To roll back a Coolify application, open the application in the dashboard, choose Rollback, select a retained image under Available images, and choose Roll back to this image. Then inspect the deployment logs and verify the domain, health check, and a critical user flow. This operation changes the application image only; it does not undo database migrations, restore persistent files, or revert external services.
Antihost offers VPS plans with one-click Coolify, NVMe storage and a 14-day free trial.
What a Coolify rollback actually changes
Coolify’s rollback feature deploys an older application image that is still retained on the server. The image contains the application code and other image contents from an earlier build. Coolify starts that image using the application’s current runtime configuration. The official documentation defines the boundary precisely: “A rollback deploys an older application image retained on the server. It does not reverse database migrations, restore persistent storage, or change external services.” See Coolify’s rollback documentation.
As an Amazon Associate I earn from qualifying purchases.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →That distinction matters during an incident. A container can start successfully while the application remains incompatible with a newer database schema, changed environment variable, altered mount, queued job, or object-storage state. Treat an image rollback as one part of recovery, not as a complete release reversal.
What is included
- The application code packaged in the selected older image.
- Image-level dependencies and files baked into that image.
- A new deployment using the selected image and the application’s current Coolify runtime settings.
What is not included
- Database rows, schema migrations, extensions, or stored procedures.
- Persistent volumes and files mounted outside the image.
- Queues, object-storage keys, webhooks, DNS, payment settings, and other external services.
- An image that has already been deleted by cleanup.
If the failed release changed any of those systems, prepare a separate restore or forward-fix procedure before selecting Rollback.
#1 Best Overall
Check these conditions before you start
A rollback is safest when the target image exists, the current runtime can run it, and persistent state remains compatible. Make these checks before triggering the deployment.
Confirm that the target image is available
Open the application’s rollback screen and inspect Available images. The rollback selector is limited to commit and pull-request image tags; the image currently running cannot be selected. If the expected revision is absent, choose Reload images and check again. Coolify cannot recreate an arbitrary image that has already been removed from the server.
Recommended Free Tools
Review image-retention settings
Rollback targets exist only while their images are retained. In the application settings, inspect Image retention and the Images to keep value. Also check whether a server-level retention setting overrides the application value. Cleanup removes images outside the retention policy, and a removed image no longer appears as a local rollback target. The retention behavior and selector limits are documented at coolify.io/docs/applications/deployments/rollbacks.
Compare the old image with today’s runtime configuration
Rollback uses the current environment variables, mounts, ports, health-check configuration, and other runtime settings rather than restoring the configuration that existed when the older image was built. Compare those settings with the older release’s requirements. A variable renamed in the failed release, a removed mount, or a changed command can prevent an otherwise valid image from starting.
Assess database and persistent-data compatibility
Identify every migration or data-writing change in the failed release. An older application may not understand a newer column, enum value, index, serialized payload, or background-job format. Confirm that the schema is backward compatible, or use a tested database restore or migration procedure. Do not assume that starting the previous container reverses a migration.
Preserve incident context
Record the failed deployment, target image or commit, operator, reason, and any data-recovery action. Keep those details linked to the rollback deployment. A later forward fix should add to the history rather than obscure the release that caused the incident.
Dashboard procedure: roll back an application image
- Open the application. In the Coolify dashboard, select the affected application and open its deployment controls.
- Choose Rollback. The rollback view lists images retained locally for that application.
- Refresh the inventory when necessary. Under Available images, select Reload images if the expected image is not immediately listed.
- Identify the working revision. Use the image tag, commit, pull-request context, and deployment history to select the last known-good image. Do not select the image that is currently running; Coolify excludes it from the rollback selector.
- Start the rollback. Select the target image and choose Roll back to this image. Coolify queues a deployment using that image and the application’s present runtime configuration.
- Open Deployment Logs. Follow the operation through image selection, configuration generation, container actions, and health checks.
- Verify the result. Check the public application domain, container and application logs, health-check status, and at least one critical user flow such as sign-in, checkout, save, or a core game action.
Do not declare success solely because the container is running. A healthy process can still serve errors, reject requests because of schema changes, or fail only when a user reaches a path that the health check does not exercise.
How to choose among the dashboard, API, and CLI
All three interfaces queue a deployment operation, but they suit different operators and identify targets differently.
| Interface | Access pattern | Target identifier | Useful evidence after the action |
|---|---|---|---|
| Dashboard | Interactive operator workflow | A retained image shown under Available images | Deployment Logs, application status, health check, and manual user-flow verification |
| API | Authenticated automation or an incident tool | The required commit JSON field, containing the image tag or commit to deploy |
The returned deployment_uuid, followed by log and health verification |
| CLI | Terminal-based operation or scripted runbook | The application UUID, with optional --commit for a Git commit SHA |
CLI output plus the corresponding Coolify deployment logs |
The comparison is based on Coolify’s rollback, API, CLI, and deployment-log documentation: rollback guide, API endpoint, CLI reference, and deployment overview.
Rolling back through the API
The authenticated endpoint is:
POST /api/v1/applications/{uuid}/rollbackReplace {uuid} with the application UUID. The request must include a JSON field named commit. Coolify defines that field as the image tag or commit to roll back to. A successful response includes a deployment_uuid, which you should retain for incident tracking and subsequent log inspection. The endpoint’s exact contract is documented at https://coolify.io/docs/api/endpoints/applications/rollback-application-by-uuid.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsPOST /api/v1/applications/{uuid}/rollback
Content-Type: application/json
Authorization: Bearer <your-authenticated-token>
{
"commit": "<retained-image-tag-or-commit>"
}The placeholder values above are syntax examples: use the real application UUID, authentication method, and retained target supported by your Coolify installation. Treat the returned deployment UUID as an operation identifier, not proof that the application is healthy. Follow the deployment in Coolify’s logs and perform the same domain, health-check, and critical-flow checks used for a dashboard rollback.
