What is the difference between "Cool down period" and "Initial delay" in the "Instance Groups" in the Google Cloud Platform (GCP)?

Question

I set up group parameters in the Instance Groups in the Google Cloud Platform.

I have the Autoscaling parameter set to On.

I wonder what is the relationship between the two other parameters Cool down period and Initial delay:

1. I would also like to know how I should choose these parameters, if in my case the virtual server is created in less than 10 seconds, and services start in less than 5 seconds?

2. What exactly do they relate to traffic monitoring for creating or deleting virtual machines in a group and the time of creating a new instance and the time of starting services on the created instance?

3. I wonder, however, whether, for example, these values sum up when creating a new machine or not, and if so in what situations?

Anticipating the answers: yes, I read the descriptions posted on the screen ;)

Thank you in advance for all your comments and help.

Answer 1

The first item that you will need is how long does it take for your instance to go from power up to a ready state. If you have Stackdriver logging setup, details on start activity are logged. Otherwise, use the serial port output to review startup times. I wrote an article on this for Debian.

I would also like to know how I should choose these parameters, if in my case the virtual server is created in less than 10 seconds, and services start in less than 5 seconds?

Choose the time it takes for your instance to go from startup to ready state and add a fudge factor. How much, depends on what activities are in your startup. If you are dependant on an external source to download and install a package, you have no control over how busy their site is, etc. As a general rule, you do not need to have great precision. If the startup varies from 5 to 10 seconds, use 30 seconds as your number.

What exactly do they relate to traffic monitoring for creating or deleting virtual machines in a group and the time of creating a new instance and the time of starting services on the created instance?

See my answer to your next question.

I wonder, however, whether, for example, these values sum up when creating a new machine or not, and if so in what situations?

No, these numbers do not sum up. They are used for different purposes. The first is how long to wait until Google starts collecting information such as CPU usage, etc. (think metrics and alerts). You do not want monitoring alarms going off while an instance is booting or your monitoring graphs show 10 minutes of zero activity during startup.

The second number is for health checks. Depending on how your systems are setup, a failure here could result in the instance being terminated and a new one launched. If your number are too small you could see an instance start, get terminated, a new one launched, get terminated ...