Good Morning, I've recently deployed the argo workflows terraform system. Tried to run the first flow and get a lot of these errors

Error downloading package from cache for '_libgcc_mutex-0.1-conda_forge':
...
Error: Could not fetch packages -- see pretty-printed errors above.

The environment is

@conda_base(libraries={}, python="3.11")

I guessing some permission didn't get setup correctly which might be related to my question above. Thanks for any help hope the talk tour is going well 🙂

@hundreds-zebra-57629 may be in a better spot to help out here 🙂

@hundreds-zebra-57629 @square-wire-39606 I got it working!! But I think I found a potential bug. So originally I used this environment

@conda_base(libraries={}, python="3.11")

and got the error

Error downloading package from cache for '_libgcc_mutex-0.1-conda_forge':
...
Error: Could not fetch packages -- see pretty-printed errors above.

I noticed in the S3 bucket that no conda_env has been cached. Even though the original output seemed to indicate that an environment had been cached. (not sure lost those logs) So I modified the env adding pandas

@conda_base(libraries={"pandas": ""}, python="3.11")

and it ran correctly. I should mention I'm using the netflix-extension package manager.

interesting. paging @dry-beach-38304 too

Just to confirm, you are saying that specifying a conda_base environment with no libs and just the python version causes an error? I can try to repro (out this week but can check this weekend/next week but confirming what the issue is).

yes thats correct.

Ok. Does it still do that? There should be no difference between the two (ie: some packages are alwasys added). I’ll try to repro but wonder if it was a problem a while ago with uploading stuff.

I'm working through another issue with @hundreds-zebra-57629 on the terraform deployment. Perhaps this is tied to that.