How to Migrate from Opsgenie to Self-Hosted OnCallAlert
Atlassian stopped selling Opsgenie on June 4, 2025, and support ends on April 5, 2027. This guide is a migration plan, not just a click path: phases, a field-by-field mapping, what does not move, time and cost estimates, and a rollback.
Timeline. Per Atlassian, Opsgenie can no longer be bought, and access is shut off on April 5, 2027. Data not migrated by then is deleted. Check Atlassian's announcement for your own dates.
What are my options now that Opsgenie is ending?
You have three: move to Jira Service Management (Atlassian's own path, per-agent pricing), move to another on-call tool, or self-host. Atlassian points customers to Jira Service Management or Compass. Self-hosting keeps alert data on your own server and replaces per-user pricing with a flat yearly price. The right choice depends on team size and who must run the server.
- Jira Service Management: Atlassian lists about $21.50 per agent per month for Standard and about $55.28 for Premium, with a Free plan for up to 3 agents (checked 2026-11; volume tiers can be lower).
- Self-hosted OnCallAlert: flat $699 to $9,999 per year by plan, viewers free. You run one Docker container and PostgreSQL. See pricing.
What does the migration plan look like, phase by phase?
Plan for five phases: inventory, import, rewire alert sources, parallel run, cutover. The importer covers phase 2 in minutes. Most of your time goes to rewiring integrations and the parallel run. Every phase before cutover is reversible, because OnCallAlert only reads from Opsgenie.
- Inventory (day 1). List users, teams, schedules, escalation policies, integrations, heartbeats, routing rules and maintenance windows in Opsgenie. Note which integrations send what, and who owns each.
- Stand up and import (day 1 to 2). Run OnCallAlert, create the read-only key, preview, import (steps below).
- Rebuild what does not map (day 2 to 5). Recreate routing logic, heartbeats and maintenance windows by hand or with the Terraform, OpenTofu or Pulumi provider.
- Rewire and parallel run (1 to 2 weeks). Point alert sources at OnCallAlert while Opsgenie keeps running. Send a test alert to every service. Compare who gets paged in each.
- Cut over and keep a rollback. Switch the senders fully, keep Opsgenie read-only for at least one more week, then turn it off. To roll back, point senders back at the old Opsgenie URL.
How long does it take?
These are our planning estimates, not measurements: about 1 to 2 days of effort for a team of 5, about 1 week for 20, and 2 to 3 weeks for 50 or more, mostly spent on integrations and the parallel run. The import itself takes minutes at any size.
| Team size | Hands-on effort | Calendar time with parallel run |
|---|---|---|
| 5 users | 1 to 2 days | About 1 week |
| 20 users | About 1 week | 2 weeks |
| 50+ users | 2 to 3 weeks | 3 to 4 weeks |
Before you start
- Run OnCallAlert. You need the Docker image and a PostgreSQL database. The Docker quickstart covers it.
- Sign in as an admin.
- Pick a quiet week. You will run both systems side by side for a few days.
Step 1. How do I create a read-only key in Opsgenie?
In Opsgenie go to Settings → Integrations → Add integration → API, with Read and Configuration access. Note the key and your region (US or EU). If Opsgenie has already moved into Jira Service Management, use the Jira Service Management option instead, with an API token from id.atlassian.com. OnCallAlert only reads from your old tool. Nothing changes there.
Step 2. How do I preview and run the import?
Open Admin → Import (/admin/import), choose Opsgenie, enter the key and region and click Preview. Read the
counts and warnings, tick the confirmation box and import. The import is one transaction: everything is imported or nothing is.
- Check which users match existing accounts and which will be created.
- Fix warnings in Opsgenie or note them for phase 3.
- Delete the Opsgenie key afterwards. OnCallAlert does not store it.
What comes across, field by field?
