Limit How Many Versions to Go Back When Upgrading From Older Version of Gitpod

[jimmybrancaccio]

Is your feature request related to a problem? Please describe

This evening I've attempted to update a Gitpod instance I have but that has been offline for a few months. This instance is using Unstable. It appears when I shut it down it was at release-2022.04.1.2. Bringing it back online this evening I believe it was running preflight checks on all releases since then up until main.3791` - or at least it attempted too. Unfortunately it looks like Docker has rate-limited me so I am unable to deploy the latest version. In fact I can't deploy any version since some of our pods use images hosted at Docker Hub so I have to wait 6 hours. 😫

failed to pull: failed to write common midstream: failed to find private images: failed to list upstream images: failed to walk upstream dir: failed to check if image 2 of 2 in "daemonset.yaml" is private: failed to retry: reading manifest 1 in docker.io/library/busybox: toomanyrequests: You have reached your pull rate limit. You may increase the limit by authenticating and upgrading: https://www.docker.com/increase-rate-limit

Describe the behaviour you'd like

I cannot think of any reason why we'd need to pull ~100 versions and preflight check them. We should likely only pull the last ~20 or so releases and only run pre-flight checks on the latest release. If someone wants to pre-flight check the other versions they could do so manually. There's no sense in wasting the bandwidth, resources and limited Docker pulls if you're likely only going to deploy the latest version anyways.

We could also allow administrators to supply their Docker Hub credentials which should allocate them more pulls.

Describe alternatives you've considered

Wait until I'm not rate-limited so I can deploy the latest version. It won't let me deploy it even sans the preflight checks because it still needs to download.

Additional context

N/A

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[jimmybrancaccio]

This happened to me the other week again. I booted up my server and had ~400 versions between what I had installed and the latest version. I ended up shutting down my cluster so it wouldn't case rate limiting at Docker.

I would say this is a low priority because I am probably one of the few using the Unstable channel, but it would be very helpful to have an option to skip these pre-flight checks, or limit the amount of versions that pre-flight checks get performed on. I'd even be happy if there was some sort of workaround.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.