Skip to main content

Maintenance & Upgrades

Version Upgrades​

Minor and patch versions are available as in-place upgrades. Major version upgrades (e.g. PostgreSQL 16 → 17) are performed as new service deploys with migration.

Check available versions​

curl -u admin:password https://api.foundrydb.com/database-versions?type=postgresql

Upgrade a service​

curl -u admin:password -X POST \
  https://api.foundrydb.com/managed-services/{id}/upgrade \
  -H "Content-Type: application/json" \
  -d '{"target_version": "17.3"}'

The service stays online during minor version upgrades on multi-node setups:

  1. Replicas are upgraded one at a time
  2. A replica is promoted to primary
  3. The old primary is upgraded and rejoins as a replica

Expect one brief failover (under 60 seconds) during the final promotion step.

Single-node services have a brief downtime during the upgrade restart.

Stop and Start​

Stop a service​

Stopping shuts down the database process. The VM and data disk remain intact. You are not billed for compute while stopped, but storage billing continues.

curl -u admin:password -X POST \
  https://api.foundrydb.com/managed-services/{id}/stop

Start a service​

curl -u admin:password -X POST \
  https://api.foundrydb.com/managed-services/{id}/start

The service will return to Running within a few seconds for most engines.

Patching​

Operating system security patches are applied automatically during a maintenance window. This involves a brief VM restart.

The default window is Sunday 02:00–04:00 UTC. To request a different window, contact support.

Multi-node services patch one node at a time, so the service remains available.

Planned Failovers​

Before performing maintenance on a primary node, trigger a manual failover to avoid unplanned downtime:

# Get replica node ID
curl -u admin:password https://api.foundrydb.com/managed-services/{id}/nodes \
  | jq '.nodes[] | select(.role == "replica") | .id'

# Trigger failover
curl -u admin:password -X POST \
  https://api.foundrydb.com/managed-services/{id}/nodes/{replica_id}/failover

DNS updates within ~10 seconds. Connection strings do not need to change.

Deleting a Service​

Deletion is permanent and removes all data, backups, and DNS records.

curl -u admin:password -X DELETE \
  https://api.foundrydb.com/managed-services/{id}

If you need to keep backups, restore them to a new service before deleting.

Engine-Specific Notes​

PostgreSQL​

  • Minor upgrades (e.g. 17.1 → 17.3) use pg_upgrade in-place
  • Extensions are preserved across upgrades
  • max_connections changes take effect after a restart

MySQL​

  • 8.0 → 8.4 upgrades require schema compatibility check (run mysqlcheck first)
  • GTID replication is maintained across rolling upgrades

MongoDB​

  • Feature compatibility version (FCV) is stepped up after all nodes are upgraded
  • Drivers with replica set awareness reconnect automatically after failover

Valkey​

  • Upgrades use binary replacement; AOF/RDB files are compatible across minor versions
  • Sentinel continues operating during rolling upgrades

Kafka​

  • Broker version is updated one at a time; partition leadership rebalances automatically
  • Consumer groups are not disrupted during rolling restarts