Troubleshoot databases

Fix database connection errors, failed authentication, full storage or memory, failed password changes and restores, greyed-out actions, and the API’s database error messages.

Most database problems come from the connection details, the database’s status, or its size. Start with the database’s Overview tab: the status should be Ready, not Suspended or a failed status, and the Connection card shows the details clients must use. If a problem persists after the fixes on this page, contact support with the database’s ID (on the Settings tab).

An application cannot connect

Typical errors: could not translate host name ... to address, Unknown MySQL server host, getaddrinfo ENOTFOUND, Connection refused.

Cause	Fix
The application has no internal connection to the database	On the database’s Networking tab, add an internal connection to the application.
The application is in another location	Internal connections work only within one location. Move one of them, or use the external connection.
The environment variables are missing or stale	Check DB_HOST, DB_PORT and DB_URL (or the REDIS_ variables) on the application’s Environment variables tab, then deploy the application: new variables reach it only with a deployment.
The application uses the external hostname	Use the internal hostname and port from the Internal connection details.
The database is suspended or not ready	Activate it, or wait until its status is Ready.

A client outside Sevalla cannot connect

Typical errors: Connection refused, Connection timed out, Can't connect to MySQL server, Could not connect to Redis.

Cause	Fix
The external connection is off	On the Networking tab, enable it. Suspending the database turns it off.
The client uses a legacy hostname	Legacy hostnames of the form <cluster>.proxy.sevalla.app, with ports from 30000 to 30999, were disabled on Oct 15, 2026. Use the hostname and port the dashboard shows now.
The client uses the internal hostname	Hostnames ending in svc.cluster.local work only inside Sevalla. Use the External details.
IP restrictions block the client	On the Networking tab, check IP restrictions and allow the client’s public IP address.
A firewall blocks the port	The external port isn’t a standard database port. Allow outbound TCP to the external hostname and port on the client’s network.

Authentication fails

Typical errors: password authentication failed for user "orders" (PostgreSQL), Access denied for user 'orders'@'...' (using password: YES) (MySQL, MariaDB), WRONGPASS invalid username-password pair or user is disabled. or NOAUTH Authentication required. (Redis, Valkey).

• The password changed. Copy the current one from the Overview tab, and update the application’s DB_PASSWORD and DB_URL (or REDIS_PASSWORD and REDIS_URL), then deploy it. Changing the password doesn’t update connected applications. See Change the database password.
• The user or database name is wrong. Copy both from the connection details; they never change after creation.
• A hand-written URL breaks on special characters in the password. Copy the URL from the dashboard, which encodes them.

PostgreSQL asks for SSL

Error: server does not support SSL, but SSL was required. PostgreSQL databases in Google Cloud locations don’t accept TLS connections. Set the client’s SSL mode to prefer or disable, for example ?sslmode=prefer at the end of the connection URL. To keep traffic off the internet, connect applications with an internal connection instead.

Connections drop

Typical errors: Lost connection to MySQL server during query, MySQL server has gone away, server closed the connection unexpectedly, Connection reset by peer.

• A restart, resize, password change or restore is running: each makes the database unavailable for a while. Check the status and wait until it’s Ready.
• The database ran out of memory and restarted. Check Memory usage on the Analytics tab. MySQL and MariaDB on DB1 often run out of memory under load: move to DB2 or larger.
• Your client keeps connections open for a long time. Use a connection pool that checks connections before use and reconnects.

Storage or memory is full

Typical errors: could not extend file ... No space left on device (PostgreSQL), The table '...' is full (MySQL, MariaDB), OOM command not allowed when used memory > 'maxmemory'. (Redis, Valkey).

Storage and memory come with the database’s size. Move to a larger size on the Settings tab. To free storage meanwhile, delete data you don’t need. For Redis and Valkey, set expiries on cache keys so they don’t fill the memory.

The status shows a failure

Status	What to do
Error	Restart the database. If it stays in Error, contact support.
Password change failed	The old password still works. Try changing it again.
Restore failed	The data may be incomplete. Check the reason in the restore history, then restore again or restore another backup.
Deletion failed	Contact support.

A button is greyed out

Point at the button to see why. The reasons are:

Reason shown	What to do
Wait for the current change to finish.	A resize, restart, password change, restore or external connection change is running. Wait until the status is Ready.
Activate the database first. (or Activate the database to ...)	The database is suspended. Activate it.
Create an application first.	Add internal connection needs an application in the company.
A missing permission	Ask a company owner or admin for access. See Users and roles.

In Add internal connection, applications marked Already connected, Suspended or In another location can’t be chosen.

A backup or restore failed

• A manual backup didn’t appear. A database keeps at most 5 manual snapshot backups, and a backup target at most 5 manual dump backups. Delete one, then create the backup again.
• The target database isn’t in the list. The list shows only compatible databases: same location and engine, the exact version for snapshots (the backup’s version or newer for dumps), status Ready or Restore failed, not suspended. See Which databases can receive a restore.
• The restore was rejected because of disk size. A snapshot restores at the source’s full storage size, so the target needs at least as much storage. Move the target to a larger size first.
• There’s no backup to restore. A database in an OCI location has backups only in the backup targets you add. See External backups.

The studio cannot connect

Failed to connect to the database means the studio couldn’t reach the database. Check that its status is Ready and it isn’t suspended, then click Try again. The studio isn’t available yet appears while a new database is still being set up.

API error messages

Message	Meaning
No valid payment method found. Add one at https://app.sevalla.com/billing/payment-methods to continue.	Creating or resizing needs a valid payment method.
The selected location is not available for this company	The location isn’t on your company’s cloud. Use a location from List clusters.
Database "<id>" is not in a ready state	Another change is running, or the status doesn’t allow the action. Wait until it’s ready.
Database "<id>" has a backup operation in progress, try again shortly	A backup is being created or deleted. Try again in a few minutes.
Source and target services are deployed to different regions	An internal connection needs both services in the same location.
Internal connection between "<id>" and "<id>" already exists	The two services are already connected.
Restore target is incompatible: ...	The target database doesn’t meet the restore rules; the message says which.
Database size cannot be downgraded	The new size has less memory or storage than the database has now.
Backup target limit reached, a database can have at most 5 backup targets	Delete a backup target before adding another.
Manual backup limit reached, at most 5 manual backups are kept per backup target	Delete a manual dump backup from that target first.