[BUG] - No zero downtime Deployment possible

1

So I have a ghost blog running - I used the exact image from render.com itself (https://render.com/docs/deploy-ghost).

I tested now multiple different settings one by one and the following happens:

  1. A deploy is triggered
  2. After render[.]com has created and uploaded the Docker image (see Deploy logs)
  3. Page goes “offline” or rather “down” with Internal Server Error 502
  4. Wait some more momenets (half a minute / minute)
  5. Reload
  6. Deployment succesful

This is really annoying since I am expecting 0-Downtime

2

hi @activenode, using Render Disks affects zero downtime deploys. We plan to remove this limitation.

However, being a CMS, Ghost doesn’t need to be deployed unless you’re upgrading Ghost itself. How frequently are you updating the image itself?

1 Like
3

Hi. Thanks for the response. I can follow-up why potentially you would do that but I can also very much see solutions for that e.g. checking the disk hashes if something cot corrupted in the meantime or like at least “readonly disk” which would already help the web-user to be able to use everything. This could e.g. inject a variable/lockfile whilst deploying the other instance but well you are the experts on this :slight_smile:

Besides that coming back to your question:

Well as given the Dockerfile does COPY . . and therefore allow to copy the custom theme into ghost. Sure thing one could upload the theme in ghost admin and therefore avoid a redeploy but that is not programmer friendly at all - especially in case the programmer would just be the service provider and not even have access to the CMS after all.

Also I am modifying Ghosts core code to fit my needs.

So tldr: I think recommending “dont redeploy” is not a good option since render should be my “deploy-whenever-you-want” provider :slight_smile: :smiley:

1 Like