After updating (pip uninstall, then re-install) and running an adhoc command, say
ansible -m ping all
I’m seeing many pages of info messages, of the form
task 1: [Continuously incrementing list of hosts]
ok: HOST
ok: HOST
ok: HOST
eventually followed by the output I’m used to seeing.
This seems to be true for playbooks too (they seem to execute fine).
Am I missing a new configuration variable
Has anyone else seen this?
To clarify, an ansible -m ping produces a very large amount of progress notes.
Each host generates far greater output
A “task 1:” header, with the host and a list of all previous hosts processed. The list starts at one host, and by the end includes all the hosts. This is followed by several “ok: [HOSTNAME]” (with other tasks there’s more detail)
By the end, there’s a long list of hosts in the “task 1:” line.
Then the end has several repetitions of the v2.9.X expected summary output
Also should note that I listed the collection version in the subject, ansible-core is at 2.10.0
I’ve tried modifying ANSIBLE_STDOUT_CALLBACK (in the config and as an environmental variable). FWIW the callback names change in 2.10 (longer qualified names).
It smells like some new output modifying feature that doesn’t make the transition for my setup, but also seems to have something to do with parallelism.
I’ve tried several versions of python, no change.
I’ve tried to keep fairly current with ansible installs, but this behavior makes the current version quite unattractive.
Functionality seems the same.
I have not seen this, and I have no idea what could cause this. Your output (although you only gave a snippet which makes it difficult) appears to be from the default callback, however when running ansible (the adhoc command) we explicitly prevent using the default callback.
Might help to see the output of ansible-config dump --only-changed
As well as some pastebin of the full output, including -vvvv
When run with -vvvv you should see a line that starts with:
Loading callback plugin
Thanks
ansible-config dump --only-changed
(I’m running a directory based config, clearly)
ANSIBLE_PIPELINING(/home/fortier/src/ansible/ansible.cfg) = True
CACHE_PLUGIN(/home/fortier/src/ansible/ansible.cfg) = jsonfile
CACHE_PLUGIN_CONNECTION(/home/fortier/src/ansible/ansible.cfg) = .tmp/facts
DEFAULT_CALLBACK_WHITELIST(/home/fortier/src/ansible/ansible.cfg) = [‘skippy’, ‘unixy’, ‘actionable’, ‘yaml’, ‘debug’, ‘dense’]
DEFAULT_FORKS(/home/fortier/src/ansible/ansible.cfg) = 75
DEFAULT_HOST_LIST(/home/fortier/src/ansible/ansible.cfg) = [‘/home/fortier/src/ansible/inventory’]
DEFAULT_LOAD_CALLBACK_PLUGINS(/home/fortier/src/ansible/ansible.cfg) = True
DEFAULT_LOG_PATH(/home/fortier/src/ansible/ansible.cfg) = /home/fortier/src/ansible/.ansible.log
DEFAULT_ROLES_PATH(/home/fortier/src/ansible/ansible.cfg) = [‘/home/fortier/src/ansible/roles’]
DEFAULT_STDOUT_CALLBACK(/home/fortier/src/ansible/ansible.cfg) = skippy
DEFAULT_TIMEOUT(/home/fortier/src/ansible/ansible.cfg) = 30
DEFAULT_VAULT_PASSWORD_FILE(/home/fortier/src/ansible/ansible.cfg) = /home/fortier/src/ansible/.ansible_vault_password
INTERPRETER_PYTHON(/home/fortier/src/ansible/ansible.cfg) = auto_silent
RETRY_FILES_ENABLED(/home/fortier/src/ansible/ansible.cfg) = False
I’m a little leary of dumping host names on a public forum
Here’s the first screen of output from non-verbose output (edited hosts)
Should give a more complete picture 
There is a small change in behavior between 2.9 and 2.10.
If you are up to testing, take a look at https://github.com/ansible/ansible/pull/59932
I’ve added your reproducer to that PR
I’ve never pulled the repo and tried to build/test ansible before.
I’d be worried that the time it would take me to set it up might be more then desirable.
Would it be useful to do over the weekend?
In the short term, you can work around the issue, by removing whitelist items for stdout plugins other than the one you have listed for stdout:
Related config entries:
DEFAULT_CALLBACK_WHITELIST(/home/fortier/src/ansible/ansible.cfg) = [‘skippy’, ‘unixy’, ‘actionable’, ‘yaml’, ‘debug’, ‘dense’]
DEFAULT_LOAD_CALLBACK_PLUGINS(/home/fortier/src/ansible/ansible.cfg) = True
DEFAULT_STDOUT_CALLBACK(/home/fortier/src/ansible/ansible.cfg) = skippy
The callback whitelist should typically be for any callback that is not a stdout callback. All of them that you have listed in the whitelist are stdout callbacks.
Only 1 should ever be loaded, that is where the bug is.
Also, if you don’t want to use skippy as the stdout callback for adhoc, then change the bin_ansible_callback value to False, instead of True.