Renku session doesn't start: "Cannot clone repository"

Dear renkuers,

on renkulab, it is currently not possible to start a session from this project [Reproducible Data Science | Open Research | Renku]

The error message reads: “Cannot clone repository: error fetching submodules”

in git-clone.txt:

2022-06-20T12:21:12.485409919Z INFO:root:/work/msei-main does not exist, creating it.
2022-06-20T12:21:12.486090945Z INFO:root:Waiting for git to become available with timeout mins None...
2022-06-20T12:21:12.712535655Z INFO:root:Git is available
2022-06-20T12:21:12.713058435Z INFO:root:Checking if the repo already exists.
2022-06-20T12:21:12.722950285Z INFO:root:Intitializing repo with email gasserli@ethz.ch and name Lili Gasser
2022-06-20T12:21:12.777787949Z INFO:root:Cloning branch master

in init-certificates:

2022-06-20T12:13:10.793438680Z Purging old certificates
2022-06-20T12:13:10.800030706Z Updating certificates
2022-06-20T12:13:11.177809199Z WARNING: ca-certificates.crt does not contain exactly one certificate or CRL: skipping
2022-06-20T12:13:11.380513290Z De-reference symlinks

What do I need to do to get this running again?

Best,
Lili

@LiliGasser do you have a submodule in this project? I.e. another repository inside your project repository?

I do not have access to your project so I cannot check myself.

If you use submodules you probably have a .gitmodules file in your repository. And its contents list all of the other git repositories that are used as submodules

Hi Tasko,

I am not aware of a submodule in the project. There is also no .gitmodules file in the repository.

You now have access to the repo.

Could it be a problem that the name of the environment in environment.yml is not base?

Best,
Lili

@LiliGasser an improperly initialized module is here:
https://renkulab.io/gitlab/msei/msei-main/-/tree/master/src/pyclassyfire

I am trying to replicate the problem locally and see how you can fix it. This pyclassyfire seems to be a python package? is it acceptable to simply install it with git instead of including a clone of the repo for that package inside your repo?

In addition to this you definitely have large files that should be in git LFS but are not. When I clone your repository without fetching LFS files the repository is still 3.3 GB. But this is not the problem (for now). This just makes working with and navigating the repository very very slow.

Eliza has been working with pyclassyfire. She’s not (yet) on renku discourse. I forwarded her your questions and either she or I will get back to you.

@LiliGasser I was able to start a session from the cleanup-submodules branch simply by removing the following directory:
src/pyclassyfire/pyclassyfire_docs. That directory seems to be a submodule added to the pyclassyfire library.

I just created this branch to test if removing that would help and it does. I suspect you should be fine if you do this even in your master branch. But I leave this up to you. You know best when it comes to this.

The other thing I wanted to suggest is simply installing the pyclassyfire library directly from the git repository with pip.

You can do this by the command: pip install git+https://github.com/JamesJeffryes/pyclassyfire

You can simply add this to your Dockerfile and then you do not have to “carry around” a whole copy of this library in your repository.

Hi @tolevski

we removed the copy from the repo and now pip install it from the environment.yml file. It is again possible to start a session.

Thanks for your support!

Best,
Lili

1 Like