doca-container-deployment

Name: doca-container-deployment
Author: NVIDIA

2.6kstars297forks

How to install doca-container-deployment?

nvidia/skills/doca-container-deployment

$npx -y skills add NVIDIA/skills --skill doca-container-deployment

Installs into the current project.

Summary

Use this skill when the user is hands-on deploying an in-bundle DOCA service container (Argus, DMS, Firefly, Flow-Inspector, OS-Inspector, UROM service) on a BlueField — kubelet standalone watching a static-pod manifests directory, YAML pod-spec drop, kubelet status / ENTRYPOINT

Files · 1

View on GitHub

SKILL.md

1	# DOCA container deployment
2
3	Where to start: This skill is for operating the cross-cutting
4	DOCA container-deployment runtime — the shared pattern every DOCA
5	service on the BlueField uses to come up (kubelet standalone agent
6	on the BlueField Arm watching a static-pod manifests directory; the
7	operator drops a YAML pod spec into that directory; kubelet schedules
8	the pod and runs the container).
9
10	If the developer has NOT yet decided container vs. bare-metal
11	("I just got a BlueField, what now?", *"my code is built, how do I
12	run it?", "how do I deploy this?"*), route them BACK to
13	[`doca-setup ## recognize`](../doca-setup/TASKS.md#recognize) first.
14	That is the front-door routing decision. The wrong failure mode is
15	to silently push every developer onto the container path because the
16	agent loaded this skill first. `## recognize` detects the system
17	shape, asks the minimum residual question, and lands the developer on
18	either this skill (when the workload is a packaged DOCA service to
19	drop on a BlueField) or the bare-metal-path sibling
20	[`doca-bare-metal-deployment`](../doca-bare-metal-deployment/SKILL.md)
21	(when the workload is a DOCA-linked application binary the developer
22	launches directly).
23
24	If the developer is already on the container path, open
25	[`TASKS.md`](TASKS.md) and start at
26	[`## configure`](TASKS.md#configure). If the question is *what shape
27	of runtime is this and what does the deployment contract look like*,
28	start at [`CAPABILITIES.md`](CAPABILITIES.md). For per-service
29	overlays, follow the per-service skill under `skills/services/` that
30	layers on top of this one — the DOCA monorepo ships six service
31	skills (Argus, DMS, Firefly, Flow-Inspector, OS-Inspector, UROM
32	service); externally-productized NVIDIA services (BlueMan, HBN, SNAP,
33	Virtio-net, DOCA Telemetry Service as productized, …) are out of
34	scope for this bundle by the strict-to-DOCA invariant — route those
35	to the public NVIDIA docs at `docs.nvidia.com/doca/sdk/`. If DOCA is
36	not installed on the BlueField target yet, route to
37	[`doca-setup`](../doca-setup/SKILL.md) first.
38
39	## Audience
40
41	This skill serves **external operators and platform teams who deploy
42	DOCA service containers on BlueField** — i.e., people who have a
43	BlueField with DOCA installed on the Arm side, a container runtime
44	plus the kubelet standalone agent already present per the BlueField
45	OS image, and the host-OS permissions the public DOCA Container
46	Deployment Guide names for the chosen service. The skill is the
47	shared deployment runtime; each per-service skill in the bundle
48	(see the list in [`## Related skills`](#related-skills)) supplies
49	the service-specific config schema, paired-workload contract, and
50	"healthy" definition.
51
52	It is not for NVIDIA developers contributing to the BlueField
53	container runtime or to kubelet itself, and it is not a generic
54	Kubernetes tutorial. Kubelet runs on the BlueField in standalone
55	mode here — no full Kubernetes control plane, no `kubectl` against
56	a cluster API server — and the substantive answer to most
57	container-deployment questions on the BlueField is the public DOCA
58	Container Deployment Guide. This skill teaches the agent which
59	guide to quote, in what order to walk it, and how to map a symptom
60	to a layer; it does NOT re-invent kubelet flags, pod-spec field
61	names, or static-pod path strings. The shared deployment runtime
62	described here is the cross-cutting layer; the per-service skill
63	(`doca-argus`, `doca-dms`, `doca-firefly`,
64	`doca-urom-svc`) supplies the per-service
65	config schema, paired-workload contract, and "healthy" definition.
66
67	## When to load this skill
68
69	Load this skill when the user is doing

