mirror of
https://github.com/ansible-collections/community.general.git
synced 2024-09-14 20:13:21 +02:00
Reworking v2 play iterator and fixing some other bugs
Still not working quite right: * dynamic includes are not adding the included tasks yet * running roles with tags not quite working right
This commit is contained in:
parent
fbc525cfb6
commit
4af2d0a907
28 changed files with 430 additions and 418 deletions
|
@ -1,66 +0,0 @@
|
|||
# (c) 2012-2014, Michael DeHaan <michael.dehaan@gmail.com>
|
||||
#
|
||||
# This file is part of Ansible
|
||||
#
|
||||
# Ansible is free software: you can redistribute it and/or modify
|
||||
# it under the terms of the GNU General Public License as published by
|
||||
# the Free Software Foundation, either version 3 of the License, or
|
||||
# (at your option) any later version.
|
||||
#
|
||||
# Ansible is distributed in the hope that it will be useful,
|
||||
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
||||
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
||||
# GNU General Public License for more details.
|
||||
#
|
||||
# You should have received a copy of the GNU General Public License
|
||||
# along with Ansible. If not, see <http://www.gnu.org/licenses/>.
|
||||
|
||||
# Make coding more python3-ish
|
||||
from __future__ import (absolute_import, division, print_function)
|
||||
__metaclass__ = type
|
||||
|
||||
from multiprocessing.managers import SyncManager, BaseProxy
|
||||
from ansible.playbook.handler import Handler
|
||||
from ansible.playbook.task import Task
|
||||
from ansible.playbook.play import Play
|
||||
from ansible.errors import AnsibleError
|
||||
|
||||
__all__ = ['AnsibleManager']
|
||||
|
||||
|
||||
class VariableManagerWrapper:
|
||||
'''
|
||||
This class simply acts as a wrapper around the VariableManager class,
|
||||
since manager proxies expect a new object to be returned rather than
|
||||
any existing one. Using this wrapper, a shared proxy can be created
|
||||
and an existing VariableManager class assigned to it, which can then
|
||||
be accessed through the exposed proxy methods.
|
||||
'''
|
||||
|
||||
def __init__(self):
|
||||
self._vm = None
|
||||
|
||||
def get_vars(self, loader, play=None, host=None, task=None):
|
||||
return self._vm.get_vars(loader=loader, play=play, host=host, task=task)
|
||||
|
||||
def set_variable_manager(self, vm):
|
||||
self._vm = vm
|
||||
|
||||
def set_host_variable(self, host, varname, value):
|
||||
self._vm.set_host_variable(host, varname, value)
|
||||
|
||||
def set_host_facts(self, host, facts):
|
||||
self._vm.set_host_facts(host, facts)
|
||||
|
||||
class AnsibleManager(SyncManager):
|
||||
'''
|
||||
This is our custom manager class, which exists only so we may register
|
||||
the new proxy below
|
||||
'''
|
||||
pass
|
||||
|
||||
AnsibleManager.register(
|
||||
typeid='VariableManagerWrapper',
|
||||
callable=VariableManagerWrapper,
|
||||
)
|
||||
|
|
@ -26,278 +26,202 @@ from ansible.utils.boolean import boolean
|
|||
|
||||
__all__ = ['PlayIterator']
|
||||
|
||||
class HostState:
|
||||
def __init__(self, blocks):
|
||||
self._blocks = blocks[:]
|
||||
|
||||
# the primary running states for the play iteration
|
||||
ITERATING_SETUP = 0
|
||||
ITERATING_TASKS = 1
|
||||
ITERATING_RESCUE = 2
|
||||
ITERATING_ALWAYS = 3
|
||||
ITERATING_COMPLETE = 4
|
||||
self.cur_block = 0
|
||||
self.cur_regular_task = 0
|
||||
self.cur_rescue_task = 0
|
||||
self.cur_always_task = 0
|
||||
self.cur_role = None
|
||||
self.run_state = PlayIterator.ITERATING_SETUP
|
||||
self.fail_state = PlayIterator.FAILED_NONE
|
||||
self.pending_setup = False
|
||||
|
||||
# the failure states for the play iteration
|
||||
FAILED_NONE = 0
|
||||
FAILED_SETUP = 1
|
||||
FAILED_TASKS = 2
|
||||
FAILED_RESCUE = 3
|
||||
FAILED_ALWAYS = 4
|
||||
def get_current_block(self):
|
||||
return self._blocks[self.cur_block]
|
||||
|
||||
class PlayState:
|
||||
|
||||
'''
|
||||
A helper class, which keeps track of the task iteration
|
||||
state for a given playbook. This is used in the PlaybookIterator
|
||||
class on a per-host basis.
