Migration from MinIO to SeaweedFS
This guide explains how to migrate Relution data from one object storage backend to another. Common scenarios include:
- Migrating from MinIO to SeaweedFS (MinIO is no longer maintained)
- Moving from cloud provider to self-hosted (or vice versa)
- Switching cloud providers
The examples below demonstrate a migration from MinIO to SeaweedFS, but the steps can be adapted for any S3-compatible source and destination.
The migration process consists of four steps:
- Set up the new storage backend
- Copy data from old to new storage
- Update Relution configuration
- Remove the old backend
Prerequisites
- Access to both storage backends simultaneously
- The
rclonecommand-line tool (installation guide →) - Credentials for both storage backends
Set Up the New Backend
If migrating to SeaweedFS, first follow SeaweedFS Setup →. Relution is stopped before the data is copied (see the section “Copy the Data”) and remains stopped until the configuration has been switched.
Ensure both old and new storage backends are running:
docker compose up -d minio seaweedfs
Configure rclone
The rclone tool provides a fast and reliable way to copy data between S3-compatible backends.
Install rclone
Follow the instructions at rclone.org/install → for the operating system in use.
Create Remote Configurations
Run the configuration wizard:
rclone config
Configure the source (MinIO):
- Type
nfor new remote - Name:
source - Storage type: Select
s3 - Provider: Select
Minio(orOtherfor generic S3) - Enter the access key and secret key
- Endpoint:
http://localhost:9000 - Accept defaults for remaining options
- Confirm with
y
Configure the destination (SeaweedFS):
- Type
nfor new remote - Name:
dest - Storage type: Select
s3 - Provider: Select
Other - Enter the new access key and secret key
- Endpoint:
http://localhost:8333 - Accept defaults for remaining options
- Confirm with
y
Type q to exit the configuration.
AES256 and the same Base64 key that is configured in Relution are set in the advanced configuration of both remotes (source and dest) (rclone options sse_customer_algorithm and sse_customer_key_base64). The new backend must support SSE-C.With client-side encryption via AWS KMS, the key data is stored in the object metadata. This metadata must be preserved during copying (rclone option
--metadata).Verify Port Access
rclone accesses the S3 APIs of both backends via the host (http://localhost:9000 and http://localhost:8333). After following Deprecated: MinIO → and SeaweedFS Setup →, the S3 ports are only reachable within the Docker network. For the duration of the migration, they are therefore published on the host in the compose.yaml:
minio:
ports:
- "127.0.0.1:9000:9000" # S3 API - for rclone during the migration
- "9001:9001" # Web console
# ...
seaweedfs:
ports:
- "127.0.0.1:8333:8333" # S3 API - for rclone during the migration
- "127.0.0.1:23646:23646" # Web console
Then recreate the containers with docker compose up -d minio seaweedfs. If rclone runs on a different host, the restriction to 127.0.0.1 is omitted. After the migration, the additional S3 port mappings are removed again.
Copy the Data
Stop Relution to prevent data changes during migration, and verify that only the storage backends are still running:
docker compose stop relution docker compose ps # Should show storage backends, but NOT relutionStart the transfer:
rclone copy source:relution dest:relution --progressWait for completion. Transfer time depends on the amount of data in the bucket.
Verify the transfer:
rclone check source:relution dest:relutionExpected output:
0 differences found
Update Relution Configuration
Update the Relution configuration to point to the new backend. Endpoint, access key and secret key change; bucket name and region only if they differ:
| Setting | Old Value | New Value |
|---|---|---|
Endpoint (customEndpoint) | http://minio:9000 | http://seaweedfs:8333 |
| Access Key | Old key | New key |
| Secret Key | Old secret | New secret |
Bucket Name (bucketName) | relution | Name of the destination bucket (only if it differs) |
Region (region) | Default eu-west-1 | Region of the destination bucket (e.g. for AWS S3) |
- If the destination bucket has a different name, the rclone commands use
source:<old-bucket> dest:<new-bucket>. The database only stores the object keys, so adjustingbucketNameis sufficient. - For AWS S3 without a custom endpoint,
customEndpointorRELUTION_STORAGE_S3_CUSTOMENDPOINTis omitted. The endpoint is then derived from the region. - The encryption settings (SSE-C key
serverSideEncryption.keyorkeyFile, or KMS) remain unchanged. - The destination bucket is used exclusively by this Relution instance. Relution regularly deletes objects in the bucket that are not assigned to one of its own resources.
See Connect object storage → for detailed instructions on where to make these changes.
Verify and Clean Up
Start Relution:
docker compose up -d relutionVerify functionality: Log into Relution and confirm that apps and resources load correctly.
Remove the old backend (only after thorough verification):
docker compose down minio # Stop the old serviceClean up configuration: Remove the old service from the
compose.yaml.Delete old data: Remove the old storage volume or directory (e.g.,
./volumes/minio/data).Clean up rclone: Optionally remove the rclone configuration:
rclone config delete source rclone config delete dest
Connection Refused / Timeout
Ensure the S3 API ports are exposed and not blocked by a firewall or network segmentation. Test connectivity:
curl http://localhost:8333 # Should return an S3 error
Access Denied / 401 / 403
Verify that the credentials are correct and the user has access to the bucket.
Missing Files After Migration
Re-run the verification:
rclone check source:relution dest:relution
If differences are found, run rclone copy again; it will only transfer missing files.