Handlers run on every host, even in ones without notified changes

[rmarchei]

Issue Type:

Bug Report

Ansible Version:

ansible 1.6.2

Environment:

Running from OS X 10.9.3
Managing RHEL 6.5, Fedora 20

Summary:

When a change is detected in a specific host handlers run on every host, even in ones without changes

@drybjed confirmed this bug and told me it is not present in 1.6.0 and it is related to this issue and commit

Steps To Reproduce:

Run a playbook like this:

---
- name: all hosts
  hosts: all
  user: root

  tasks:
    - name: foo
      copy: src=/tmp/foo.conf dest=/tmp/foo.conf owner=root group=root mode=0644
      notify:
        - restart ntpd
  handlers:
    - include: "../../handlers/restart_services.yml"

Expected Results:

handlers run only on notified hosts

Actual Results:

handlers run on every host in the playbook, even in ones without changes:

PLAY [all hosts] **************************************************************

GATHERING FACTS *************************************************************** 
ok: [host1.domain]
ok: [host2.domain]
ok: [host3.domain]

TASK: [foo] ******************************************************************* 
ok: [host1.domain] => {"changed": false, "dest": "/tmp/foo.conf", "gid": 0, "group": "root", "md5sum": "d41d8cd98f00b204e9800998ecf8427e", "mode": "0644", "owner": "root", "path": "/tmp/foo.conf", "secontext": "unconfined_u:object_r:admin_home_t:s0", "size": 0, "state": "file", "uid": 0}
ok: [host2.domain] => {"changed": false, "dest": "/tmp/foo.conf", "gid": 0, "group": "root", "md5sum": "d41d8cd98f00b204e9800998ecf8427e", "mode": "0644", "owner": "root", "path": "/tmp/foo.conf", "secontext": "unconfined_u:object_r:admin_home_t:s0", "size": 0, "state": "file", "uid": 0}
changed: [host3.domain] => {"changed": true, "dest": "/tmp/foo.conf", "gid": 0, "group": "root", "md5sum": "d41d8cd98f00b204e9800998ecf8427e", "mode": "0644", "owner": "root", "secontext": "unconfined_u:object_r:admin_home_t:s0", "size": 0, "src":     "/root/.ansible/tmp/ansible-tmp-1401171805.31-244401036446552/source", "state": "file", "uid": 0}

NOTIFIED: [restart ntpd] ******************************************************
changed: [host1.domain] => {"changed": true, "item": "", "name": "ntpd", "state": "started"}
changed: [host2.domain] => {"changed": true, "item": "", "name": "ntpd", "state": "started"}
changed: [host3.domain] => {"changed": true, "item": "", "name": "ntpd", "state": "started"}

host2.domain   : ok=3    changed=1    unreachable=0    failed=0
host1.domain : ok=3    changed=1    unreachable=0    failed=0   
host3.domain : ok=3    changed=2    unreachable=0    failed=0  

[mpdehaan]

This is quite unacceptable to us, we're adding a test for this specific thing and will update the release ASAP when ready once confirmed.

[jimi-c]

Thanks, I've reproduced this and will work on a fix today.

[jimi-c]

I've created the above patch to address this issue, if you'd like to test it yourself. I'm continuing testing on this, and will merge it in as soon as my testing is complete.

[drybjed]

Looks like patch works correctly. +1

[rmarchei]

patch tested, everything works fine with it!

[jimi-c]

Great, just merged this in. Please let us know if you have any further problems related to this issue. Thanks!