|
||||
'''
|
||||
|
||||
# FIXME: this class is the representation of a finite state machine,
|
||||
# so we really should have a well defined state representation
|
||||
# documented somewhere...
|
||||
|
||||
def __init__(self, parent_iterator, host):
|
||||
'''
|
||||
Create the initial state, which tracks the running state as well
|
||||
as the failure state, which are used when executing block branches
|
||||
(rescue/always)
|
||||
'''
|
||||
|
||||
self._parent_iterator = parent_iterator
|
||||
self._run_state = ITERATING_SETUP
|
||||
self._failed_state = FAILED_NONE
|
||||
self._gather_facts = parent_iterator._play.gather_facts
|
||||
#self._task_list = parent_iterator._play.compile()
|
||||
self._task_list = parent_iterator._task_list[:]
|
||||
self._host = host
|
||||
|
||||
self._cur_block = None
|
||||
self._cur_role = None
|
||||
self._cur_task_pos = 0
|
||||
self._cur_rescue_pos = 0
|
||||
self._cur_always_pos = 0
|
||||
self._cur_handler_pos = 0
|
||||
|
||||
def next(self, peek=False):
|
||||
'''
|
||||
Determines and returns the next available task from the playbook,
|
||||
advancing through the list of plays as it goes. If peek is set to True,
|
||||
the internal state is not stored.
|
||||
'''
|
||||
|
||||
task = None
|
||||
|
||||
# save this locally so that we can peek at the next task
|
||||
# without updating the internal state of the iterator
|
||||
run_state = self._run_state
|
||||
failed_state = self._failed_state
|
||||
cur_block = self._cur_block
|
||||
cur_role = self._cur_role
|
||||
cur_task_pos = self._cur_task_pos
|
||||
cur_rescue_pos = self._cur_rescue_pos
|
||||
cur_always_pos = self._cur_always_pos
|
||||
cur_handler_pos = self._cur_handler_pos
|
||||
|
||||
|
||||
while True:
|
||||
if run_state == ITERATING_SETUP:
|
||||
if failed_state == FAILED_SETUP:
|
||||
run_state = ITERATING_COMPLETE
|
||||
else:
|
||||
run_state = ITERATING_TASKS
|
||||
|
||||
if self._gather_facts == 'smart' and not self._host.gathered_facts or boolean(self._gather_facts):
|
||||
self._host.set_gathered_facts(True)
|
||||
task = Task()
|
||||
# FIXME: this is not the best way to get this...
|
||||
task.set_loader(self._parent_iterator._play._loader)
|
||||
task.action = 'setup'
|
||||
break
|
||||
elif run_state == ITERATING_TASKS:
|
||||
# if there is any failure state besides FAILED_NONE, we should
|
||||
# change to some other running state
|
||||
if failed_state != FAILED_NONE or cur_task_pos > len(self._task_list) - 1:
