Health checks

A health check tells Plexivo App Runner when an instance is ready. Traffic goes only to instances that pass it, so deploys, restarts and scale-outs never send visitors to an app that is still starting.

Two kinds of check#

Use an HTTP path for anything that serves HTTP. It catches an app that is running but stuck.

Set an HTTP health check#

1. Open the app’s Settings tab. In Health check, enter the HTTP path, for example /healthz. Leave it empty to go back to the port check.
2. Apply. Choose Apply changes. Plexivo restarts the instances one at a time, and each new one gets traffic once the check answers.

A path starts with /, has no spaces and is at most 256 characters.

curl -s -X PUT $API/deployments/$APP/health-check \
  -H "Authorization: Bearer $PLEXIVO_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"path": "/healthz"}'
# {"health_check": {"path": "/healthz", "kind": "http"}}

Send {"path": null} to return to the port check.

Timing#

• Plexivo checks every second.
• A new instance has up to 5 minutes to pass its first check before it is restarted.
• An instance that fails for 6 seconds stops receiving traffic until it answers again.
• With an HTTP path, an instance that has not answered for a minute is restarted.

A good health endpoint#

• Answers fast, with 200, and does no heavy work.
• Checks what the app cannot serve without, such as its database connection, but not other services that have their own checks.
• Needs no sign-in.

app.get('/healthz', (req, res) => res.status(200).send('ok'));

Next steps#