CODEWITHSUNDEEP
node.jsgoogle-app-enginegoogle-cloud-platformslack
Jim Factor
Jim Factor

Reputation: 1535

Node.js on GCP App Engine with B1 Instance restarts every 15 minutes

I currently have a slack bot running on Google App Engine B1 Instance with manual scaling and instances set to 1.

I have read other reports of people having their instances restart and they mention the scaling type or they make a kind of keep alive function that will request an App URL every minute or so. I have been using manual_scaling which I've read will try to keep your app running indefinitely, but I am seeing regular restarts (every 15 minutes).

app.yaml

runtime: nodejs16 # or another supported version

instance_class: B1

manual_scaling:
  instances: 1

env_variables:
  SLACK_BOT_TOKEN: "tokenhere"

Is there a configuration or setting I'm missing somewhere that is causing the frequent reboots?

down every 10-15 minutes

Edit:

It worked 3 hours after I deleted extra app.yaml config while asking the question. Seems as though the extra config settings I had in my app.yaml were overriding the default behavior of B1 manual_scaling. When I removed the yaml code shown below the instance stayed up for 3 hours before eventually restarting again every 10-15 minutes.

code removed from app.yaml

handlers:
- url: /stylesheets
  static_dir: stylesheets

- url: /.*
  secure: always
  redirect_http_response_code: 301
  script: auto

enter image description here

Upvotes: 0

Views: 262

Answers (1)

Jim Factor
Jim Factor

Reputation: 1535

edit: Fix was only short term

It got fixed for 3 hours while I was writing the question. Seems as though the extra config settings I had in my app.yaml were overriding the default behavior of B1 manual_scaling. When I removed the yaml code shown below the instance stayed up for 3 hours before eventually restarting again every 10-15 minutes.

app.yaml

handlers:
- url: /stylesheets
  static_dir: stylesheets

- url: /.*
  secure: always
  redirect_http_response_code: 301
  script: auto

edit2: Switching to an F1 instance and doing the keep alive strategy of requesting its own URL every minute seems to be working.

Upvotes: 1

Related Questions