|
||||
# if there is a block (and there always should be), start running
|
||||
# the rescue portion if it exists (and if we haven't failed that
|
||||
# already), or the always portion (if it exists and we didn't fail
|
||||
# there too). Otherwise, we're done iterating.
|
||||
if cur_block:
|
||||
if failed_state != FAILED_RESCUE and cur_block.rescue:
|
||||
run_state = ITERATING_RESCUE
|
||||
cur_rescue_pos = 0
|
||||
elif failed_state != FAILED_ALWAYS and cur_block.always:
|
||||
run_state = ITERATING_ALWAYS
|
||||
cur_always_pos = 0
|
||||
else:
|
||||
run_state = ITERATING_COMPLETE
|
||||
else:
|
||||
run_state = ITERATING_COMPLETE
|
||||
else:
|
||||
task = self._task_list[cur_task_pos]
|
||||
if cur_block is not None and cur_block != task._block:
|
||||
run_state = ITERATING_ALWAYS
|
||||
continue
|
||||
else:
|
||||
cur_block = task._block
|
||||
cur_task_pos += 1
|
||||
|
||||
# Break out of the while loop now that we have our task
|
||||
break
|
||||
|
||||
elif run_state == ITERATING_RESCUE:
|
||||
# If we're iterating through the rescue tasks, make sure we haven't
|
||||
# failed yet. If so, move on to the always block or if not get the
|
||||
# next rescue task (if one exists)
|
||||
if failed_state == FAILED_RESCUE or cur_block.rescue is None or cur_rescue_pos > len(cur_block.rescue) - 1:
|
||||
run_state = ITERATING_ALWAYS
|
||||
else:
|
||||
task = cur_block.rescue[cur_rescue_pos]
|
||||
cur_rescue_pos += 1
|
||||
break
|
||||
|
||||
elif run_state == ITERATING_ALWAYS:
|
||||
# If we're iterating through the always tasks, make sure we haven't
|
||||
# failed yet. If so, we're done iterating otherwise get the next always
|
||||
# task (if one exists)
|
||||
if failed_state == FAILED_ALWAYS or cur_block.always is None or cur_always_pos > len(cur_block.always) - 1:
|
||||
cur_block = None
|
||||
if failed_state == FAILED_ALWAYS or cur_task_pos > len(self._task_list) - 1:
|
||||
run_state = ITERATING_COMPLETE
|
||||
else:
|
||||
run_state = ITERATING_TASKS
|
||||
else:
|
||||
task = cur_block.always[cur_always_pos]
|
||||
cur_always_pos += 1
|
||||
break
|
||||
|
||||
elif run_state == ITERATING_COMPLETE:
|
||||
# done iterating, return None to signify that
|
||||
return None
|
||||
|
||||
if task._role:
|
||||
# if we had a current role, mark that role as completed
|
||||
if cur_role and task._role != cur_role and not peek:
|
||||
cur_role._completed = True
|
||||
|
||||
cur_role = task._role
|
||||
|
||||
# if the current role has not had its task run flag set, mark
|
||||
# clear the completed flag so we can correctly determine if the
|
||||
# role was run
|
||||
if not cur_role._had_task_run and not peek:
|
||||
cur_role._completed = False
|
||||
|
||||
# If we're not just peeking at the next task, save the internal state
|
||||
if not peek:
|
||||
self._run_state = run_state
|
||||
self._failed_state = failed_state
|
||||
self._cur_block = cur_block
|
||||
self._cur_role = cur_role
|
||||
self._cur_task_pos = cur_task_pos
|
||||
self._cur_rescue_pos = cur_rescue_pos
|
||||
self._cur_always_pos = cur_always_pos
|
||||
self._cur_handler_pos = cur_handler_pos
|
||||
|
||||
return task
|
||||
|
||||
def mark_failed(self):
|
||||
'''
|
||||
Escalates the failed state relative to the running state.
|
||||
'''
|
||||
if self._run_state == ITERATING_SETUP:
|
||||
self._failed_state = FAILED_SETUP
|
||||
elif self._run_state == ITERATING_TASKS:
|
||||
self._failed_state = FAILED_TASKS
|
||||
elif self._run_state == ITERATING_RESCUE:
|
||||
self._failed_state = FAILED_RESCUE
|
||||
elif self._run_state == ITERATING_ALWAYS:
|
||||
self._failed_state = FAILED_ALWAYS
|
||||
|
||||
def add_tasks(self, task_list):
|
||||
if self._run_state == ITERATING_TASKS:
|
||||
before = self._task_list[:self._cur_task_pos]
|
||||
after = self._task_list[self._cur_task_pos:]
|
||||
self._task_list = before + task_list + after
|
||||
elif self._run_state == ITERATING_RESCUE:
|
||||
before = self._cur_block.rescue[:self._cur_rescue_pos]
|
||||
after = self._cur_block.rescue[self._cur_rescue_pos:]
|
||||
self._cur_block.rescue = before + task_list + after
|
||||
elif self._run_state == ITERATING_ALWAYS:
|
||||
before = self._cur_block.always[:self._cur_always_pos]
|
||||
after = self._cur_block.always[self._cur_always_pos:]
|
||||
self._cur_block.always = before + task_list + after
|
||||
def copy(self):
|
||||
new_state = HostState(self._blocks)
|
||||
new_state.cur_block = self.cur_block
|
||||
new_state.cur_regular_task = self.cur_regular_task
|
||||
new_state.cur_rescue_task = self.cur_rescue_task
|
||||
new_state.cur_always_task = self.cur_always_task
|
||||
new_state.cur_role = self.cur_role
|
||||
new_state.run_state = self.run_state
|
||||
new_state.fail_state = self.fail_state
|
||||
new_state.pending_setup = self.pending_setup
|
||||
return new_state
|
||||
|
||||
class PlayIterator:
|
||||
# the primary running states for the play iteration
|
||||
ITERATING_SETUP = 0
|
||||
ITERATING_TASKS = 1
|
||||
ITERATING_RESCUE = 2
|
||||
ITERATING_ALWAYS = 3
|
||||
ITERATING_COMPLETE = 4
|
||||
|
||||
'''
|
||||
The main iterator class, which keeps the state of the playbook
|
||||
on a per-host basis using the above PlaybookState class.
