All skills
dynamo-troubleshoot
Collect a read-only debug bundle and classify a failed Dynamo deployment into one problem class with a next command.
- Part of NVIDIA/skills
- 1.9k installs
- DevOps & Self-hosting · Containers and clusters
When to use it
Your Dynamo pods, model-download job, or endpoint are failing and you need to know which layer to check first.
For example
You say
Use dynamo-troubleshoot on namespace dynamo-demo; the qwen-vllm-disagg deployment has worker pods stuck in CrashLoopBackOff and I need to know why.
You get
- Problem class
- Strongest signal
- Next command or patch
How it works
- Checks layers top down, from namespace and PVC to pod logs and the /v1/models endpoint.
- Read-only: it returns remediation commands instead of changing the cluster.
Before installing
Needs Python 3.10+ and kubectl with read access to pods, events, jobs, PVCs, and DynamoGraphDeployment resources in the target namespace.
This skill includes a script your assistant may run on your computer.
More Containers and clusters skills
See all- kubernetes-specialist14.9kCreates and manages Kubernetes workloads, manifests, Helm charts, RBAC rules and pod debugging.
- container-layer16Builds and caches a custom container environment so installs and config survive ephemeral sessions.
- cel-validation-advisorGuides you through writing and registering Kubernetes webhooks and CEL admission validation rules in the Run:ai repo.
- list-servicesLists all buildable Run:ai services grouped by type, reading from CLAUDE.md without scanning the repo.
Also in this pack
- accelerated-computing-cudf3.2kGuides GPU DataFrame work with cuDF, including pandas acceleration, joins, groupby, file I/O and multi-GPU jobs.
- cuopt-developer2.8kModifies, builds, tests, and debugs NVIDIA cuOpt source code and prepares contributions.
- cuopt-install2.6kInstalls cuOpt for Python, C, or server via pip, conda, or Docker and verifies it works.
- data-designer2.6kGenerates a synthetic dataset from your description of the data you need.