Reference › Changelog

Changelog

What changed for users in 1.5.0, and since 1.1.

1.5.0

Monitoring replaces the checks list as the workspace home. Each row is a monitoring subject with the status of its rules, filtered by All or With incidents, and links straight into that subject’s Timeline.
The old checks list and the per-rule history page are gone. /checks redirects to Monitoring, so existing links still land. Removed with them, and not yet replaced: the uptime history strip and sorting by name, status or up-since.

The form is now Website monitoring. Saving it creates a monitoring subject with one built-in HTTP signal and one Reachability rule — you no longer create those separately.
Location agreement gains “All locations”, alongside Any and Majority.

Timeline — what happened to a subject, newest first, with how long a closed incident lasted.
Observation log — every report each location sent, with a day picker and paging. An Unaccepted filter shows rows that were stored but not used for a verdict, and why. This is what answers “the workers are running but the rule says no data”.
Rule page — one input per selected location, what each is reporting, the agreement in force, and the confirm/close holds.

Slack alerts now link to the subject timeline instead of a rule’s history.
Monitoring warns when the workspace has no Slack webhook: incidents are still detected and recorded, but nobody is notified until one is added.

API v2 speaks the product’s language — locations, website monitoring, incidents — and is self-contained. GET /v2/incidents answers “what is wrong right now”. API v1 is unchanged and supported.
Python SDK 1.5.0 targets v2 only. Its version now tracks the server it talks to. Staying on v1? Pin uptimer-python-sdk<1.

Success and error notifications could disappear before they could be read.
Security: a workspace member could read another workspace’s monitor configuration, or delete it, by using its identifier directly. Deleting it also removed its subject, signal, rule and history. All identifier-addressed routes are now scoped to the workspace in the URL.

A monitor’s status is now decided across all of its locations, not by the newest check to arrive. The form gains Locations Required to Fail — Majority of locations (default) or At least one location. A location that never reports counts as unknown and stays inside the decision, so a half-silent monitor reads No Data instead of trusting whichever location answered. See Monitors & incidents.
A problem must last 2 minutes before it is confirmed and alerted, and a recovery must last 2 minutes before the incident closes. Short blips no longer page you, though they do show up in the history.
History is now a timeline of incidents — pending, problem, no data, recovering, closed — with the failing and silent locations on each event and the error they reported, replacing the old list of individual check results.
The monitor list gains two badges — Pending and Recovering — shown while a monitor is inside one of the 2-minute holds.
Slack alerts were rewritten. Each one now says how many locations are failing and which, quotes the error the check returned, and says how long the problem ran before the alert or how long the outage lasted. The monitor’s URL left the title, and the no-data alert is amber.

HTTP(S) checks now send Uptimer/<version> (+https://uptimer.myuptime.info) instead of Go-http-client/1.1, so you can recognise and filter monitoring traffic in logs and analytics. The value carries no workspace, worker or location, and is not configurable.

Signing out now also ends the session at your OIDC provider, so the next login no longer signs the same person straight back in. Providers that advertise end_session_endpoint need no configuration; for the rest, set server.auth.oidc.end_session_endpoint and, if it renames the return-URL parameter, server.auth.oidc.post_logout_redirect_param. See Authentication.

A worker keeps its rules in memory and receives them over gRPC. worker.db.dsn and UPTIMER__WORKER__DB__DSN are no longer read — still accepted and ignored, so existing configurations keep working. You can delete the setting and drop the second uptimer_worker database. A worker still needs its /data volume for worker.uuid and worker.pem.

Patched vulnerable dependencies: golang.org/x/text and google.golang.org/grpc.
The worker is now a stateless library shared by every build, which removes a class of drift between how self-hosted and hosted workers behave.
Fixed missing icons on the debug page.

The image moved to GitHub Container Registry — ghcr.io/myuptime-info/uptimer. Pull a pinned :X.Y.Z tag (or :edge for pre-releases). The old Docker Hub image is retired.

New uptimer migrate command plus server.db.boot_migrate — run schema migrations as a one-shot job and gate rollouts on a fully-migrated database.
PostgreSQL migrations are now versioned (with data backfills) and upgrade-safe; SQLite stays auto-migrated for dev. See Choosing a database.

server --services … replaces the old run command — select services with --services api,ui,grpc,availabilities.
New server.sqids_salt setting (give it a unique value in production).

Worker-regions management in the dashboard (still being refined).
Assign regions when creating or updating a rule over the API — the rule payload takes a regions field (region names), so a headless/SDK workflow can create a rule that actually runs instead of sitting at “No Data”.
New grpc_request_duration_seconds metric on /metrics.

A pluggable architecture (job registry + route/gRPC extension seams) lets a build add services and routes without forking core.