|
||||
'''
|
||||
# the failure states for the play iteration, which are powers
|
||||
# of 2 as they may be or'ed together in certain circumstances
|
||||
FAILED_NONE = 0
|
||||
FAILED_SETUP = 1
|
||||
FAILED_TASKS = 2
|
||||
FAILED_RESCUE = 4
|
||||
FAILED_ALWAYS = 8
|
||||
|
||||
def __init__(self, inventory, play):
|
||||
# FIXME: should we save the post_validated play from below here instead?
|
||||
self._play = play
|
||||
self._inventory = inventory
|
||||
self._host_entries = dict()
|
||||
self._first_host = None
|
||||
|
||||
# Build the per-host dictionary of playbook states, using a copy
|
||||
# of the play object so we can post_validate it to ensure any templated
|
||||
# fields are filled in without modifying the original object, since
|
||||
# post_validate() saves the templated values.
|
||||
|
||||
# FIXME: this is a hacky way of doing this, the iterator should
|
||||
# instead get the loader and variable manager directly
|
||||
# as args to __init__
|
||||
# post validate the play, as we need some fields to be finalized now
|
||||
# so that we can use them to setup the iterator properly
|
||||
all_vars = inventory._variable_manager.get_vars(loader=inventory._loader, play=play)
|
||||
new_play = play.copy()
|
||||
new_play.post_validate(all_vars, fail_on_undefined=False)
|
||||
|
||||
self._task_list = new_play.compile()
|
||||
self._blocks = new_play.compile()
|
||||
self._host_states = {}
|
||||
for host in inventory.get_hosts(new_play.hosts):
|
||||
if self._first_host is None:
|
||||
self._first_host = host
|
||||
self._host_entries[host.get_name()] = PlayState(parent_iterator=self, host=host)
|
||||
self._host_states[host.name] = HostState(blocks=self._blocks)
|
||||
|
||||
# FIXME: remove, probably not required anymore
|
||||
#def get_next_task(self, peek=False):
|
||||
# ''' returns the next task for host[0] '''
|
||||
#
|
||||
# first_entry = self._host_entries[self._first_host.get_name()]
|
||||
# if not peek:
|
||||
# for entry in self._host_entries:
|
||||
# if entry != self._first_host.get_name():
|
||||
# target_entry = self._host_entries[entry]
|
||||
# if target_entry._cur_task_pos == first_entry._cur_task_pos:
|
||||
# target_entry.next()
|
||||
# return first_entry.next(peek=peek)
|
||||
|
||||
def get_next_task_for_host(self, host, peek=False):
|
||||
''' fetch the next task for the given host '''
|
||||
if host.get_name() not in self._host_entries:
|
||||
def get_host_state(self, host):
|
||||
try:
|
||||
return self._host_states[host.name].copy()
|
||||
except KeyError:
|
||||
raise AnsibleError("invalid host (%s) specified for playbook iteration" % host)
|
||||
|
||||
return self._host_entries[host.get_name()].next(peek=peek)
|
||||
def get_next_task_for_host(self, host, peek=False, lock_step=True):
|
||||
s = self.get_host_state(host)
|
||||
|
||||
task = None
|
||||
if s.run_state == self.ITERATING_COMPLETE:
|
||||
return None
|
||||
else:
|
||||
while True:
|
||||
try:
|
||||
cur_block = s._blocks[s.cur_block]
|
||||
except IndexError:
|
||||
s.run_state = self.ITERATING_COMPLETE
|
||||
break
|
||||
|
||||
if s.run_state == self.ITERATING_SETUP:
|
||||
s.run_state = self.ITERATING_TASKS
|
||||
if self._play._gather_facts == 'smart' and not host.gathered_facts or boolean(self._play._gather_facts):
