Skip to content

mcp-data-platform-v1.69.1

Choose a tag to compare

View all tags
@github-actions github-actions released this 29 May 08:17
· 61 commits to main since this release
v1.69.1
e91f0e0
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

mcp-data-platform v1.69.1

A patch release that makes the admin observability dashboards from v1.69.0 work out of the box. In v1.69.0 the API Gateway and Health tabs could redirect a logged-in admin to the login page and required manual Prometheus configuration. This release fixes the authentication path and adds Prometheus auto-discovery so the tabs work with zero config and degrade gracefully when no metrics backend is present.

Fixes

API Gateway and Health tabs no longer redirect to login (#496)

The portal SPA calls the PromQL proxy (/api/v1/observability/*) directly using the browser session cookie. The proxy was mounted with header-only authentication, which accepts a Bearer token or API key but not the session cookie, so cookie-authenticated requests received a 401 and the portal redirected to the login page.

The proxy now accepts the portal session cookie (the same way the admin and portal APIs do) in addition to Bearer/API-key tokens. The proxy authorizer resolves a cookie-authenticated user first and falls back to token authentication for programmatic callers. Authentication (401) and the observability:read capability (403) are still enforced, so access control is unchanged. Programmatic Bearer/API-key callers are unaffected.

Features

Prometheus auto-discovery and graceful "metrics unavailable" state (#496)

  • The PromQL proxy now defaults to the in-cluster service http://mcp-data-platform-prometheus:9090 when observability.prometheus.url is not set. The observability tabs work with no configuration; set observability.prometheus.url only to point at a Prometheus deployed under a different name.
  • With auto-discovery, an absent or unreachable backend surfaces as a connection failure (502) or timeout (504) rather than the not-configured 503. The portal's "metrics unavailable" detection now covers 502/503/504, so the API Gateway and Health tabs render a clean empty state for any backend-availability condition instead of a generic error.

Upgrade notes

  • No configuration change is required. If you previously set observability.prometheus.url, it continues to take precedence over the auto-discovered default.
  • To serve live metrics, run a Prometheus reachable at mcp-data-platform-prometheus:9090 (or set observability.prometheus.url) that scrapes the platform's /metrics endpoint. Without one, the observability tabs show the "metrics unavailable" state.

Installation

Homebrew (macOS)

brew install txn2/tap/mcp-data-platform

Claude Code CLI

claude mcp add mcp-data-platform -- mcp-data-platform

Docker

docker pull ghcr.io/txn2/mcp-data-platform:v1.69.1

Verification

All release artifacts are signed with Cosign. Verify with:

cosign verify-blob --bundle mcp-data-platform_1.69.1_linux_amd64.tar.gz.sigstore.json \
  mcp-data-platform_1.69.1_linux_amd64.tar.gz

Full Changelog: v1.69.0...v1.69.1

Assets 25
Loading