deployment-admin
Deploy and administer self-hosted CVAT with Docker Compose, serverless overlays, Helm orientation, superusers, and service troubleshooting.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- deployment-admin
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Deploy and administer self-hosted CVAT with Docker Compose, serverless overlays, Helm orientation, superusers, and service troubleshooting.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/configuration.md2.6 KB
- 参考文档references/deployment-guide.md2.7 KB
- 参考文档references/troubleshooting.md2.5 KB
- 脚本scripts/docker_compose_command_builder.py2.7 KB
03 SKILL.md 正文
CVAT deployment and administration
Use this sub-skill when the user asks how to start a self-hosted CVAT Community stack, choose Docker Compose overlays, set CVAT_HOST or CVAT_VERSION, create an admin account, enable serverless/analytics/cloud-related components, orient to Helm/Kubernetes deployment, or troubleshoot service startup and access problems.
Route first
- Read
references/deployment-guide.mdfor self-hosted Docker Compose and high-level Helm/Kubernetes orientation. - Read
references/configuration.mdfor environment variables, compose overlays, admin commands, and service boundaries. - Read
references/troubleshooting.mdfor Docker, port, browser, migration, version, serverless, and worker issues. - Use
scripts/docker_compose_command_builder.pyto generate reviewable Compose commands without executing them.
Minimal community stack
# Optional: expose through a host/domain rather than localhost.
export CVAT_HOST=your-host-or-domain
docker compose up -d
docker exec -it cvat_server bash -ic 'python3 ~/manage.py createsuperuser'
Open the web UI in a Chromium-based browser. CVAT primarily targets Chrome/Chromium; Safari/WebKit is not supported and Firefox may have caveats.
Common overlays
# Development-style stack with source builds.
docker compose -f docker-compose.yml -f docker-compose.dev.yml up -d --build
# Add serverless infrastructure for automatic annotation models.
docker compose -f docker-compose.yml -f components/serverless/docker-compose.serverless.yml up -d
# Use a specific published image tag instead of the default branch behavior.
CVAT_VERSION=v2.71.0 docker compose up -d
The exact overlay set must match the user's goal. Do not start or rebuild services unless the user explicitly approves those side effects.
Boundaries
- Use
../sdk-automation/SKILL.mdfor Python API automation against a running server. - Use
../cli-automation/SKILL.mdforcvat-clioperations. - Use
../dataset-ops/SKILL.mdfor import/export/data-format issues. - Use
../auto-annotation/SKILL.mdfor AA function implementation and Nuclio model deployment details.
Safety defaults
- Treat
docker compose down -v, volume removal, database reset, and cleanup commands as destructive; ask before suggesting execution. - Record whether a command builds images, downloads images, starts services, mutates a database, or opens a public network endpoint.
- For public access, require a deliberate
CVAT_HOST, TLS/reverse proxy, and credential plan. - Keep enterprise/online plan features distinct from CVAT Community self-hosted operations.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).