NVIDIA Mission Control 2.5
NVIDIA's AI cluster operations software, now with autonomous job and hardware recovery for Kubernetes workloads managed by Run:ai.
Overview
NVIDIA Mission Control 2.5, published October 26, 2026 and available that day, extends its autonomous recovery engine from Slurm to Kubernetes workloads managed by NVIDIA Run:ai. Three layers cooperate: the Run:ai scheduler places and queues jobs, an autonomous job recovery service picks restart, retry or requeue from job context and failure history, and an autonomous hardware recovery service isolates suspect nodes or GPUs, records why, and runs diagnostics. Hardware faults trigger restart while the node is cordoned; transient software errors are retried without hardware remediation; degrading nodes leave the scheduling pool early. The release also adds beta AI network anomaly detection, central Keycloak authentication, hardening updates and early access for Vera Rubin NVL72. NVIDIA gives no benchmark figures.
This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.
Key features
- Autonomous recovery for Kubernetes workloads under Run:ai, previously Slurm only
- Separate job-recovery and hardware-recovery services
- Early removal of at-risk nodes before they disrupt jobs
- Beta AI-enabled network anomaly detection
- Central Keycloak authentication and early access for Vera Rubin NVL72
Use cases
- Keeping large Kubernetes training clusters running through GPU and node faults
- Reducing manual triage of failed AI jobs in AI factories
Pricing
unknown; the post gives no pricing.
Pricing reflects the source record and may change. Confirm current terms with the provider.
Dates and sources
- Released
- 26 Oct 2026
- Record updated
- 29 Oct 2026
- Record last checked
- 29 Oct 2026