Users, contact methods, teams, schedules with rotations, escalation policies and integrations come across. Integrations become one service each. This table is taken from the migration docs; nothing outside it is imported.
| Opsgenie | OnCallAlert | Notes |
|---|---|---|
| Users (not blocked) | Users | Matched to existing users by email, others created. No password: sign in with GitHub or Google using the same verified email, or OIDC. |
| Contact methods, notification rules | Contact methods and notification rules | Email, SMS and voice. |
| Teams | Teams and members | |
| Schedules and rotations | Schedules and rotations | Including overnight and multi-day shifts. |
| Escalation policies | Escalation policies | Same steps, delays and repeats. |
| Integrations | One service per integration, with an integration key | Uses the policy from the team's default routing rule. |
What does not migrate?
Anything not in the table above does not come across. Expect to recreate these by hand: routing rules beyond a team's default, heartbeats, maintenance windows, alert and incident history, and custom actions or automations. Export history from Opsgenie first if you need it, because Atlassian deletes it after April 5, 2027.
- Routing rules: only the default rule's policy is used. Model extra routing as separate services, one per integration key.
- Heartbeats: recreate as heartbeat monitors (created through the GraphQL API) on the target service, with a timeout in minutes.
- Maintenance windows: use service maintenance mode (pause for 1, 2 or 4 hours, or until a time you pick).
- Alert history: not imported. Keep an export for audit needs.
Step 3. How do I point my alerts at OnCallAlert?
The result page lists every old integration with its new URL and what to change. This is the step that takes the most time.
- Grafana, Prometheus Alertmanager and Site24x7: swap in the new URL.
- Email integrations: send to the new address. This needs inbound email set up on your server.
- Everything else, such as Datadog or CloudWatch: OnCallAlert does not accept other tools' alert formats. Change the sender to post JSON with
summaryanddetailsto the new URL. See the alert ingestion API.
Step 4. How do I cut over safely?
Test every service, run both tools in parallel for one to two weeks, then switch senders fully and keep Opsgenie for a rollback window. Imported users land in their accounts with schedules and contact methods already in place.
- Send a test alert to every service and confirm the right person is paged.
- Have each on-call user pair the mobile app and acknowledge a test alert.
- Switch Opsgenie off only after a quiet week with no missed pages.
How much does it cost compared with Opsgenie and Jira Service Management?
For 5, 20 and 50 users, OnCallAlert Team costs $1,499 a year flat, against $1,197 to $11,970 for Opsgenie Standard and $1,290 to $12,900 for Jira Service Management Standard at list price. Opsgenie is no longer sold, so existing customers pay until April 2027 only.
| Users | Opsgenie Essentials / Standard (per year) | Jira Service Management Standard (per year) | OnCallAlert Team (per year) |
|---|---|---|---|
| 5 | $567 / $1,197 | $1,290 | $1,499 |
| 20 | $2,268 / $4,788 | $5,160 | $1,499 |
| 50 | $5,670 / $11,970 | $12,900 | $1,499 |
Opsgenie list prices: $9.45 Essentials and $19.95 Standard per user per month billed annually. Jira Service Management: about $21.50 per agent per month. Atlassian offers volume discounts, so your quote may be lower. OnCallAlert Team includes the mobile app and up to 50 users; add roughly $20 a month for your own server if you do not have one. At 5 users a small team may be cheaper on Opsgenie Essentials until it shuts off, and Core ($699 a year, up to 15 users) has no mobile push. Jira Service Management also bundles service desk features OnCallAlert does not have. Try your own numbers in the cost calculator.
Why does OnCallAlert cost less to own than a per-user plan?
One flat license covers your whole team: $1,499 a year (Team) for up to 50 on-call users, with unlimited free viewers. The server is a single small container, it uses the PostgreSQL you probably already run, and SMS and voice go through your own Twilio account at carrier cost.
How much does the flat license save?
