Add git ssh auth support

[ron1]

What problem are you facing?

Some environments rely on gith ssh based auth for private repositories instead of the currently supported token one.

The Upbound provider-terraform shares this problem as described here.

How could Crossplane help solve your problem?

Extend the current code around credentials setup with ssh support.

This code is similar to the code in provider-terraform around credentials setup.

[glennpratt]

If you'd like a monstrously terrible hack workaround:

{{- $now := now | unixEpoch -}}

---
apiVersion: ansible.crossplane.io/v1alpha1
kind: ProviderConfig
metadata:
  name: {{ $myName }}
spec:
  credentials:
  - filename: .git-credentials-{{ $now }}
    source: Secret
    secretRef:
      namespace: mynamespace
      name: git-deploy-secret
      key: sshPrivateKey
  vars:
    - key: GIT_SSH_COMMAND
      value: ssh -i $(find /ansibleDir -name '.git-credentials-{{ $now }}' -maxdepth 2 -mindepth 2 | head -n1) -o StrictHostKeyChecking=accept-new

[WintonHuang1893]

If you'd like a monstrously terrible hack workaround:

{{- $now := now | unixEpoch -}}

---
apiVersion: ansible.crossplane.io/v1alpha1
kind: ProviderConfig
metadata:
  name: {{ $myName }}
spec:
  credentials:
  - filename: .git-credentials-{{ $now }}
    source: Secret
    secretRef:
      namespace: mynamespace
      name: git-deploy-secret
      key: sshPrivateKey
  vars:
    - key: GIT_SSH_COMMAND
      value: ssh -i $(find /ansibleDir -name '.git-credentials-{{ $now }}' -maxdepth 2 -mindepth 2 | head -n1) -o StrictHostKeyChecking=accept-new

@glennpratt Thanks, but still no luck to me using GIT_SSH_COMMAND

  status:
    atProvider: {}
    conditions:
    - lastTransitionTime: "2024-01-29T03:39:47Z"
      message: |-
        connect failed: failed to install galaxy collections/roles: [WARNING]: - role-repo was NOT installed successfully: - command
        /usr/bin/git clone [email protected]:group/role-repo.git
        role-repo failed in directory /home/ansible/.ansible/tmp/ansible-
        local-5600490048dvjn/tmp4fc9p3aq (rc=128) - Cloning into 'role-repo'
        ... error: cannot fork() for ssh -i $(find /ansibleDir -name '.git-
        credentials-ssh' -maxdepth 2 -mindepth 2 | head -n1) -o
        StrictHostKeyChecking=accept-new: Resource temporarily unavailable fatal:
        unable to fork
        ERROR! - you can use --ignore-errors to skip failed roles and finish processing the list.
        No config file found; using defaults
        Starting galaxy role install process
        : exit status 1
      reason: ReconcileError
      status: "False"
      type: Synced

[glennpratt]

@WintonHuang1893

Resource temporarily unavailable fatal: unable to fork

That sounds like #304 which is fixed but not yet released. Restarting the deployment or deleting the pod would help if that is the issue.

[lukas-andrade]

Up!