Nomad Lens

HashiCorp Nomad operations inside VS Code — the place where you already edit your job specs.

🌐 Website → · 📖 Full documentation → (source: docs/GUIDE.md)

_{Illustration of the Nomad view: job health with OOM/restart-loop flags, "won't place" diagnostics, plan diff repo↔running, and live resource usage.}

Features

• Cluster explorer — jobs with real health (a running job missing allocations shows as degraded), allocations with restart counts, tasks, nodes (ready/drain), active deployments. Multi-cluster via settings, one click to switch.
• Cluster dashboard — a webview panel with a job-health donut, the problem list, nodes needing attention, and active deployments with progress bars. Zero-dependency SVG charts, theme-aware, one Refresh away.
• Detail panels & log console — per-job and per-node webviews (allocations, resource gauges + sparklines, version history, drain/eligibility) with action buttons that reuse the confirmed commands, plus a log console with level colouring, a live filter and follow/wrap. Dashboard and job panels update in place on each poll tick.
• Global search — one QuickPick over every job of every configured cluster; pick one to switch cluster and open its panel.
• Plan diff, repo vs running — open a .nomad/.hcl job spec and run Plan Current Job File: the spec is parsed server-side and planned against the running job, and the diff opens beside your editor. If the diff contains more than you expected, you know before deploying.
• Job version history & revert — every registered version of a job with the diff of what each one changed, and a revert that first plans the old spec against what is running, so you see the rollback before you confirm it (confirmation by typing the job id).
• Live task logs — follow stdout/stderr of any task in an Output channel (streaming, not polling); multiple streams side by side.
• Incident bundle in one click — for a failed allocation: incidents/<date>-<job>-<alloc>/ with a ready-to-fill report.md (task event timeline, node, restart counts) plus the tail of stdout/stderr as attached log files. Your incident report is half-written before you start.
• Cluster snapshot — one command generates a markdown health report: problems first (degraded jobs, drained nodes, stuck deployments), full job table after. Perfect for the morning check or as a preflight baseline.

Go vulncheck auto-fix

If you also use the Go extension, its go.diagnostic.vulncheck setting ships a default of "Prompt" that the gopls language server rejects (Invalid settings: … invalid option "Prompt"). On activation Nomad Lens detects this and sets a valid value ("Off" by default), with a dismissable notification and an undo action. Opt out with nomadLens.autoFixGoVulncheck: false, or prefer active scanning with nomadLens.goVulncheckFixValue: "Imports".

Security

ACL tokens are read from an environment variable (you configure the variable name per cluster via tokenEnv) — tokens are never stored in settings and never displayed.

Settings

"nomadLens.clusters": [
  { "name": "dev",     "address": "http://nomad-dev.example:4646" },
  { "name": "prod",    "address": "https://nomad.example:4646", "namespace": "default", "tokenEnv": "NOMAD_TOKEN_PROD" }
]

Development

npm install
npm test             # unit tests + integration against a throwaway `nomad agent -dev`
npm run build        # bundle to dist/
npm run site         # generate site/guide.html from docs/GUIDE.md
# F5 in VS Code launches the Extension Development Host

With Docker

docker compose run --rm tests      # full suite (unit + integration) in a container, pinned Nomad
docker compose --profile demo up   # dev Nomad on http://127.0.0.1:4646 + sample job

tests runs the whole suite reproducibly (the same checks as CI, with an internal throwaway nomad agent -dev). The demo profile starts a persistent Nomad with a sample job (two allocations that log, including error lines to try the grep out): point the extension at it via nomadLens.clusters → http://127.0.0.1:4646. Requires Docker Desktop (the services run privileged with a writable cgroup, because the Nomad client creates its own cgroups).

The core (API client + report renderers) has no VS Code dependency and lives in src/core/. The integration test spins up nomad agent -dev on a random port, registers a sample job, and verifies plan diffs; if nomad is not installed the integration tests are skipped. Zero runtime dependencies: the Nomad HTTP API is consumed with Node's native fetch.

Release & automation

Everything is driven by CI (.github/workflows/):

• Publish on tag — pushing a v* tag runs the full test suite, packages the .vsix onto the GitHub Release, then publishes to the VS Code Marketplace (vsce publish) and, if configured, Open VSX (ovsx publish). A guard aborts if the tag doesn't match version in package.json.
  • Required secret: VSCE_PAT — an Azure DevOps Personal Access Token for the Marketplace publisher named in package.json (publisher). See the vsce publishing docs.
  • Optional secret: OVSX_PAT — an Open VSX token. Omit it and that step is skipped.
  • The publish job uses the marketplace environment — add required reviewers there if you want a manual approval gate before each store publish.
• Backlog sync — editing BACKLOG.md on main mirrors it to GitHub milestones + issues (scripts/backlog-sync.mjs, zero deps, native fetch). Each ## … heading becomes a milestone, each NOM-n item an issue (anchored by a hidden <!-- backlog:NOM-n --> marker so retitling never duplicates). Checked items close their issue; a fully-checked section closes its milestone. BACKLOG.md stays the single source of truth — GitHub is a read-only mirror. Run it manually (with a dry-run toggle) from the Actions tab.

To cut a release: bump version in package.json, add a CHANGELOG.md entry, commit, then git tag -a vX.Y.Z -m "Release X.Y.Z" && git push --follow-tags.

License

MIT