OpsKnight 2.1.0 introduces automated runbook execution (Beta) with sandboxed agent pools, Service Automation (Beta), REST API v2, enterprise OIDC single sign-on (compatible with Microsoft Entra ID, Okta, Google Workspace, and Keycloak), SCIM 2.0 directory sync, and verified PostgreSQL 16 disaster recovery.
Service Automation (Beta): rules engine with first-match evaluation, payload mapping, shadow simulation, unattended-write authorization, and automated remediation links with full audit trails.
New
REST API v2
Documented REST API v2 endpoints for incident lifecycle, services, on-call schedules, and runbook trigger operations with scoped API keys and token authentication.
Scoped API keys, granular permission enforcement, rate-limiting, and machine-readable OpenAPI documentation.
New
Enterprise SSO & Identity
Enterprise OpenID Connect (OIDC) authentication supporting Microsoft Entra ID, Okta, Google Workspace, and generic OIDC/Keycloak providers (SAML is not supported; SSO is OIDC-only).
Automated SCIM 2.0 user and group synchronization, dynamic team assignment, and automated offboarding.
Session security registry: active session monitoring, device fingerprinting, and one-click remote session revocation.
Change
Disaster Recovery & Operational Hardening
PostgreSQL 16 verified backup, point-in-time recovery, and automated rollback tooling.
Docker Swarm secret-file resolution for DATABASE_URL_FILE across migration and bootstrap scripts.
Updated Kubernetes Helm charts (opsknight and opsknight-agent) pinned to v2.1.0.
Performance
Certification & Quality Gates
Continuous E2E validation verifying terminal state assertion across notification queuing, provider acceptance, and delivery.
Cross-browser Playwright test suite validating authenticated routes, modal interactions, and zero console errors.
A major milestone release introducing split production runtimes, Microsoft Teams ChatOps, Twilio voice paging, incident response policy engine, SCIM 2.0 provisioning, Docker Swarm HA, ManageEngine ingestion, and a durable notification delivery control plane.
Split production runtime: independently scalable Web, Scheduler, General Worker, Critical Worker, Bulk Worker, and Status Projector processes.
Docker Swarm deployment stacks (integrated and split) with Raft secrets, zero-downtime topology switching, and direct-database lifecycle.
PgBouncer pooling overlays, published sizing profiles (Small, Medium, Large, Storm), and load/correctness scenarios for deployment planning. Benchmark results are planning evidence, not guaranteed capacity.
New
ChatOps & Paging
Native Microsoft Teams ChatOps: Entra/Azure Bot integration, Adaptive Cards, incident war rooms, identity linking, and up to 3 Teams destinations per service.
Twilio voice paging (VOICE channel) for triggered-incident escalation with responder acknowledgement input and signed webhook callbacks.
Up to 3 Slack channel destinations per service with improved war room participant sync.
New
Governance & Delivery Control
Incident response policy engine: versioned SLA policies, classification, and support-hours rules at workspace and service scope.
SCIM 2.0 provisioning: Users, Groups, discovery endpoints, team membership, and bearer token rotation.
Auditor role with session registry (browser, OS, device, activity tracking, and instant revoke controls).
Durable notification delivery control plane: logical intents, traffic classes, admission/defer behavior, and operator evidence ledgers.
Status Page V3 with custom themes, announcements, subscriber verify/unsubscribe, API tokens, and snapshot exports.
Responder-grade installable mobile PWA with per-device Web Push registration and iOS recovery flows.
Custom dashboard templates, configurable widgets, live refresh, PDF export, and fullscreen NOC mode.
First stable release distributed under AGPL-3.0-only; historical v1.4.0 and earlier artifacts retain Apache-2.0.
Performance
Performance & Scale
Integrated and split runtime roles let operators scale Web, Scheduler, Critical, General, Bulk, and Status processing independently where the deployment model supports it.
PgBouncer pooling overlays and published sizing profiles reduce database-connection pressure and give operators a documented starting point for capacity planning.
Load and correctness scenarios exercise alert ingestion, incident lifecycle, escalation, notifications, status fan-out, realtime traffic, user workload, security-under-load, recovery, and queue-drain behavior. Results are planning evidence, not guaranteed capacity.
Change
2.0 Release Boundaries
One supported status page per installation; multiple independent status pages are not a released 2.0 capability.
Service Objectives/SLO UI is deferred and its route redirects; it is not advertised as a released feature.
Mobile is an installable PWA with Web Push, not a native App Store or Google Play application.
Voice paging is for initial triggered-incident alerting and responder acknowledgement; later lifecycle updates do not place additional calls.
OpsKnight remains self-hosted; v2.0.0 does not introduce a hosted OpsKnight Cloud service.
The 28 inbound contracts are the total supported inbound catalog, not 28 integrations newly introduced in 2.0.
A safer release foundation with reliable delayed paging, an administrator Health Center, hardened upgrades, complete versioned docs, and multi-architecture stable images.
Delayed escalation steps retain their configured delay instead of paging the fallback team immediately.
Orphaned escalations recover through one atomic processing claim.
Fallback delivery no longer duplicates service notifications.
New
Operational confidence
The administrator Health Center consolidates supported database, scheduler, worker, paging, provider, integration, URL, encryption, and version signals.
The release contract validates installation, upgrades, migration failure, restore, incident delivery, deployment rendering, and documentation coverage.
Stable images publish for linux/amd64 and linux/arm64; the continuously updated test image remains amd64-only.
Security
Release hardening
Disabled accounts cannot be reactivated through OIDC linking.
First-administrator bootstrap is serialized and public readiness errors are redacted.
Next.js, Auth.js, Nodemailer, and vulnerable transitive dependencies are updated to patched releases.
When an incident is serious, OpsKnight can open a Slack room, invite who is on call, and drop a video link. A Node 20 clock bug that paged the whole roster is fixed.
OpsKnight