Issues

• [Feature]: GpuThermalMarginWatch fails on one instantaneous sample, so transient T.Limit spikes become CONTACT_SUPPORT events

  area/health-monitorspriority/P1
  Enhancement

  NVIDIA/NVSentinel#1980 · Filter by author lfriedman-netllama opened on Oct 6, 2026
  1

  

• [Feature]: nic-health-monitor goes silently blind when sysfs reads stall, because liveness restarts it and nothing is published

  area/health-monitorspriority/P1
  Enhancement

  NVIDIA/NVSentinel#1979 · Filter by author lfriedman-netllama opened on Oct 6, 2026
  2

• [Feature]: remote-mode DCGM probe hangs ignore runtimeDebounce and probeStoreOnly, so they cannot be kept out of remediation

  area/health-monitorspriority/P1
  Enhancement

  NVIDIA/NVSentinel#1978 · Filter by author lfriedman-netllama opened on Oct 6, 2026
  1

• [Bug]: post remediation validation fails on nodes with a preexisting NoExecute taint

  bugpriority/P1
  Bug

  NVIDIA/NVSentinel#1975 · Filter by author lalitadithya opened on Oct 5, 2026
  

• feat(fault-quarantine): retain quarantine taints during post-remediation validation

  area/fault-managementpriority/P1
  Enhancement

  NVIDIA/NVSentinel#1961 · Filter by author HarshavardhanK opened on Oct 4, 2026
  1

  

• [Feature]: on-prem host diagnostics in log-collector (supportconfig / sos), and an optional Rancher support bundle

  area/fault-managementpriority/P1
  Enhancement

  NVIDIA/NVSentinel#1955 · Filter by author abonillabeeche opened on Oct 3, 2026
  1

  

• [Bug]: platform-connectors and nic-health-monitor fail on SELinux-enforcing nodes (RKE2 SELinux support)

  area/deploymentpriority/P1
  Bug

  NVIDIA/NVSentinel#1953 · Filter by author abonillabeeche opened on Oct 3, 2026
  

• [Feature]: Kubernetes-native remediation pipeline without a database

  area/health-monitorsenhancementpriority/P1
  Enhancement

  NVIDIA/NVSentinel#1950 · Filter by author abhipth opened on Oct 2, 2026
  1

  

• [Feature]: NVLink BER magnitude is only available as prose, so operators cannot threshold on severity

  area/health-monitorspriority/P1
  Enhancement

  NVIDIA/NVSentinel#1944 · Filter by author lfriedman-netllama opened on Oct 1, 2026
  3

• [Bug]: labeler expected device count ratchets upward and cannot recover, so two MIG nodes permanently mislabel an entire fleet

  area/fault-managementpriority/P1
  Bug

  NVIDIA/NVSentinel#1914 · Filter by author lfriedman-netllama opened on Sep 24, 2026
  2

• heterogenous clusters (device dimension): scoping and high level design

  priority/P1
  Task

  NVIDIA/NVSentinel#1908 · Filter by author lalitadithya opened on Sep 22, 2026
  

• heterogenous clusters (device dimension): scoping and high level design

  priority/P1
  Task

  NVIDIA/NVSentinel#1907 · Filter by author lalitadithya opened on Sep 22, 2026
  

• heterogenous clusters (node dimension): scoping and high level design

  priority/P1
  Task

  NVIDIA/NVSentinel#1906 · Filter by author lalitadithya opened on Sep 22, 2026
  

• heterogenous clusters (node dimension): scoping and high level design

  priority/P1
  Task

  NVIDIA/NVSentinel#1905 · Filter by author lalitadithya opened on Sep 22, 2026
  

• [Feature]: Support for heterogenous clusters (device dimension)

  area/fault-managementenhancementpriority/P1
  Epic

  NVIDIA/NVSentinel#1904 · Filter by author lalitadithya opened on Sep 22, 2026
  

• [Feature]: Support for heterogenous clusters (node dimension)

  area/fault-managementenhancementpriority/P1
  Epic

  NVIDIA/NVSentinel#1903 · Filter by author lalitadithya opened on Sep 22, 2026
  

• [Bug]: NVSwitch health events reach no Prometheus metric: STORE_ONLY hardcoded and dcgm_health_active_events never populated

  area/health-monitorsbugpriority/P1
  Bug

  NVIDIA/NVSentinel#1902 · Filter by author lfriedman-netllama opened on Sep 22, 2026
  2

• [Feature]: DRA-aware gpu-allocatable readiness criterion in lifecycle-manager for GPUCluster mode

  area/health-monitorsenhancementpriority/P1
  Enhancement

  NVIDIA/NVSentinel#1899 · Filter by author tanishagoyal2 opened on Sep 22, 2026
  

• [Bug]: health-events-analyzer --workers has no effect; only one event is ever in flight

  area/fault-managementpriority/P1
  Bug

  NVIDIA/NVSentinel#1897 · Filter by author lfriedman-netllama opened on Sep 22, 2026
  20

• [Feature]: [labeler] Automatic GPUs detection in both GPU operator mode

  area/coreenhancementpriority/P1
  Enhancement

  NVIDIA/NVSentinel#1892 · Filter by author tanishagoyal2 opened on Sep 22, 2026
  

• Publish the chart's CRDs as a separate nvsentinel-crds chart

  area/deploymentenhancementpriority/P1
  Enhancement

  NVIDIA/NVSentinel#1882 · Filter by author lockwobr opened on Sep 18, 2026
  2

  

• KOM: three-state / restart-aware condition handling for NPD-sourced policies

  area/fault-managementpriority/P2
  Task

  NVIDIA/NVSentinel#1876 · Filter by author dmvevents opened on Sep 18, 2026
  13

  

• [Feature]: Pin CI and UAT to GPU Operator 26.7

  enhancementpriority/P1
  Enhancement

  NVIDIA/NVSentinel#1870 · Filter by author tanishagoyal2 opened on Sep 17, 2026
  

• [Bug]: Janitor GPU reset overwrites operand node labels without checking their prior values

  area/health-monitorsbugpriority/P1
  Bug

  NVIDIA/NVSentinel#1869 · Filter by author tanishagoyal2 opened on Sep 17, 2026
  

• [Feature]: Add nvidia.com/gpu.deploy.client node selector for GPU-client DaemonSets

  area/health-monitorsenhancementpriority/P1
  Enhancement

  NVIDIA/NVSentinel#1868 · Filter by author tanishagoyal2 opened on Sep 17, 2026