People who liked this also viewed

6 picks

Type

doca-container-deployment

2.6kstars297forks

View on GitHub

How to install doca-container-deployment?

nvidia/skills/doca-container-deployment

$npx -y skills add NVIDIA/skills --skill doca-container-deployment

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# DOCA container deployment
2
3	Where to start: This skill is for operating the cross-cutting
4	DOCA container-deployment runtime — the shared pattern every DOCA
5	service on the BlueField uses to come up (kubelet standalone agent
6	on the BlueField Arm watching a static-pod manifests directory; the
7	operator drops a YAML pod spec into that directory; kubelet schedules
8	the pod and runs the container).
9
10	If the developer has NOT yet decided container vs. bare-metal
11	("I just got a BlueField, what now?", *"my code is built, how do I
12	run it?", "how do I deploy this?"*), route them BACK to
13	[`doca-setup ## recognize`](../doca-setup/TASKS.md#recognize) first.
14	That is the front-door routing decision. The wrong failure mode is
15	to silently push every developer onto the container path because the
16	agent loaded this skill first. `## recognize` detects the system
17	shape, asks the minimum residual question, and lands the developer on
18	either this skill (when the workload is a packaged DOCA service to
19	drop on a BlueField) or the bare-metal-path sibling
20	[`doca-bare-metal-deployment`](../doca-bare-metal-deployment/SKILL.md)
21	(when the workload is a DOCA-linked application binary the developer
22	launches directly).
23
24	If the developer is already on the container path, open
25	[`TASKS.md`](TASKS.md) and start at
26	[`## configure`](TASKS.md#configure). If the question is *what shape
27	of runtime is this and what does the deployment contract look like*,
28	start at [`CAPABILITIES.md`](CAPABILITIES.md). For per-service
29	overlays, follow the per-service skill under `skills/services/` that
30	layers on top of this one — the DOCA monorepo ships six service
31	skills (Argus, DMS, Firefly, Flow-Inspector, OS-Inspector, UROM
32	service); externally-productized NVIDIA services (BlueMan, HBN, SNAP,
33	Virtio-net, DOCA Telemetry Service as productized, …) are out of
34	scope for this bundle by the strict-to-DOCA invariant — route those
35	to the public NVIDIA docs at `docs.nvidia.com/doca/sdk/`. If DOCA is
36	not installed on the BlueField target yet, route to
37	[`doca-setup`](../doca-setup/SKILL.md) first.
38
39	## Audience
40
41	This skill serves **external operators and platform teams who deploy
42	DOCA service containers on BlueField** — i.e., people who have a
43	BlueField with DOCA installed on the Arm side, a container runtime
44	plus the kubelet standalone agent already present per the BlueField
45	OS image, and the host-OS permissions the public DOCA Container
46	Deployment Guide names for the chosen service. The skill is the
47	shared deployment runtime; each per-service skill in the bundle
48	(see the list in [`## Related skills`](#related-skills)) supplies
49	the service-specific config schema, paired-workload contract, and
50	"healthy" definition.
51
52	It is not for NVIDIA developers contributing to the BlueField
53	container runtime or to kubelet itself, and it is not a generic
54	Kubernetes tutorial. Kubelet runs on the BlueField in standalone
55	mode here — no full Kubernetes control plane, no `kubectl` against
56	a cluster API server — and the substantive answer to most
57	container-deployment questions on the BlueField is the public DOCA
58	Container Deployment Guide. This skill teaches the agent which
59	guide to quote, in what order to walk it, and how to map a symptom
60	to a layer; it does NOT re-invent kubelet flags, pod-spec field
61	names, or static-pod path strings. The shared deployment runtime
62	described here is the cross-cutting layer; the per-service skill
63	(`doca-argus`, `doca-dms`, `doca-firefly`,
64	`doca-urom-svc`) supplies the per-service
65	config schema, paired-workload contract, and "healthy" definition.
66
67	## When to load this skill
68
69	Load this skill when the user is doing

People who liked this also viewed

6 picks

Type