Azure Database for PostgreSQL Flexible Server
Does ZopNight manage Azure Database for PostgreSQL Flexible Server?
PostgreSQL Flexible Server supports native stop/start, and non-production Postgres servers scheduled off nights and weekends routinely save 60-70 percent of their compute cost. ZopNight issues the stop through the Flexible Server API from the Ready state, watches a 60-day utilization lookback, and accounts for storage billing at full rate while stopped.
Rules that fire on Azure Database for PostgreSQL Flexible Server
At a glance
| Field | Value |
|---|---|
| Scheduling notes | stopped and started via the Flexible Server API, but only when the server is in the Ready state. |
PostgreSQL Flexible Server is the current managed PostgreSQL offering with native stop/start. Non-production Postgres servers scheduled off nights and weekends routinely save 60-70% of their compute cost.
Where the Postgres Flexible Server bill comes from
Three meters run on a Flexible Server: compute for the chosen vCore tier, provisioned storage, and backups beyond the included retention. Compute dominates on most non-production servers and is the only meter a stop affects. Storage keeps accruing whether the server is Ready or Stopped, which sets the floor cost for a scheduled server, a point worth checking before promising a savings number to anyone.
Stop/start mechanics and the Ready-state gate
The stop and start operations go through the Flexible Server API, and they only succeed when the server is in the Ready state. A server that is mid-restart, applying maintenance, or already stopped rejects the call. ZopNight honors that gate rather than retrying blindly, so a schedule fires cleanly when the server is eligible and skips when it is not. Azure also restarts a stopped Flexible Server after 30 days on its own; recurring schedules make that irrelevant, since the server cycles well inside the window.
What ZopNight automates for Postgres servers
Discovered via Azure Resource Graph. Azure Monitor metrics (60-day lookback) track utilization, Cost Management billing attributes spend, recommendations cover rightsizing and reservations, and schedules stop it outside working hours. ML auto-tagging covers this type, predicting environment tags that route servers into the right schedule automatically.
Postgres-specific waste to look for
Frequent offenders: high-availability configurations on development servers, which run a paired standby and roughly double compute for an environment nobody would page over; review and preview databases that outlive the branch they were created for; and vCore tiers inherited from a production template rather than sized for the actual dev workload.
Verifying a stopped Postgres server
Azure portal → Azure Database for PostgreSQL Flexible Servers → select the server. The Overview blade’s Status field reads Stopped once the schedule has run, and the compute meter stays quiet until the next start, while the storage line on the invoice carries on unchanged.