"no migration found for version" Error Encountered After Using the Restore API to Restore Terraform Enterprise – HashiCorp Help Center

Problem

When using the Restore API endpoint to restore a backup of a Terraform Enterprise (TFE) installation, the application fails to start and reports the following error in the logs.

[ERROR] tfe-task-worker: error running database migrations: err="migrate: no migration found for version X: read down for version X migrations: file does not exist"

Cause

This error occurs when the backup was created from a different version of Terraform Enterprise than the version you are restoring to. This commonly happens when attempting to roll back to a previous version of Terraform Enterprise after an upgrade.

Solution

To resolve this issue, you must manually rename the existing PostgreSQL database. This action forces Terraform Enterprise to create a new, clean database on the next startup, allowing the restore operation to complete successfully.

Prerequisites

The postgresql-client package must be installed on the Terraform Enterprise host to run pg_dump and psql commands.
- Debian/Ubuntu:
```
$ sudo apt-get install postgresql-client
```
- RHEL/CentOS:
```
$ sudo yum install postgresql-client
```
You will need your PostgreSQL connection details (PG_USERNAME, PG_PASSWORD, PG_NETLOCATION, PG_DBNAME). You can retrieve these values by running the following command.
```
$ replicatedctl app-config export --hidden
```

Procedure

Stop the Terraform Enterprise application.
```
$ replicatedctl app stop
```
Monitor the application status until it reports a stopped state.
```
$ watch replicatedctl app status
```

Create a local backup of the current database.

$ pg_dump postgres://<PG_USERNAME>:<PG_PASSWORD>@<PG_NETLOCATION>:5432/<PG_DBNAME>?sslmode=require > tfe_pgdump.backup

Connect to the PostgreSQL server using a default database like postgres.

$ psql postgres://<PG_USERNAME>:<PG_PASSWORD>@<PG_NETLOCATION>:5432/postgres?sslmode=require

Rename the current Terraform Enterprise database by appending _old to its name.

ALTER DATABASE "CurrentDBName" RENAME TO "CurrentDBName_old";

If this command fails due to active connections, terminate them first and then retry the rename command.

SELECT pg_terminate_backend (pg_stat_activity.pid) \
FROM pg_stat_activity \
WHERE pg_stat_activity.datname = 'CurrentDBName';

Create a new, empty database with the original name.
```
CREATE DATABASE "OriginalDBName";
```
Exit the psql session.
```
exit
```
Start the Terraform Enterprise application.
```
$ replicatedctl app start
```
If the application fails to start, review the application logs for further errors before proceeding.
Execute the API request to the restore endpoint again to restore the backup.
Restart the application to apply the restored data.
```
$ replicatedctl app stop
$ watch replicatedctl app status
## Wait for the state to become 'stopped'
$ replicatedctl app start
```
If the application fails to start after the restore, review the application logs for further errors.
Verify that the data was restored successfully by logging into the Terraform Enterprise UI.

Optional Cleanup

After confirming a successful restore, you can remove the old database and the local backup file.

Connect to the PostgreSQL server again.

$ psql postgres://<PG_USERNAME>:<PG_PASSWORD>@<PG_NETLOCATION>:5432/postgres?sslmode=require

Drop the old database.
```
DROP DATABASE "DatabaseName_old";
```
Exit the psql session.
```
exit
```
Delete the local database backup file.
```
$ rm tfe_pgdump.backup
```

Outcome

After completing the procedure, Terraform Enterprise starts successfully with all data from the backup restored.