Release notes
OpsProbe v0.3.56: evidence-first reliability.
A single-server reliability release that makes monitor configuration, alert delivery, and public status history easier to understand without claiming coverage the current infrastructure cannot provide.
Quick answer
Verified websites now expose selectable recommended checks, HTTP monitors can enforce bounded assertions, alert delivery attempts are visible without exposing message contents, and public status pages show 90-day uptime and response-time summaries.
What changed
- Added a website health profile for HTTP, DNS, TLS, TCP, domain lifecycle, and security-header signals, with idempotent enablement for selected scheduled checks.
- Added HTTP status, keyword presence/absence, JSON-path, and response-header latency assertions with a 512 KiB response-body limit.
- Persisted structured assertion and failure-confirmation evidence without storing response bodies or sensitive headers.
- Added an Alerts console view with status counts, attempts, last/next delivery time, and safe actionable error summaries.
- Added bounded 90-day uptime and average HTTP response-time summaries to public status pages, with explicit unavailable states.
- Expanded scheduled-target rejection for special-purpose IPv4 and IPv6 ranges.
Capability boundaries
Scheduled probes still run from one default region. Multi-node voting, browser journeys, AI diagnosis, and MCP-triggered actions remain planned, not active capabilities.
Migration and rollback
Apply additive database migration 024_monitor_assertions_and_evidence.sql before starting the API. No new environment variables are required. Rollback to the previous API binary and static bundle is compatible; leave migration 024 applied because it only adds nullable or defaulted fields.