Install rpm from control machine to ansible managed hosts

[satheesh kumar]

I tried to install rpm in ansible control machine to target machines using playbook

- name: Install local rpm in server
  yum: pkg=files/mod_jk-1.2.40-5.el6.x86_64.rpm state=present
  when: ansible_os_family == 'RedHat'


I got error :
failed: [xyz.cloudapp.net] => {"changed": false, "failed": true, "rc": 0, "results": []}
msg: No Package file matching 'files/mod_jk-1.2.40-5.el6.x86_64.rpm' found on system

When I copied the same file in server and installed as local installation it works.


      - name: copy mod_jk rpm to server
        copy: src=files/mod_jk-1.2.40-5.el6.x86_64.rpm dest=/home/user/mod_jk-1.2.40-5.el6.x86_64.rpm owner=user group=azureuser mode=755
        when: ansible_os_family == 'RedHat'

      - name: Install mod_jk local rpm in server
        yum: name=/home/user/mod_jk-1.2.40-5.el6.x86_64.rpm state=present
        when: ansible_os_family == 'RedHat'


Is there a way to directly install from control machine without copying it

[Brian Coca]

no, currently the yum module only operates with files already on the
remote machine.


--
Brian Coca

[satheesh kumar]

Thanks !

[Sam Hartsfield]

I'm installing packages from the control machine, but it's a bit of work (so it probably won't be worth it for just a few packages), and I'm still figuring out the details.

Basically, I create a local yum repository, then temporarily run an HTTP server on the control machine (maybe with Ansible 2.0 I can reliably start and kill it as part of the playbook, but right now it's a manual process); for example using "python -m SimpleHTTPServer 8000"

I set the host to the control server using the SSH_CLIENT environment variable (if there's a better way to access that, I'd like to know):

yum_local_host: "{{ (ansible_connection is defined and ansible_connection == 'local') | ternary('localhost', ansible_env.SSH_CLIENT.split()[0]) }}"
yum_local_port: 8000
yum_repo_base: "http://{{ yum_local_host }}:{{ yum_local_port }}"

Then I use templates to set up the repositories on the remote side. Currently I'm looking at using a custom yum.conf so that it doesn't interfere with the normal configuration.

If anyone wants more details, or has suggestions for how to improve this setup, let me know.

[Reza Ghafari]

Is this still the case in ansible 2.2? The module document is not really clear.