|
||||
# mark the host as having gathered facts
|
||||
host.set_gathered_facts(True)
|
||||
|
||||
task = Task()
|
||||
task.action = 'setup'
|
||||
task.set_loader(self._play._loader)
|
||||
|
||||
elif s.run_state == self.ITERATING_TASKS:
|
||||
# clear the pending setup flag, since we're past that and it didn't fail
|
||||
if s.pending_setup:
|
||||
s.pending_setup = False
|
||||
|
||||
if s.fail_state & self.FAILED_TASKS == self.FAILED_TASKS:
|
||||
s.run_state = self.ITERATING_RESCUE
|
||||
elif s.cur_regular_task >= len(cur_block.block):
|
||||
s.run_state = self.ITERATING_ALWAYS
|
||||
else:
|
||||
task = cur_block.block[s.cur_regular_task]
|
||||
s.cur_regular_task += 1
|
||||
break
|
||||
elif s.run_state == self.ITERATING_RESCUE:
|
||||
if s.fail_state & self.FAILED_RESCUE == self.FAILED_RESCUE:
|
||||
s.run_state = self.ITERATING_ALWAYS
|
||||
elif s.cur_rescue_task >= len(cur_block.rescue):
|
||||
if len(cur_block.rescue) > 0:
|
||||
s.fail_state = self.FAILED_NONE
|
||||
s.run_state = self.ITERATING_ALWAYS
|
||||
else:
|
||||
task = cur_block.rescue[s.cur_rescue_task]
|
||||
s.cur_rescue_task += 1
|
||||
break
|
||||
elif s.run_state == self.ITERATING_ALWAYS:
|
||||
if s.cur_always_task >= len(cur_block.always):
|
||||
if s.fail_state != self.FAILED_NONE:
|
||||
s.run_state = self.ITERATING_COMPLETE
|
||||
break
|
||||
else:
|
||||
s.cur_block += 1
|
||||
s.cur_regular_task = 0
|
||||
s.cur_rescue_task = 0
|
||||
s.cur_always_task = 0
|
||||
s.run_state = self.ITERATING_TASKS
|
||||
else:
|
||||
task= cur_block.always[s.cur_always_task]
|
||||
s.cur_always_task += 1
|
||||
break
|
||||
|
||||
if task and task._role:
|
||||
# if we had a current role, mark that role as completed
|
||||
if s.cur_role and task._role != s.cur_role and s.cur_role._had_task_run and not peek:
|
||||
s.cur_role._completed = True
|
||||
|
||||
s.cur_role = task._role
|
||||
|
||||
if not peek:
|
||||
self._host_states[host.name] = s
|
||||
|
||||
return (s, task)
|
||||
|
||||
def mark_host_failed(self, host):
|
||||
''' mark the given host as failed '''
|
||||
if host.get_name() not in self._host_entries:
|
||||
raise AnsibleError("invalid host (%s) specified for playbook iteration" % host)
|
||||
s = self.get_host_state(host)
|
||||
if s.pending_setup:
|
||||
s.fail_state |= self.FAILED_SETUP
|
||||
s.run_state = self.ITERATING_COMPLETE
|
||||
elif s.run_state == self.ITERATING_TASKS:
|
||||
s.fail_state |= self.FAILED_TASKS
|
||||
s.run_state = self.ITERATING_RESCUE
|
||||
elif s.run_state == self.ITERATING_RESCUE:
|
||||
s.fail_state |= self.FAILED_RESCUE
|
||||
s.run_state = self.ITERATING_ALWAYS
|
||||
elif s.run_state == self.ITERATING_ALWAYS:
|
||||
s.fail_state |= self.FAILED_ALWAYS
|
||||
s.run_state = self.ITERATING_COMPLETE
|
||||
self._host_states[host.name] = s
|
||||
|
||||
self._host_entries[host.get_name()].mark_failed()
|
||||
def get_failed_hosts(self):
|
||||
return dict((host, True) for (host, state) in self._host_states.iteritems() if state.run_state == self.ITERATING_COMPLETE and state.failed_state != self.FAILED_NONE)
|
||||
|
||||
def get_original_task(self, task):
|
||||
def get_original_task(self, host, task):
|
||||
'''
|
||||
Finds the task in the task list which matches the UUID of the given task.
