Sign in Start free trial
MODULES / SCHEDULER

Lights out for idle cloud.

The Scheduler stops dev, test and staging resources outside working hours and starts them before your team logs in. EC2, RDS, ECS and EKS on AWS, plus Google Cloud and Azure VMs, on one schedule screen.

Start your 30-day free trialWhat it schedules
See it work

Watch the Scheduler and cm-bot at 2 AM

What it schedules

Six kinds of resources, one schedule

Pick the resources, the days and the stop and start times. Each scheduler handles the details of its service, so a database, a container service and a node group are each switched off the right way.

EC2 instances

ALL PLANS
  • Stops and starts instances in every region of the account.
  • Stop or start one by hand from the dashboard; the page follows the state until AWS settles.
  • Take single instances off the schedule without touching the rest.
  • Shows who stopped each instance: you, a teammate or cm-bot.

RDS and Aurora

ALL PLANS
  • Standalone databases and whole Aurora or Multi-AZ clusters.
  • Read replicas, Aurora Serverless v1 and Multi-AZ SQL Server can't be stopped, so they're shown but not selectable.
  • AWS starts a database again after 7 days stopped; the next scheduled stop switches it off again.

ECS services

ADD-ON $19/MO
  • Set a task count for working hours and a lower one for off hours.
  • Services with auto scaling have it suspended at night and resumed in the morning. Your min and max are never changed.
  • "Set tasks" scales a service right now.

EKS node groups

ADD-ON $29/MO
  • Managed node groups get a working-hours and an off-hours node count.
  • At night min, desired and max are lowered so Cluster Autoscaler can't add nodes back. The original sizes are saved and restored.
  • "Set nodes" resizes a node group right now.

Google Cloud VMs

ALL PLANS
  • Compute Engine VMs in the project you connect.
  • GKE nodes, managed instance group members and VMs with local SSDs are left alone.

Azure VMs

ALL PLANS
  • Virtual machines are deallocated, not just stopped, so compute billing stops.
  • Scale-set VMs are left alone.
How a schedule runs

Predictable, and it stays out of your way

  • Checks every minute. When a schedule moves from working hours to off hours, cm-bot stops what's on it. In the morning it starts it again.
  • Acts on changes only. If someone starts a machine at 2 a.m. to fix something, cm-bot leaves it running until the next scheduled change.
  • Safe first run. A new schedule inside its off hours stops right away, but never starts anything you had switched off.
  • Retries on its own. A failed cloud call is retried every minute, up to 5 times, and the failure shows on the dashboard.
  • Pause any time. One click pauses a schedule; nothing is stopped or started until you resume.
  • Full history. Every stop, start, approval and change is in History with who did it.

Weekday nights and weekends off

runningoff

Stop at 19:00, start at 07:00, Monday to Friday. Off all weekend. That's 108 of 168 hours.

1 minbetween schedule checks
5×retries before it reports a failure
cm-bot suggestsFROM 28 DAYS OF DATA

Stop 19:00 · start 07:00 · Mon–Fri + weekends

14EC2 instances
3RDS databases
$1,240saved a month
Summary · by cm-bot

Your staging and QA machines are idle every night after 19:00 and all weekend. Two analytics instances run batch jobs at 02:00, so they're left out of this schedule.

Example suggestion.

Schedule suggestions

Not sure which hours? Let cm-bot look.

"Scan my account" reads 28 days of hourly CloudWatch metrics and finds the hours each resource sits idle.

  • Idle means idle. An EC2 hour with CPU under 5% and network under 5 MB, an RDS hour with no connections, an ECS hour under 5% CPU, EKS nodes under 10% CPU.
  • Every schedule tried. Stops from 17:00 to 01:00, starts from 05:00 to 10:00, any days. It keeps the one that saves the most, counting only resources idle for all of its off hours.
  • Explained in plain words, then applied in one click. You can change it first.
  • More you can schedule. The dashboard also lists machines, databases, containers and VMs in your connected clouds that aren't on a schedule yet, with the saving for each.
WhatsApp approvals

cm-bot asks your team before it powers off

Pick a WhatsApp team for each scheduler. Ten minutes before a stop, everyone in the team gets the list of machines that are about to go off.

  • Approve all, or choose machines from a checklist in the chat.
  • Remind me in 1 hour when someone is still working. The power-off moves back an hour.
  • Cancel keeps everything running until the next scheduled night.
  • A report afterwards. Successes come together; each failure comes on its own, with the reason.
  • Opt-in. Only teammates who tapped Join get messages. Works with EC2, RDS, Google Cloud and Azure schedules.
cm-botCloud Modules
⏻ 3 machines power off in 10 minutes (schedule “Nights & weekends”):
  1. web-staging-1 · EC2
  2. api-dev · GCP VM
  3. reports-db · RDS
18:50
Savings

Work out your saving

Move the sliders to your numbers. The estimate assumes weekday nights 19:00–07:00 and weekends off.

Estimated saving per month$3,571108 h off a week · 4.35 weeks a month
Connect

Connected in minutes, with no keys to hand over

You create the access in your own account and can remove it at any time. It can list, stop and start, and nothing else.

AWS

  • Launch one CloudFormation stack. Your External ID is filled in for you.
  • Each scheduler gets its own role, so you choose what cm-bot can touch.
  • Deleting instances is off unless you turn it on in the stack.

Google Cloud

  • Run one command in Cloud Shell and paste back the line it prints.
  • It creates a service account with a custom role. We impersonate it: no service account keys exist.

Microsoft Azure

  • Run one command in Azure Cloud Shell and paste back the line it prints.
  • It adds a custom role on the subscription: read VMs, start, deallocate.

Someone else runs your cloud account? Each setup page has a PDF guide with your connection ID, every step and the exact permissions, ready to send.

Plans

What's included

SchedulerFree trial (30 days)Business $49/mo
EC2Up to 20 instancesUnlimited
RDS and AuroraUp to 10 databasesUnlimited
Google Cloud and Azure VMsIncludedIncluded
ECS servicesConnect and previewAdd-on, $19 a month
EKS node groupsConnect and previewAdd-on, $29 a month
CloudsOne of AWS, GCP or AzureAll three, up to 5 AWS accounts
WhatsApp approvals1 team5 teams
History and reports30 days1 year
FAQ

Questions teams ask

What if someone needs a machine at night?

Start it from the dashboard or your cloud console. cm-bot acts only when the schedule changes phase, so it stays on until the next scheduled stop. With WhatsApp approvals, anyone in the team can postpone or cancel tonight's power-off.

Will it ever delete anything?

No. The Scheduler stops, starts, deallocates and scales. Deleting EC2 instances from the dashboard is possible only if you turn it on in your CloudFormation stack, and it asks you to type the instance name.

What happens to disks and IP addresses?

Stopped instances keep their disks. Public IPs that aren't Elastic IPs can change when an instance starts again, so use Elastic IPs or DNS for anything that needs a fixed address.

Which time zone do schedules use?

Each schedule has its own time zone, so a team in Tel Aviv and a team in New York can each have their own nights.

What happens when the trial ends?

Schedules stop running until you choose a plan. Nothing in your cloud is changed or removed.

Cloud Modules professional services

Need hands-on help? Our DevOps, SecOps and cloud architecture engineers can do it with you.

Professional servicescloud-modules.com ↗

Switch the lights off tonight.

Start your free trialNext module: Attack surface map