The price does not change as you add responders. Team works out at about $12.49 per user a month for 10 users, $6.25 for 20 and $2.50 for 50, so the bill stays predictable while per-user plans keep climbing. Team costs less than each plan below from the number of paid users in the last column.
| Compared with | List price | Team ($1,499 a year) costs less from |
|---|---|---|
| Opsgenie Essentials | $9.45 per user a month, annual | 14 users |
| Opsgenie Standard | $19.95 per user a month, annual | 7 users |
| Opsgenie Enterprise | $31.90 per user a month, annual | 4 users |
These figures count license fees at the list prices on our Sources list (checked November 2026) and ignore taxes and add-ons. Very small teams can start on a free plan elsewhere, such as PagerDuty Free for up to 5 users, and move to OnCallAlert as the rotation grows. The Core plan ($699 a year, up to 15 users) suits teams that alert by SMS and voice, and the mobile app comes with Team and above. Past 50 users, extra 10-user packs cost $300 each on yearly plans.
What does it cost to run?
OnCallAlert is one container plus PostgreSQL 16 or newer. In our deployments the server uses roughly half a vCPU and PostgreSQL about the same at typical on-call load, and most teams already have a PostgreSQL instance, so the database often adds no new machine. We have not published a benchmark, so size it for your own alert volume. You control upgrades, backups and placement: run it on infrastructure separate from the systems it watches.
How do you keep SMS and voice bills low?
You use your own Twilio account and pay Twilio directly, with no markup from us. In the US an SMS costs roughly $0.013 per segment with carrier fees and voice about $0.014 a minute, plus a monthly number fee (list prices, so check Twilio). Hosted tools meter SMS and voice too, so this cost exists everywhere; here you see it on your own bill. See our full-screen call versus direct call breakdown for the arithmetic.
The hybrid approach is what cuts it. When the responder's phone is online, the alert arrives as a push notification from the mobile app and rings through Do Not Disturb, so no call or SMS is sent. SMS and voice fire only as a fallback when the phone is offline or the alert is not acknowledged. Every page delivered by push is a call and two SMS segments you do not pay for, about $0.04 at the list prices above, and the saving grows with alert volume. Read how mobile push replaces most SMS and voice for details. Ring-through-Do-Not-Disturb works on Android today; Apple's Critical Alerts approval for iPhone is still pending.
Why does self-hosting matter for data privacy?
Your alert content, schedules and contact details stay on a server you control. The OnCallAlert server sends no telemetry about your alerts. Optional features call out: the license check, the push relay for mobile notifications, and the dead-man's switch heartbeat. The privacy page lists exactly what leaves your server.
What are the weak spots?
You run the server, so you own upgrades, backups and uptime. The importer does not move routing rules, heartbeats, maintenance windows or history, and there is no hosted Managed Cloud yet. Opsgenie has integrations OnCallAlert does not accept natively, and those senders must post JSON to OnCallAlert.
After the move
Install the mobile app and pair it for push alerts (the Team, Business and Unlimited plans include the hosted push relay). If you manage on-call as code, the Terraform, OpenTofu and Pulumi provider can take over from there (see infrastructure as code).
Frequently asked questions
Can I keep using Opsgenie until the end?
Yes. Per Atlassian, access works until April 5, 2027. Run both in parallel until you are confident, then switch off.
Does the importer change anything in Opsgenie?
No. It reads through a read-only key. Delete the key after the import.
What happens to my Opsgenie alert history?
It is not imported. Export what you need before April 5, 2027, when Atlassian deletes un-migrated data.
Can I migrate to Jira Service Management instead?
Yes. Atlassian offers that path. OnCallAlert can also import from Opsgenie data that already moved into Jira Service Management.
See what your team would pay. Put your own numbers into the calculator, then compare the plans.
Sources and notes (checked 2026-11)
- Atlassian: evolution of IT operations and moving Opsgenie data to Jira Service Management.
- Atlassian: Opsgenie pricing and end-of-sale notice and Jira Service Management pricing. Checked 2026-11. Prices are list prices per user or agent, billed annually; yearly figures are monthly price x users x 12.
- Time estimates are our own planning estimates, not measurements.
- Importer coverage from the migration docs.