How to fix this ssh-keyscan problem in my playbook?

I have an Ansible playbook for provisioning servers. After the server is provisioned, I want to remove the old FQDN and/or hostname alias for that server from my local SSH known_hosts file, pause until the server is up, and then run the ssh-keyscan command to add the new server’s public SSH key to my known_hosts file. However, my task for running the keyscan command is resulting in an error that I haven’t been able to figure out.

Here are my tasks:

(tasks that provision server run first here)

  • name: remove old server’s public key from known_hosts
    known_hosts:
    path: “{{ local_home_dir }}/{{ me }}/.ssh/known_hosts”
    name: “{{ item }}”
    state: absent
    loop:

  • “{{ hostname }}”

  • “{{ hostname }}.{{ domain_name }}”
    become: true
    become_user: “{{ me }}”

  • name: pause until host is up
    local_action: command ping -c 1 {{ hostname }}.{{ domain_name }}
    register: result
    until: result.rc == 0
    retries: 30
    delay: 5

  • name: add new server’s public key to local known_hosts file
    local_action: command ssh-keyscan -t ecdsa {{ hostname }}.{{ domain_name }} >> {{ local_home_dir }}/{{ me }}/.ssh/known_hosts
    become: true
    become_user: “{{ me }}”

The error can be seen here (I abbreviated the key that was returned with …)

TASK [add new server’s public key to local known_hosts file] ************************************************************
task path: /Users/smith/playbooks/gc/test.yml:36
Using module file /Users/smith/.virtualenvs/provision/lib/python3.7/site-packages/ansible/modules/commands/command.py
Pipelining is enabled.
ESTABLISH LOCAL CONNECTION FOR USER: smith
EXEC /bin/sh -c ‘/Users/smith/.virtualenvs/provision/bin/python && sleep 0’
changed: [127.0.0.1 → localhost] => {
“changed”: true,
“cmd”: [
“ssh-keyscan”,
“-t”,
“ecdsa”,
example.com”,
“>>”,
“/Users/smith/.ssh/known_hosts”
],
“delta”: “0:00:00.241807”,
“end”: “2020-01-22 14:50:09.699630”,
“invocation”: {
“module_args”: {
“_raw_params”: “ssh-keyscan -t ecdsa example.com >> /Users/smith/.ssh/known_hosts”,
“_uses_shell”: false,
“argv”: null,
“chdir”: null,
“creates”: null,
“executable”: null,
“removes”: null,
“stdin”: null,
“stdin_add_newline”: true,
“strip_empty_ends”: true,
“warn”: true
}
},
“rc”: 0,
“start”: “2020-01-22 14:50:09.457823”,
“stderr”: “getaddrinfo >>: nodename nor servname provided, or not known\r\ngetaddrinfo /Users/smith/.ssh/known_hosts: nodename nor servname provided, or not known\r\n# example.com:22 SSH-2.0-OpenSSH_7.9p1 Debian-10+deb10u1”,
“stderr_lines”: [
“getaddrinfo >>: nodename nor servname provided, or not known”,
“getaddrinfo /Users/smith/.ssh/known_hosts: nodename nor servname provided, or not known”,
“# example.com:22 SSH-2.0-OpenSSH_7.9p1 Debian-10+deb10u1”
],
“stdout”: “example.com ecdsa-sha2-nistp256 AAAAE2VjZHNhL…”,
“stdout_lines”: [
example.com ecdsa-sha2-nistp256 AAAAE2VjZHNhL…”
]
}
META: ran handlers
META: ran handlers

PLAY RECAP **************************************************************************************************************
127.0.0.1 : ok=3 changed=1 unreachable=0 failed=0 skipped=0 rescued=0 ignored=0

I know that the playbook’s pause task is working because I can see the server come up. I can also run the keyscan command that the playbook is running from the command line with no problems. It’s just the Ansible playbook that gets the “nodename nor servname provided, or not known error”. What’s more puzzling is that the stdout lines above show that the public key has been fetched from the server. Thanks!

I have an Ansible playbook for provisioning servers. After the server is provisioned, I want to remove the old FQDN
and/or hostname alias for that server from my local SSH known_hosts file, pause until the server is up, and then run the
ssh-keyscan command to add the new server's public SSH key to my known_hosts file. However, my task for running the
keyscan command is resulting in an error that I haven't been able to figure out.

Hello,

you can not use redirections with the command module. The shell module can do that. Another alternative would
be to fetch the SSH key file from the server instead of running ssh-keyscan.

Regards
        Racke

Racke, thank you very much. Changing ‘command’ to ‘shell’ fixed the problem. One follow-up… I thought ssh-keyscan was fetching the SSH key file from the server. Was I mistaken? What technique were you thinking of? Thanks again!