SigNoz v0.135.0 rolls out the redesigned Dashboards experience and the new Dashboards V2 APIs. Your existing dashboards, queries, variables, and panels are migrated to the new Perses-compatible schema automatically, so no action is needed if you only use dashboards through the UI.
Who needs to act #
| If you | Then |
|---|---|
| Use dashboards only through the UI | Nothing: dashboards are migrated automatically. Check afterwards for any dashboard marked Legacy |
Have scripts calling /api/v1/dashboards |
Migrate them to the V2 APIsbefore upgrading : the V1 endpoints stop responding |
| Manage dashboards with the Terraform provider | Upgrade the provider and migrate resources |
| Use the SigNoz MCP server with an AI assistant | Upgrade it to v0.10.0 |
Upgrade Self-Hosted SigNoz #
Step 1: Back up your data
Dashboards live in the SigNoz Metastore (SQLite or Postgres) and are converted to the new schema automatically when the server starts up. Back up the Metastore before upgrading. If you are more than one release behind, check the Upgrade Path Tool for required stops (v0.131 requires a ClickHouse upgrade) before jumping to v0.135.0.
Step 2: Upgrade SigNoz
Upgrade foundryctl to pick up the v0.135.0 defaults, then re-apply your existing casting.yaml:
curl -fsSL https://signoz.io/foundry.sh | bash
foundryctl cast -f casting.yaml
If your casting.yaml pins image versions, the new defaults will not override them (your config wins), so bump the SigNoz image pin to v0.135.0 yourself.
Update the chart and upgrade:
helm repo update
helm -n signoz upgrade signoz signoz/signoz
helm upgrade takes the newest chart; add --version 0.135.0 to pin the v0.135.0 chart.
Step 3: Verify the upgrade
- Pods and containers are healthy (
kubectl get pods -n signoz,docker compose ps, orsystemctl status 'signoz-*'). - SigNoz reports v0.135.0 under Settings .
- Open Dashboards : the redesigned list page loads, and your existing dashboards render with data.
- Scan the list for rows marked Legacy . The conversion runs per dashboard, and a dashboard whose stored data cannot be converted is left on the old schema and flagged rather than failing the upgrade. Legacy dashboards are listed but cannot be opened in the new experience. To fix one, seeRecover a legacy dashboard .
What changes after migration #
- Automatic conversion. Existing dashboards, inbuilt dashboards, and templates convert to the new schema when the server starts. SVG dashboard icons stored as percent-encoded inline data URIs are preserved during conversion.
- Dashboards that can't be converted stay on the legacy format. A dashboard whose stored data cannot be converted is logged and left in the V1 format instead of blocking startup. It is listed and flagged asLegacy , and cannot be opened in the new experience. SeeRecover a legacy dashboard .
- Saved list-filter URLs no longer apply filters. The dashboards list page now filters with a query expression. The old URL parameters (
search,createdBy,updated, andtags) no longer apply filters, so bookmarks or shared links using them open the list unfiltered. SeeFind and filter dashboards . - Layouts are validated on write. Programmatic creates and updates must satisfy the grid rules. SeeDashboard layout constraints .
Recover a legacy dashboard #
From v0.136.0, Editors and Admins can re-run the migration for a dashboard flagged Legacy:
- Open Dashboards and click the row flaggedLegacy .
- In the dialog, click Retry migration .
- Wait for the Dashboard migrated to the new experience message. The list refreshes, and you can open the dashboard.
If the retry fails, or you have the Viewer role, copy the Dashboard ID from the same dialog, then click Contact Support and share the ID.
To retry through the API, see Retry a dashboard migration.
Migrate API scripts to the V2 Dashboards APIs #
The V1 dashboard endpoints are retired in v0.135.0. They respond with 501 Not Implemented, the error code dashboard_deprecated, and a message naming the V2 replacement. Move your scripts to the V2 endpoints before you upgrade:
V1 endpoint (now 501 ) |
V2 endpoint |
|---|---|
GET /api/v1/dashboards |
GET /api/v2/dashboards |
POST /api/v1/dashboards |
POST /api/v2/dashboards |
GET /api/v1/dashboards/{id} |
GET /api/v2/dashboards/{id} |
PUT /api/v1/dashboards/{id} |
PUT /api/v2/dashboards/{id} |
DELETE /api/v1/dashboards/{id} |
DELETE /api/v2/dashboards/{id} |
PUT /api/v1/dashboards/{id}/lock |
PUT /api/v2/dashboards/{id}/lock ,DELETE /api/v2/dashboards/{id}/lock |
In v0.135.0 the public sharing routes (/api/v1/dashboards/{id}/public and /api/v1/public/dashboards/{id}) are not affected and keep working. This changed later: /api/v1/public/dashboards/{id} (and its widget query sub-path) is removed in v0.143.0. See the v0.143.0 upgrade guide for the public-route migration. The share-config route /api/v1/dashboards/{id}/public still works.
One compatibility shim remains, on create only: POST /api/v2/dashboards still accepts a V1 payload (a body carrying version and no schemaVersion) and converts it to V2, so import flows and existing create automation keep working during the rollout window. This shim will be removed in a few releases (signoz#12177). PUT /api/v2/dashboards/{id} does not accept V1 payloads; it requires schemaVersion: "v6".
The V2 APIs also add a PATCH endpoint for partial updates, split lock and unlock endpoints, a pin-aware per-user listing, and pagination, sorting, and a query DSL on the list endpoint. The payload changes shape too, from flat JSON to a Perses spec tree. See the Dashboards V2 API guide for the full endpoint list, schema, and PATCH examples.
Upgrade the SigNoz MCP server #
The MCP server's dashboard tools moved to the V2 APIs in v0.10.0, so the two upgrades are coupled. On SigNoz Cloud's hosted MCP endpoint this happens for you; self-hosted MCP deployments need the version bump.
| signoz-mcp-server | SigNoz | Result |
|---|---|---|
| v0.9.x or older | v0.134.x or older | Works on the V1 dashboard API |
| v0.9.x or older | v0.135.0 or newer | Dashboard tools fail: they call the retired V1 routes |
| v0.10.0 | v0.134.x or older | Dashboard tools fail: the V2 routes do not exist yet |
| v0.10.0 | v0.135.0 or newer | Supported |
Only the dashboard tools are affected. The metrics, traces, logs, alerts, services, and docs tools work across both versions. Tool names are unchanged, so MCP client configs need no edits beyond the version bump. v0.10.0 also adds signoz_patch_dashboard, which applies an RFC 6902 JSON Patch so an assistant can edit one panel without resending the whole dashboard.
For the full list of changes and known limitations, see signoz-mcp-server#262.
Migrate Terraform-managed dashboards #
Terraform provider v0.1.0 rebuilds signoz_dashboard on the V2 APIs as a fully typed resource, so the jsonencode(...) attributes become a typed spec tree and existing config and state must be migrated. It requires SigNoz v0.135.0 or newer. This is a breaking change tracked in terraform-provider-signoz#122.
Follow the provider's v0.0.x to v0.1.0 upgrade guide for the full migration workflow. For general usage, see the Terraform provider guide.