search

gush-labs / kazan

VueJS web-app to practice Japanese
https://gush-labs.github.io/kazan/
GNU General Public License v3.0
1 stars 1 forks source link

Create a separate pipeline for production deployment #108

Closed VadimGush closed 1 year ago

VadimGush commented 1 year ago

I have a suspicion that the main cost of running Kazan in the cloud will come from the deployment itself (it seems like every time Cloud Function is updated, Google Cloud uses Compute Engine to build a new image which leads to additional costs). So it might be reasonable to create a separate deployment pipeline that will be triggered only manually when we will want.

bill876 commented 1 year ago

do you pay for a deploy of a lambda for some reason? do you wanna discuss how this all works in gcp?

VadimGush commented 1 year ago

@bill876 Looking at the billing, its expenses a little bit higher than I calculated. The only explanation for that might be Google's use of Compute Engine for building new images... But I'm might be wrong. I think it will be better to wait until the end of the month and analyze billing once more.

bill876 commented 1 year ago

@VadimGush can you explain in more detail? what is your estimate? how did you get it? how much is the actual bill? what happens in gcp when you deploy a cloud function? where is the code to do that (i dont see it in the repo)?

available for a call or to meet irl also

VadimGush commented 1 year ago

@bill876 I have found the issue. Cost of 0.6$ per day comes from Load Balancer (which somewhere around 15$ per month). It is not Cloud Functions or anything related to them.

bill876 commented 1 year ago

This is exactly why it makes no sense to deploy any cloud functions for a small app unless you have a free tier for a load balancer. For google's 0.025$/hr I can get 4 VMs with a "well-known" provider. Or 3 VMs for free with a certain less known one :)