I'm trying to run ArgoCD on a local Rancher Desktop single node K3s cluster with a local Git repo

Versions:

• Rancher Desktop: 1.18.2
• Kubernetes: 1.31
• Argo CD: v3.1.0+9f37d43

The local Git repo is in this path on my laptop: /Users/me/srv/local/git/test-repo.git

Following a similar approach to that described here, I've first created a PV and PVC like below:

apiVersion: v1
kind: PersistentVolume
metadata:
  finalizers:
    - kubernetes.io/pv-protection
  labels:
    type: local
  name: local-git-pv
  namespace: argocd
spec:
  storageClassName: manual
  capacity:
    storage: 10Gi
  accessModes:
    - ReadWriteOnce
  hostPath:
    path: "/Users/me/srv/local/git"
    type: Directory
  persistentVolumeReclaimPolicy: Retain
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  labels:
    type: local
  name: local-git-pvc
  namespace: argocd
spec:
  storageClassName: manual
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 3Gi
  volumeName: local-git-pv

Then I installed argocd with the manifest from here, but first modified the argocd-repo-server Deployment descriptor, to include the mount in the argocd-repo-server container:

          volumeMounts:
          ...
            # Local git repo
            - mountPath: "/usr/local/share/git"
              name: local-git-volume
              readOnly: true
...
      volumes:
      ...
        # Local git repo
        - name: local-git-volume
          persistentVolumeClaim:
            claimName: local-git-pvc

And I created the below Argo CD Application:

apiVersion: argoproj.io/v1alpha1
kind: Application
metadata:
  name: test-app
  namespace: argocd
spec:
  destination:
    namespace: apps
    server: ''
  source:
    path: test-app
    repoURL: 'file:///usr/local/share/git/test-repo.git'
    targetRevision: ''
  project: default
  syncPolicy:
    automated:
      prune: true
      selfHeal: true

However, I got this error in the UI regarding sync:

Failed to fetch default: `git fetch origin --tags --force --prune` failed exit status 128: 
fatal: detected dubious ownership in repository at '/usr/local/share/git/test-repo.git' 
To add an exception for this directory, call: 
git config --global --add safe.directory /usr/local/share/git/test-repo.git 
fatal: Could not read from remote repository. Please make sure you have the correct access rights and the repository exists.

Based on the error message, I creaetd a ConfigMap to wrap a .gitconfig file:

apiVersion: v1
kind: ConfigMap
metadata:
  name: argocd-gitconfig
  namespace: argocd
data:
  .gitconfig: |+
    [safe]
        directory = *
    

And mounted it to the argocd-repo-server container:

          volumeMounts:
          ...
            - mountPath: /home/argocd/.gitconfig
              name: gitconfig
              subPath: .gitconfig
      ...
      volumes:
      ...
        - name: gitconfig
          configMap:
            name: argocd-gitconfig

This didn't fix the issue

When I exec into the argocd-repo-server Pod's container as the argocd user, the /tmp/_argocd-repo directory and its subdirectories that hold the cloned repo is initially not accessible. So I ran the following commands:

$ cd /tmp
$ ls -l _argocd-repo
ls: cannot open directory '_argocd-repo': Permission denied
$ chmod +r _argocd-repo
$ ls -l _argocd-repo
total 4
drwxr-xr-x 3 argocd argocd 4096 Mar 20 10:40 2f92a276-0180-47cc-9da5-8018b8effcda
$ cd _argocd-repo/2f92a276-0180-47cc-9da5-8018b8effcda
$ ls
$ git fetch origin --tags --force --prune
...
Unpacking objects: 100% (19/19), 2.12 KiB | 362.00 KiB/s, done.
From file:///usr/local/share/git/test-repo
 * [new branch]      main       -> origin/main

After a short time, the app appears as successfully synced in the UI, and I can see the application has been deployed ok

Subsequent changes to the repo and attempts to sync get the same error again. If I again exec into the Pod and manually run git fetch origin --tags --force --prune, the sync works ok

I'm not sure what is going on here. It looks like the .gitconfig is not being used in the argocd-repo-server's sync process

This is a local scratch type of lab, so I could add a script to manually run the sync commands each time I push a change

But, if anyone can advise on what is going on and how to fix the setup to avoid the manual work around, I'd be very grateful