This was causing a bug in the env module, due to the fact that we now
pass variables for the module through the templating engine combined
with the fact that we split-up the hostvars and setup variables. As a
result, if a variable in the env lookup had the same name as the variable
in Ansible, it would try and template itself over and over again until
the recursion limit would be hit, at which time an empty string was
returned.
Fixes#7396
Previously we assumed the shell on the target systems were 'sh'-
compliant when formatting environment variables preceding command
strings. This patch corrects that by basing the target shell type
on the DEFAULT_EXECUTABLE setting, which can be overridden on a
per-host basis using the inventory variable 'ansible_shell_type'.
Fixes#7352
Since some of the vars contained in there may be used in the templating
of other variables. This also reverts e83a494 which originally fixed
issue #6979 but broke some other variable related things.
Fixes#6979Fixes#7321Fixes#7332
Previously, we set the LANG (and LC_CTYPE) environment variables
directly in the module code and applied them with os.environ().
Instead, we are now pre-pending those variables to the environment
string used to execute the command which allows the user to
override the localization values by setting the environment values
directly (even on a per-task basis):
- subversion: repo=file:///path/to/repos/svn_über dest=/tmp/svntest
environment:
LANG: "C"
LC_CTYPE: "en_US.UTF-8"
So if a user wishes to default their LANG back to C, they can still
avoid unicode issues by doing the above.
Fixes#7060
If someone add ssh_args = " " to his .ansible.cfg, it will result into
strange failure later :
<server.example.org> ESTABLISH CONNECTION FOR USER: misc
<server.example.org> REMOTE_MODULE ping
<server.example.org> EXEC ['ssh', '-C', '-tt', '-q', ' ', '-o', 'KbdInteractiveAuthentication=no',
'-o', 'PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey', '-o', 'PasswordAuthentication=no',
'-o', 'ConnectTimeout=10', 'server.example.org', "/bin/sh -c 'mkdir -p /tmp/ansible-tmp-1397947711.21-5932460998838
&& chmod a+rx /tmp/ansible-tmp-1397947711.21-5932460998838 && echo /tmp/ansible-tmp-1397947711.21-5932460998838'"]
server.example.org | FAILED => 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
The root cause is the empty string between -q and -o, who kinda break mkdir.
If a delegated host is not found in the inventory specified
private_key_file for primary host was not used.
This allows running playbooks without having to define any inventory at
all and to use the same ssh private key for both primary host and
delegated one.
