Caching problem on OVH k8s cluster deployment

Hello everyone,

I tried to deploy the full binder stack starting from repository and even if it seems to globally work I still have a small problem.

Indeed the image cacheing doesn’t seem to work. It re-triggered a build everytime.

You can test and reproduce the problem here :

The configuration I used for that deployment is available here on my PR :

Does anyone has an idea about what I missed ?


1 Like

The bit of code that checks the image registry for an existing image is called from:

and the code itself is:

My guess would be that something with checking the registry isn’t quite right. I think there was an issue previously with how the name that is looked up in the registry is formatted or something. The point being that we have had things break around the registry handling before.

The other thing to double check (maybe even first) is if your BinderHub is indeed configured to use a docker registry.

1 Like

Indeed the problem was on my configuration key binderhub.config.BinderHub.image_prefix which need to be in 3 parts separated by / for the cache to work properly.

Example of value that was not working :

  • (.*)/binder-r2d-f18835fd-

Example of new value that work :

  • (.*)/binder-ovh/r2d-f18835fd-

Thank’s for your help Tim !
This issue can be close