicas if not set (#517) — ensures the CSI controller Deployment always has replicas set, preventing a case where the controller plugin could fail to deploy when the field was absent from the generated spec. This is a meaningful correctness fix for the RBD driver used by rook-ceph-block
• CI-only changes (tag naming, release prep) — no runtime impact

Repo impact assessment:

• Only the RBD driver is enabled in rook-ceph/csi/ceph-csi-drivers.yaml; CephFS, NFS, and NVMeoF are all enabled: false
• The configured values (kubeletDirPath, tolerations, hostNetwork, operatorConfig) are unchanged and remain fully compatible
• The HelmRepository source (https://ceph.github.io/ceph-csi-operator) is unaffected
• No CRD schema changes in this range; the OperatorConfig and Driver CRs consumed by rook-ceph v1.20.1 are unaffected

Known issues: No open bugs or regressions reported against v1.0.3 in the upstream issue tracker.

Sources consulted:

• v1.0.2 release notes
• v1.0.3 release notes
• v1.0.1 → v1.0.3 compare

[github-actions]

Claude Review Usage

When (UTC)	Version	Model	Turns	Cost
2026-07-01 20:56	1.0.1 -> 1.0.3	claude-sonnet-4-6	14	$0.2345
2026-07-07 14:54	1.0.1 -> 1.0.4	claude-sonnet-4-6	12	$0.2421

Superseded by updated review

[claude]

ceph-csi-drivers: v1.0.1 → v1.0.4

Verdict: Safe to merge

This PR extends the previously reviewed v1.0.1 → v1.0.3 range by one additional patch release (v1.0.4). Findings for v1.0.2 and v1.0.3 are unchanged from the prior review; this update covers v1.0.4 only.

v1.0.4 fix (backport):

• remove the default replicas from validation (#538, merged in #539) — removes the CRD-level default of 2 replicas for the controller plugin that was inadvertently introduced by the v1.0.3 fix for #517. The v1.0.3 default caused two regressions reported in #532:
  1. Single-node clusters always got 2 controller-plugin replicas instead of 1 (node-count logic was bypassed by the CRD default).
  2. The replicas field in OperatorConfig was ignored — the CRD default always won. With v1.0.4, the default is removed so the operator's node-count heuristic and OperatorConfig.replicas propagate correctly again.
• CI-only change (release prep) — no runtime impact.

Repo impact assessment:

• rook-ceph/csi/ceph-csi-drivers.yaml sets no explicit replicas field on either driverSpecDefaults or the RBD driver CR. Under v1.0.3 the CRD default of 2 was silently applied; after this upgrade the operator will correctly determine replicas from node count and any OperatorConfig defaults.
• Only the RBD driver is enabled (drivers.rbd.enabled: true); CephFS, NFS, and NVMeoF remain enabled: false — unchanged.
• No CRD schema changes in this range; OperatorConfig and Driver CRs are unaffected.
• No changes to kubeletDirPath, tolerations, hostNetwork, or any other configured values.

Breaking changes: None.

Deprecations: None.

Known issues: No open bugs or regressions reported against v1.0.4 in the upstream issue tracker.

Sources consulted:

• v1.0.4 release notes
• upstream issue #532 — regression report for the v1.0.3 replicas default
• upstream issue #538 — fix: remove default replicas from CRD validation
• v1.0.3 → v1.0.4 compare