Rolling back with the CLI
The documented command is:
coolify app rollback run <uuid>Here, <uuid> is the application UUID. The command also accepts an optional --commit argument for a Git commit SHA:
coolify app rollback run <uuid> --commit <git-commit-sha>CLI flags can change between releases. Run the installed command’s help and treat that output as authoritative for the exact CLI version in use. The command reference is https://coolify.io/docs/cli/command-reference/commands/app/rollback-run.
After the command returns, locate the deployment it created in Coolify, open its logs, and verify the application externally. A successful command means Coolify accepted or queued the operation; it does not establish that the old image built, started, passed health checks, or serves correct data.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Reading deployment logs when something fails
Coolify deployment logs record the selected source revision, image build or pull activity, generated configuration, and container actions. Start at the first failing command rather than the last cascading error. The deployment overview explains the log flow at https://coolify.io/docs/applications/deployments/overview.
Failure during source or image build
A failed source build does not replace the running application. Correct the build problem, choose another retained image, or deploy a fixed revision. Do not interpret a build failure as evidence that production has already switched images.
Failure after the image is built
Failures during configuration generation, container replacement, startup, or health checks depend on the replacement path and health-check settings. Compare the selected image’s expected ports, start command, environment variables, mounts, and readiness behavior with the current application configuration.
Health check passes but users still fail
Check application logs and exercise a critical user path manually. Health checks often cover only a narrow endpoint; they may not reveal authentication, database-query, queue, payment, or asset failures. Keep the rollback deployment and the failed release associated in the incident record.
Database migrations and persistent data
Rollback does not undo a migration. If the failed release added a column, changed a constraint, rewrote data, or advanced a migration table, those changes remain after the older image starts. The same applies to files in persistent volumes, queue messages, object-storage content, and data held by external services.
Use an explicit data-recovery plan
- Determine whether the schema remains readable by the target image.
- For backward-incompatible changes, use a tested database backup restore or a forward migration that makes the schema compatible again.
- Restore persistent files from the appropriate backup or snapshot rather than expecting the container image to contain them.
- Reconcile queues, webhooks, and external-service state separately; an application image cannot retract requests already sent elsewhere.
Run destructive data operations only under the procedures appropriate to your database and service providers. Coolify’s rollback feature is not a database restore mechanism.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why a previous deployment may be missing
The image was cleaned up
The most common explanation is retention. Review Images to keep under Image retention and check for a server-level override. Once cleanup removes an image, it disappears from the local rollback list and cannot be selected by the dashboard.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →The image list is stale
Use Reload images on the rollback screen, then inspect the deployment history and logs to confirm which tags actually exist locally.
The target is outside the supported selector
Coolify’s rollback selector supports commit and pull-request image tags, while the running image is not selectable. A revision that does not meet those tagging conditions will not be offered through that selector.
The deployment was never retained locally
A deployment record can exist in history even when its image has been removed. In that case, you need a separately available image or a new deployment of the desired revision; the rollback control cannot reconstruct a deleted local image.
Best Value
Rollback versus downgrading Coolify itself
Rolling back an application and downgrading the self-hosted Coolify control plane are different operations. A Coolify downgrade does not roll back applications, databases, services, managed servers, or their data. The control-plane procedure is covered at https://coolify.io/docs/core/instance-management/downgrade.
Free tools Windows power users keep installed
One-click scans. No signup required.
Before changing the Coolify version, Coolify advises creating an instance backup, reviewing release notes, checking active deployments, and considering database compatibility. If the downgrade fails, restoring a known-good instance backup may be required. Keep this plan separate from an application-image rollback: one changes the management platform, while the other deploys an older application image.
A practical incident checklist
- Identify the failed release and the last known-good image or commit.
- Confirm the target image is listed, or reload the image inventory.
- Check image-retention settings and any server-level override.
- Review current environment variables, mounts, commands, ports, and health checks against the older image.
- Assess database-schema and persistent-data compatibility.
- Prepare database, file, queue, and external-service recovery actions if the release changed them.
- Choose the dashboard, API, or CLI path appropriate to the operator and audit requirements.
- Record the target, operator, reason, and resulting deployment UUID or deployment record.
- Watch Deployment Logs from the first command through container replacement and health checks.
- Verify the public domain, logs, health-check status, and a critical user flow.
- Keep the rollback and failed-release records linked until the forward fix is complete.
Designing releases so rollback remains safe
Rollback is most reliable when releases are deliberately reversible. Retain enough images to cover the period in which a defect is likely to be discovered, and monitor whether server-level retention policies shorten that window. Use expand-and-contract database changes: add structures before the code needs them, deploy compatible readers and writers, and remove old structures only after older images are no longer needed.
Keep runtime configuration changes under the same compatibility discipline as image changes. If a release requires a new variable, mount, port, or health-check behavior, make the older image tolerate the interim configuration or document the exact restoration step. For queues and external APIs, design idempotent operations and versioned payloads so an older application does not misinterpret work created by a newer one.
Finally, test the rollback path itself. Verify that the expected image remains selectable, that the operator can find the deployment logs, and that backups for databases and persistent storage are restorable. A rollback plan that depends on an image already removed by cleanup is not an available recovery plan.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
