How to prevent git clone timeouts in playbook?

[Robert F]

I have a task in my Ansible playbook that uses the git module to clone my Git repository which is hosted on bitbucket.com to my Linode-hosted web server.  The playbook is run from my local laptop.  My problem is that this task seems to timeout quit frequently.  When it does, I'll get the following error:

    fatal: [web02.example.com]: FAILED! => {... "msg": "Cloning into '/srv/http/example.com/repo'...\nHost key verification failed.\r\nfatal: Could not read from remote repository.\\n\nPlease make sure you have the correct access rights\nand the repository exists", "rc" 128,...}

I'm pretty certain this is a timeout error because I haven't changed my SSH keys, the access rights are correct, and the repository does exist.  Furthermore, if I run the entire playbook again it will sometimes succeed and other times I'll have to re-run it a couple of times before it succeeds.

Here's the task:

    # webservers.yml

    - name: check out dev version of django project repo

      git: >

        repo={{ repo_url }}

        version=dev

        dest={{ repo_path }}

        accept_hostkey=yes

      become: true

      become_user: "{{ me }}"

Here's my config file:

    # ansible.cfg

    [defaults]

    inventory = inventories/development

    remote_user = smith

    roles_path = $HOME/playbooks/roles:$HOME/playbooks/community/roles

    vault_password_file = vault_pass

    retry_files_enabled = False

    retry_files_save_path = "/tmp"

    host_key_checking = False

    # Try to prevent git clone from timing out

    ssh_args = -o ControlPersist=360s

    [ssh_connection]

    pipelining = True

I read that setting ssh_args as shown above might prevent this problem but it doesn't seem to help.  Is there something else I can do to prevent these timeouts so that I don't have to keep re-running my playbook to get my repo cloned?  Thanks.

[J Hawkesworth]

Well you could use ansible blocks to retry if the first attemp fails

http://docs.ansible.com/ansible/playbooks_blocks.html

I would be tempted to change things around somewhat and spin up an ansible instance 'near' the other machines you want to manage (i.e. same datacenter/cloud/network - you get the idea), then you reduce likelihood of timeouts in the first place.

Hope this helps,

Jon

[Robert F]

Thanks Jon.  I was thinking the same thing, to set up a small server near my other machines solely for build purposes in order to minimize latency.  I'll also take a look at blocks.