max_failed_percentage exit 0

[dominis]

max_failed_percentage always sends a non zero exit status even if the percentage is not exceed.

---
- hosts: app-pool
  gather_facts: False
  user: dominis
  serial: 32
  max_fail_percentage: 50

  tasks:
    - action: ping

dominis@shuriken:~/work/temp$ ansible-playbook a.yml -f 32

PLAY [app-pool] ***************************************************************

TASK: [ping] ******************************************************************
ok: [app01.bfc.kinja-ops.com]
fatal: [app10.bfc.kinja-ops.com] => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
ok: [app08.bfc.kinja-ops.com]
ok: [app02.xyz.kinja-ops.com]
ok: [app11.xyz.kinja-ops.com]
ok: [app13.xyz.kinja-ops.com]
ok: [app02.bfc.kinja-ops.com]
ok: [app12.xyz.kinja-ops.com]
ok: [app15.bfc.kinja-ops.com]
ok: [app05.xyz.kinja-ops.com]
ok: [app09.bfc.kinja-ops.com]
ok: [app16.bfc.kinja-ops.com]
ok: [app04.xyz.kinja-ops.com]
ok: [app06.xyz.kinja-ops.com]
ok: [app09.xyz.kinja-ops.com]
ok: [app11.bfc.kinja-ops.com]
ok: [app14.bfc.kinja-ops.com]
ok: [app14.xyz.kinja-ops.com]
ok: [app15.xyz.kinja-ops.com]
ok: [app16.xyz.kinja-ops.com]
ok: [app07.bfc.kinja-ops.com]
ok: [app03.xyz.kinja-ops.com]
ok: [app12.bfc.kinja-ops.com]
ok: [app13.bfc.kinja-ops.com]
ok: [app05.bfc.kinja-ops.com]
ok: [app03.bfc.kinja-ops.com]
ok: [app07.xyz.kinja-ops.com]
ok: [app10.xyz.kinja-ops.com]
ok: [app04.bfc.kinja-ops.com]
ok: [app01.xyz.kinja-ops.com]
ok: [app06.bfc.kinja-ops.com]
ok: [app08.xyz.kinja-ops.com]

PLAY RECAP ********************************************************************
           to retry, use: --limit @/Users/dominis/a.retry

app01.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app01.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app02.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app02.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app03.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app03.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app04.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app04.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app05.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app05.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app06.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app06.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app07.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app07.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app08.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app08.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app09.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app09.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app10.bfc.kinja-ops.com    : ok=0    changed=0    unreachable=1    failed=0
app10.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app11.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app11.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app12.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app12.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app13.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app13.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app14.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app14.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app15.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app15.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app16.bfc.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0
app16.xyz.kinja-ops.com    : ok=1    changed=0    unreachable=0    failed=0

dominis@shuriken:~/work/temp$ echo $?
3

I don't think this should be the same as this:

---
- hosts: app10.bfc.kinja-ops.com
  gather_facts: False
  user: dominis

  tasks:
    - action: ping

dominis@shuriken:~/work/temp$ ansible-playbook a.yml -f 32

PLAY [app10.bfc.kinja-ops.com] ************************************************

TASK: [ping] ******************************************************************
fatal: [app10.bfc.kinja-ops.com] => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue

FATAL: all hosts have already failed -- aborting

PLAY RECAP ********************************************************************
           to retry, use: --limit @/Users/dominis/a.retry

app10.bfc.kinja-ops.com    : ok=0    changed=0    unreachable=1    failed=0

dominis@shuriken:~/work/temp$ echo $?
3

ansible version:

dominis@shuriken:~/work/temp$ ansible-playbook --version
ansible-playbook 1.4 (devel 41d382b36d) last updated 2013/10/09 12:14:59 (GMT +200)

[mpdehaan]

Hi Nandor,

In the above you have set max_failure_percentage to 50% and you have had one failure.

So yes, it's doing the right thing in going on -- but I can see what you mean about having a nice way to make sure this failure shows up in Jenkins.

Perhaps we should set a global flag and if we would have returned 0 and that flag is set, we can return a new exit code that we have previously unused (like, I believe, 4) and print a message at the bottom.

This is a very good idea, so I'm tagging this as a priority feature request.

[dominis]

Hey Michael,

thanks for the quick response!

[mpdehaan]

Ordinarily speaking we should return non-zero if there are any failures, so this seems more like max_fail just eating the error code.

But yes, agree that there should be a code here.

[mpdehaan]

Reclassifying this as a bug report so we can investigate and close this out.

[jctanner]

@dominis I did find a bug in how the failed percentage was calcuated when serial is set, which is now fixed by 5b3b9ba

http://docs.ansible.com/playbooks_delegation.html#maximum-failure-percentage
'' it may be desirable to abort the play when a certain threshold of failures have been reached"

Max fail percentage was only intended to abort the play if the threshold was reached, but has never been a way to alter the final exit code and override detected host failure/unreachable. We are going to keep that behavior as is, because a change could be very confusing for users who rely on the exit code 0 to mean a perfect execution.