Automatic Lambda Idle GPU Instance Alerts
Every hour, WebRun opens Lambda, lists every running instance with its type, region, and uptime, compares each against the expected life you set for that instance type, appends the hourly snapshot to Google Sheets, and messages you on WhatsApp about any machine still up long past when it should have been shut down.
How do I catch cloud GPU instances that are still running unused?
WebRun opens Lambda every hour and lists every running instance with its type, region, and uptime, comparing each against the expected life you set. It appends the snapshot to Google Sheets and messages you on WhatsApp about machines nobody shut down, so forgotten GPUs are terminated before the hours become a bill.
- Forgotten instances are named within the hour they overrun
- Monthly GPU hours accumulate in a sheet you already read
- Attached filesystems are flagged before anyone terminates a machine
Built for ML engineers · research teams · AI startups · infrastructure leads
What does WebRun do on every run?
The exact actions WebRun takes, in order - in plain language, so you can adjust anything.
-
WebRun signs in and gets to work
Opens
cloud.lambdalabs.comin a real browser with your saved login - no setup, no API keys. -
1
Lambda - list running instances and uptime
lambdalabs.com
WebRun opens Lambda to list running instances and uptime. - Open Lambda and list every instance currently running, with its name, instance type, region, and launch time
- Work out the uptime for each and note any attached filesystems that would persist after termination
- Compare each instance against the expected life you set for that type and mark the ones that have run past it
Done when Every running instance has an uptime, a type, and a verdict against its expected life.
-
2
Google Sheets - log hours and cost
google.com
WebRun opens Google Sheets to log hours and cost. - Append this hour's snapshot to your usage sheet with one row per running instance
- Record the instance name, type, region, hours up so far, and the cumulative hours for the month
- Total the hours by instance type so the heaviest spend is visible without opening the console
Done when Google Sheets holds this hour's running instances and the month's cumulative hours.
-
3
WhatsApp - flag forgotten machines
whatsapp.com
WebRun opens WhatsApp to flag forgotten machines. - Message the team group when an instance has run past its expected life, longest overrun first
- Name the instance, the type, the region, the hours up, and whether it has an attached filesystem
- Stay quiet when everything is inside its window. WebRun never terminates or stops an instance
Done when Every overrunning instance has been flagged in WhatsApp, or the hour passed quietly.
How is each run configured?
Secure by default
Connect once, stays signed in
WebRun signs in once and keeps each session in a persistent environment, so every run picks up right where it left off.
Every action is checked against this policy before it runs.
Questions, answered
Will it terminate an instance to save money?
No. WebRun only reads the instance list and reports what has overrun. Stopping or terminating a machine stays a human decision, so a long training run is never killed by an automation.
How does it know an instance has run too long?
You set an expected life per instance type, for example eight hours for an interactive box. WebRun compares each launch time against that window and flags anything past it.
Does it warn about data on the machine?
Yes. Each flagged instance notes whether it has an attached filesystem, so you know before shutting down whether anything persists after termination.
Related templates
Put this on autopilot.
Turn it on in minutes - or have our team set it up for you.