# Volcano v1.16.0

> Orchestration & Management · 2026-10-14

A largely additive release expands multi-tenant queueing, topology-aware scheduling, observability, platform support, and Helm and CLI capabilities, with most new behavior opt-in. Operators should review the security updates and the Kubernetes and Go compatibility requirements before upgrading.

## Action needed
- **[security]** Dependency updates for upstream security advisories
  - Updates `Prometheus`, `golang.org/x/crypto`, `x/net`, `x/text`, `OpenTelemetry`, `etcd`, `gRPC`, and other indirect dependencies to address upstream security advisories.
- **[breaking]** The `--csi-storage` flag default
  - The `--csi-storage` flag now defaults to true.

## Check if affected
- **[security]** DRA capacity aggregation fix, `GHSA-j38h-7pfq-cxmw`
  - Applies if you run DRA capacity accounting.
  - The DRA capacity aggregation path no longer permits tenant-controlled device or minimum-task counts to trigger excessive iteration. This addresses `GHSA-j38h-7pfq-cxmw`.
- **[breaking]** Cluster Queue `spec.allowedNamespaces` authorization
  - Applies if you use `NamespaceQueue`.
  - Cluster Queue authorization now uses `spec.allowedNamespaces`: omitted or empty lists deny `NamespaceQueue` attachment, while `["*"]` allows all namespaces.
- **[breaking]** `NamespaceQueue` authorization and readiness checks
  - Applies if you use `NamespaceQueue`.
  - `NamespaceQueue` creation now requires the namespace and an authorized parent Queue, namespace-local RBAC permissions for `namespacequeues`, and confirmation of `Ready=True` and `Authorized=True` before workloads are submitted.
- **[breaking]** `custom.hypernode_controller_mode` migration sequence
  - Applies if you configure `custom.hypernode_controller_mode`.
  - Integrated installations must disable `custom.hypernode_controller_mode` and wait for the controller-manager rollout before enabling `standalone`, so only one process reconciles HyperNodes.
- **[breaking]** HyperNode topology plugin interfaces
  - Applies if you use `HyperNodeGradientForJobFn`, `HyperNodeGradientForSubJobFn`, or `HyperNodeCandidateFn`.
  - Custom topology plugins using these HyperNode extension points must adapt to the new behavior and rebuild: `HyperNodeGradientForJobFn`, `HyperNodeGradientForSubJobFn`, and `HyperNodeCandidateFn`.
- **[breaking]** `Kubernetes 1.36.1` and `Go 1.26` build requirements
  - Applies if you do not use `Go 1.26`.
  - The release builds against `Kubernetes 1.36.1` and requires `Go 1.26` for source builds.
- **[breaking]** `Kubernetes 1.23` deployment support
  - Applies if you use `Kubernetes 1.23`.
  - Deployments on `Kubernetes 1.23` are unsupported.
- **[breaking]** HyperNode gradient return types
  - Applies if you use `HyperNode gradient return types`.
  - Custom topology plugins must be rebuilt for the new HyperNode gradient return types.
- **[breaking]** TTL cleanup for Helm admission-init Jobs
  - Applies if you use Helm admission-init Jobs.
  - Helm admission-init Jobs now use TTL cleanup instead of immediate deletion after a successful hook.

## Other recorded changes
- 89 (added 34, defect_corrected 32, value_changed 12, constraint_changed 8, default_changed 2, removed 1)

[Full analysis](https://ratatosk.io/en/releases/volcano/v1.16.0)

[Original release notes](https://github.com/volcano-sh/volcano/releases/tag/v1.16.0)