Suspending Non-Production Environments and Databases

Non-production Environments — dev, staging, demo, and any other
non-production stage — can be automatically suspended during
off-hours to reduce infrastructure costs, on both AWS and GCP.
Databases can be put on their own schedule in the same way.

A suspended Environment has its workloads scaled to zero — no pods
running, no compute costs. Resuming restores all workloads.

Scheduled suspension can reduce non-production infrastructure costs by up
to 40%.

Scheduling automatic suspension

Suspension is configured per application, on its Schedule tab in
SVIEW. All times are in UTC.

1. Open the Environment from the Dashboard.
2. Open the Schedule tab.
3. Set Schedule type to Cron (recurring) or Date (one-time).
4. Configure the SUSPEND row:
   • Every: day, or week on a chosen day or day range
   • At: the UTC time the Environment should suspend
5. Configure the RESUME row the same way.
6. Click Save.

The schedule takes effect within about a minute. There is nothing to
restart.

How the two times work together

The Environment is suspended for the whole period between a suspend time
and the next resume time — the suspend time does not have to recur on the
same days as the resume time. That is what lets a single schedule cover
both nights and full weekends.

Example — nights only. Suspend every day at 20:00, resume every day at 07:00: the Environment runs 07:00–20:00 UTC, every day including
weekends.

Example — nights plus the whole weekend. Set both rows to week on
MON-FRI — suspend at 21:30, resume at 04:30. Each weekday night the
Environment goes down at 21:30 and comes back at 04:30. No suspend is
scheduled on Saturday or Sunday, so after Friday night's suspend the
Environment simply stays down until Monday's resume.

Suspending or resuming immediately

The Override buttons on the same tab — Suspend now and Resume
now — take effect straight away and hold until you clear them.

• Suspend now keeps the Environment down regardless of the schedule.
• Resume now keeps it up regardless of the schedule — use it when you
  need to work outside the usual window.

Clear the override and save to hand control back to the schedule.

What happens during suspension

• All application workloads scale to zero — no running containers.
• Persistent storage remains intact — data is not lost.
• The Kubernetes control plane stays running — only workloads are
  suspended.
• Databases (RDS, Cloud SQL) remain running unless separately
  scheduled — see below.

Resuming typically takes 5–10 minutes from the resume time before all
Components are healthy — nodes have to start before pods can be
scheduled, and a suspended database takes a few minutes to accept
connections again.

Suspending databases separately

Individual database instances can also be put on their own suspend/resume
schedule, independently of the Environment they belong to — supported for
AWS RDS, GCP Cloud SQL, PostgreSQL, and MySQL.

1. Open the database instance in SVIEW.
2. Configure its own SUSPEND / RESUME schedule on the Schedule
   tab, the same way as for an Environment.

Production Environments

Automatic suspension should only be configured for non-production
Environments. If you accidentally suspend a production Environment, use
Resume now immediately or raise a request.

Related articles

• Key Concepts
• SVIEW — Overview
• Decommission an Application, Database or Environment — removing an Environment permanently instead