grafana-json-to-jsonnet

Grafana JSON Export to Jsonnet

Convert Grafana JSON exports (UI: Share → Export) to Jsonnet using grafana-code unified libraries. Produce a single self-contained file, modernize legacy panels, and preserve row structure.

Not suitable for: Refactoring existing Jsonnet (use grafana-jsonnet-refactor), content optimization (use grafana-dashboard-optimize), or Python report migration (use grafana-report-to-dashboard).

Workflow with validation

Copy this checklist and track your progress:

Conversion Progress:
- [ ] Step 1: Analyze source JSON and create inventory
- [ ] Step 2: Convert variables (verify count matches)
- [ ] Step 3: Convert rows (preserve structure)
- [ ] Step 4: Convert panels (verify count and placement)
- [ ] Step 5: Compile and fix build errors
- [ ] Step 6: Verify completeness (run validation checks)
- [ ] Step 7: Fix any missing elements

Step 1: Analyze source JSON and create inventory

Count all elements in the source JSON (panels, variables, rows). See references/verification-guide.md for inventory scripts. Capture the dashboard title and generate a new UID derived from the name (do not reuse the source UID). If the dashboard belongs to a specific repo or stack, review the local Jsonnet defaults and docs in the working directory (datasource config, time range, panel types, variables) before converting.

Step 2: Convert variables

Convert all variables with g.dashboard.variable.* constructors. After conversion, verify count matches inventory.

Step 3: Convert rows

Create rows with panels.rowPanel() or g.panel.row.new() and keep collapsed + gridPos aligned to the source JSON.

Step 4: Convert panels and assign to rows

Convert panels with unified constructors (panels.*Panel()). Add id and gridPos via panels.withIdAndPatches(...) or + { id, gridPos }. Set each panel's gridPos.y to its row gridPos.y and keep the original x/w/h unless standardizing with layouts.*.

Step 5: Compile and fix build errors

Run mixin/build.sh or mixin/build.ps1. Fix any errors.

Step 6: Verify completeness

Run verification checks from references/verification-guide.md. Ensure panel count, variable count, and row structure match source.

Step 7: Fix any missing elements

If verification fails, return to the appropriate step, add missing elements, recompile, and verify again.

Modernization guidelines

• graph -> timeseries
• singlestat -> stat
• Prefer standards.legend.* and themes.timeseries.*
• Use newer tooltip modes and legend placements
• For repo-specific styling and table patterns, load references/visual-style-guides.md.

Manual import support

• Use ${DS_*} for datasource UID in manual import mode.
• Add __inputs and __requires so Grafana can prompt for datasources.
• Keep provisioning mode (real UID) as the default line, and comment the manual line.

UID generation (required)

• Always create a new dashboard UID based on the dashboard name.
• Prefer a stable, name-derived UID (e.g., slugified name), and keep it concise (Grafana UID max length is 40).
• Do not carry over the source export UID.

Row structure preservation

CRITICAL: Grafana rows organize panels. Always preserve row structure from source JSON.

Panels belong to a row based on gridPos.y coordinate. Set each panel's Y to match its row's Y.

Example:

// Row at Y=0
local overviewRow = panels.rowPanel('Overview', collapsed=true)
+ g.panel.row.gridPos.withY(0)
+ g.panel.row.withPanels([panel1, panel2]);

// Panels at Y=0 belong to overviewRow
local panel1 = panels.statPanel(...)
+ g.panel.stat.gridPos.withY(0);  // Same Y as row

For detailed row handling, see references/full-conversion-playbook.md section 3.6.

Minimal structure (single file)

local g = import 'github.com/grafana/grafonnet/gen/grafonnet-latest/main.libsonnet';
local helpers = import '../lib/helpers.libsonnet';
local layouts = import '../lib/layouts.libsonnet';
local panels = import '../lib/panels.libsonnet';
local prom = import '../lib/prometheus.libsonnet';
local standards = import '../lib/standards.libsonnet';
local themes = import '../lib/themes.libsonnet';

// Provisioning mode (real UID). For manual import, switch to ${DS_*}.
local DATASOURCE_UID = 'prometheus-thanos';
// local DATASOURCE_UID = '${DS_PROMETHEUS}';

local config = {
  datasource: { type: 'prometheus', uid: DATASOURCE_UID },
  pluginVersion: '12.3.0',
  timezone: 'browser',
  timeFrom: 'now-6h',
  timeTo: 'now',
};

local qpsStat = panels.statPanel(
  title='QPS',
  targets=[prom.instantTarget('sum(rate(http_requests_total[1m]))', '')],
  datasource=config.datasource,
  unit=standards.units.qps,
  pluginVersion=config.pluginVersion
);

g.dashboard.new('Dashboard Name')
+ g.dashboard.withUid('dashboard-name')
+ g.dashboard.withPanels([qpsStat])

Handling complex configs

• For unsupported panel types, use Grafonnet directly and still apply standards.units.* and standards.thresholds.*.
• For advanced options, layer .with*() on top of unified constructors.

Optional scaffold script

scripts/convert_grafana_json.py generates a scaffold (entrypoint + lib + raw files). Use it only as a scratchpad: inline all panels and variables into the single file and delete raw JSON files.

Example:

python scripts/convert_grafana_json.py \
  --input <export.json> \
  --output-dir <mixin/system> \
  --system <system> \
  --datasource-type <type> \
  --datasource-uid <uid>

Completeness verification

CRITICAL: After conversion, run the functional completeness checks in the quality checklist below. For verification scripts and detailed instructions, see references/verification-guide.md.

If verification fails, return to the appropriate workflow step, fix issues, recompile, and verify again.

Quality checklist

Code quality:

• All panels use panels.*Panel() and helper libs (prom.*, standards.*, themes.*)
• Units and thresholds use standards.*
• Legacy panels modernized (graph -> timeseries, singlestat -> stat)
• Dashboard UID is regenerated from the dashboard name (no UID reuse)
• No dashboard-specific lib files or raw JSON panels remain

Functional completeness:

• Panel count matches source JSON (verified with script)
• All variables converted and present (verified with script)
• Row structure preserved (panels organized in correct rows)
• Variables populate with data when dashboard is imported
• No panels missing compared to source dashboard
• Build succeeds without errors

Formatting guardrail

• Do not run jsonnetfmt / jsonnet fmt on generated Jsonnet files. Keep formatting manual and consistent with grafana-code mixin style.

References (load as needed)

• references/visual-style-guides.md
• references/full-conversion-playbook.md
• references/lib-api-reference.md
• references/mapping.md
• references/verification-guide.md
• references/common-issues.md
• references/examples.md
• references/style-and-practices.md