[2.0] wait_for fails with timeout

[tbondarchuk]

Playbook (wait_for task example straight from documentation):

    - name: Reboot
      command:
        "shutdown -r now 'Rebooted by Ansible'"
      async: 0
      poll: 0
      ignore_errors: true

    - name: Wait for server's sshd to start
      local_action:
        wait_for port="22" host="{{ ansible_host }}" search_regex="OpenSSH" delay=60

result:

TASK [Reboot] ******************************************************************
changed: [test1]

TASK [Wait for server's sshd to start] *****************************************
fatal: [test1 -> localhost]: FAILED! => {"changed": false, "elapsed": 300, "failed": true, "msg": "Timeout when waiting for search string OpenSSH in localhost:22"}

looks like 'wait_for' is checking localhost instead of remote host.
Removed 'local_action' from task, fails immediately with:

TASK [Wait for server's sshd to start] *****************************************
fatal: [test1]: UNREACHABLE! => {"changed": false, "msg": "ERROR! SSH Error: data could not be sent to the remote host. Make sure this host can be reached over ssh", "unreachable": true}

[bcoca]

This is working as expected:
[test1 -> localhost] <= this means the wait_for has been delegated to 'localhost' on behalf of 'test1', using host={{ansible_host}} should point the check 'from localhost to test1'.

That it is hitting the default timeout (300) indicates that the test1 host was not up in that time, increase the timeout if it is just taking the host longer to boot and get ssh up and running.

Closing this as it is not a bug, please post to IRC or the mailing lists first as you can probably get help much faster:

• IRC: #ansible on irc.freenode.net
• mailing list: https://groups.google.com/forum/#!forum/ansible-project

[tbondarchuk]

Guess my description was really poor..

The issue is, host is up and accessible long before timeout ends, every time I've run this test.

Here is timeout error on version 1.9.4, when host is indeed not accessible:

TASK: [Wait for server's sshd to start] ***************************************
failed: [test1 -> 127.0.0.1] => {"elapsed": 301, "failed": true}
msg: Timeout when waiting for search string OpenSSH in test1:22

And here is error message from example above:
"msg": "Timeout when waiting for search string OpenSSH in localhost:22"

[bcoca]

can you use debug on ansible_host? you might be wanting to use inventory_hostname

[tbondarchuk]

Run more tests with following playbook:

    - name: show host
      debug:
        msg: "{{ ansible_host }}"

    - name: Reboot
      command:
        "shutdown -r now 'Rebooted by Ansible'"
      async: 0
      poll: 0
      ignore_errors: true

    - name: Wait for server's sshd to start
      local_action:
        wait_for port="22" host="{{ ansible_host }}" search_regex="OpenSSH" delay=60

First result was the same:

TASK [show host] ***************************************************************
ok: [test1] => {
    "msg": "192.168.1.13"
}

TASK [Reboot] ******************************************************************
changed: [test1]

TASK [Wait for server's sshd to start] *****************************************
fatal: [test1 -> localhost]: FAILED! => {"changed": false, "elapsed": 120, "failed": true, "msg": "Timeout when waiting for search string OpenSSH in localhost:22"}

But then I've looked at inventory, and yes it was set to use deprecated variable :(
test1 ansible_ssh_host=192.168.1.13

changed it to new variable test1 ansible_host=192.168.1.13 and finally all worked:

TASK [show host] ***************************************************************
ok: [test1] => {
    "msg": "192.168.1.13"
}

TASK [Reboot] ******************************************************************
changed: [test1]

TASK [Wait for server's sshd to start] *****************************************
ok: [test1 -> localhost]

Looks like variable ansible_host is substituted for deprecated ansible_ssh_host everywhere but wait_for task.

Then I've run slightly more complicated playbook:

    - name: show host
      debug:
        msg: "{{ ansible_host | default(inventory_hostname,boolean=True) }}"

    - name: Reboot
      command:
        "shutdown -r now 'Rebooted by Ansible'"
      async: 0
      poll: 0
      ignore_errors: true

    - name: Wait for server's sshd to start
      local_action:
        wait_for port="22" host="{{ ansible_host | default(inventory_hostname,boolean=True) }}" search_regex="OpenSSH" delay=60

with inventory without ansible_host at all:
192.168.1.13

Got error again:

TASK [show host] ***************************************************************
ok: [192.168.1.13] => {
    "msg": "192.168.1.13"
}

TASK [Reboot] ******************************************************************
changed: [192.168.1.13]

TASK [Wait for server's sshd to start] *****************************************
fatal: [192.168.1.13 -> localhost]: FAILED! => {"changed": false, "elapsed": 120, "failed": true, "msg": "Timeout when waiting for search string OpenSSH in localhost:22"}

So wait_for seems to work only with strictly declared ansible_host but not with any other variable.

[edrzmr]

this really works for me.

ansible: 2.1.1.0

[vinaykonanki]

If you use same variable name (ansible_ssh_host or ansible_host) in both inventory and module then it is working fine or else wait_for will take it as localhost.

I tested with both ansible 2.1.1.0 and 2.1.0.0.

[shaiksulaiman]

I set the exact host IP / hostname to wait_for task to avoid confusion.

My snippet below. Hope it helps someone

• name: Set facts
  set_fact:
  node_ip: "{{ ansible_host | default(inventory_hostname) }}"

• name: Print node IP
  debug:
  msg: "node ip is {{ node_ip }}"

• name: Verify the Node reachability
  wait_for:
  port: 893
  #host: "{{ ansible_host | default(inventory_hostname) }}"
  host: "{{ node_ip }}"
  delay: 90
  timeout: 300
  delegate_to: localhost

[jlebon]

The tricky part here is that inside a wait_for task, the ansible_host variable becomes localhost, which is most definitely not what you want (you can verify this by running with -vvv). What we ended up doing was saving the value of ansible_host beforehand and using that:

- set_fact:
    real_ansible_host: "{{ ansible_host }}"

- name: restart hosts
  shell: sleep 3 && shutdown -r now
  async: 1
  poll: 0
  ignore_errors: true

- name: wait for hosts to go down
  local_action:
    wait_for host={{ real_ansible_host }}
    port=22 state=absent delay=1 timeout=120
  become: false

- name: wait for hosts to come back up
  local_action:
    wait_for host={{ real_ansible_host }}
    port=22 state=started delay=30 timeout=120
  become: false

[chrowe]

Looks like #21269 was filed for the above issue.

[revmischa]

I'm getting this issue too on EC2