|
||||
The executor engine serializes/deserializes objects as they are passed through
|
||||
the different processes, and not all data structures are preserved. This method
|
||||
allows us to find the original task passed into the executor engine.
|
||||
'''
|
||||
|
||||
for t in self._task_list:
|
||||
for block in self._blocks:
|
||||
if block.block:
|
||||
for t in block.block:
|
||||
if t._uuid == task._uuid:
|
||||
return t
|
||||
if block.rescue:
|
||||
for t in block.rescue:
|
||||
if t._uuid == task._uuid:
|
||||
return t
|
||||
if block.always:
|
||||
for t in block.always:
|
||||
if t._uuid == task._uuid:
|
||||
return t
|
||||
|
||||
return None
|
||||
|
||||
def add_tasks(self, host, task_list):
|
||||
if host.name not in self._host_entries:
|
||||
raise AnsibleError("invalid host (%s) specified for playbook iteration (expanding task list)" % host)
|
||||
def add_tasks(self, task_list):
|
||||
if self._run_state == self.ITERATING_TASKS:
|
||||
before = self._task_list[:self._cur_task_pos + self._tasks_added]
|
||||
after = self._task_list[self._cur_task_pos + self._tasks_added:]
|
||||
self._task_list = before + task_list + after
|
||||
elif self._run_state == self.ITERATING_RESCUE:
|
||||
before = self._cur_block.rescue[:self._cur_rescue_pos + self._tasks_added]
|
||||
after = self._cur_block.rescue[self._cur_rescue_pos + self._tasks_added:]
|
||||
self._cur_block.rescue = before + task_list + after
|
||||
elif self._run_state == self.ITERATING_ALWAYS:
|
||||
before = self._cur_block.always[:self._cur_always_pos + self._tasks_added]
|
||||
after = self._cur_block.always[self._cur_always_pos + self._tasks_added:]
|
||||
self._cur_block.always = before + task_list + after
|
||||
|
||||
# set this internal flag now so we know if
|
||||
self._tasks_added += len(task_list)
|
||||
|
||||
self._host_entries[host.name].add_tasks(task_list)
|
||||
|
|
|
@ -137,12 +137,13 @@ class ResultProcess(multiprocessing.Process):
|
|||
result_items = [ result._result ]
|
||||
|
||||
for result_item in result_items:
|
||||
if 'include' in result_item:
|
||||
include_variables = result_item.get('include_variables', dict())
|
||||
if 'item' in result_item:
|
||||
include_variables['item'] = result_item['item']
|
||||
self._send_result(('include', result._host, result._task, result_item['include'], include_variables))
|
||||
elif 'add_host' in result_item:
|
||||
#if 'include' in result_item:
|
||||
# include_variables = result_item.get('include_variables', dict())
|
||||
# if 'item' in result_item:
|
||||
# include_variables['item'] = result_item['item']
|
||||
# self._send_result(('include', result._host, result._task, result_item['include'], include_variables))
|
||||
#elif 'add_host' in result_item:
|
||||
if 'add_host' in result_item:
|
||||
# this task added a new host (add_host module)
|
||||
self._send_result(('add_host', result_item))
|
||||
elif 'add_group' in result_item:
|
||||
|
|
|
@ -132,13 +132,14 @@ class TaskExecutor:
|
|||
res = self._execute(variables=task_vars)
|
||||
(self._task, tmp_task) = (tmp_task, self._task)
|
||||
|
||||
# FIXME: we should be sending back a callback result for each item in the loop here
|
||||
|
||||
# now update the result with the item info, and append the result
|
||||
# to the list of results
|
||||
res['item'] = item
|
||||
results.append(res)
|
||||
|
||||
# FIXME: we should be sending back a callback result for each item in the loop here
|
||||
print(res)
|
||||
|
||||
return results
|
||||
|
||||
def _squash_items(self, items, variables):
|
||||
|
|
|
@ -26,7 +26,6 @@ import sys
|
|||
|
||||
from ansible.errors import AnsibleError
|
||||
from ansible.executor.connection_info import ConnectionInformation
|
||||
#from ansible.executor.manager import AnsibleManager
|
||||
from ansible.executor.play_iterator import PlayIterator
|
||||
from ansible.executor.process.worker import WorkerProcess
|
||||
from ansible.executor.process.result import ResultProcess
|
||||
|
@ -36,7 +35,6 @@ from ansible.utils.debug import debug
|
|||
|
||||
__all__ = ['TaskQueueManager']
|
||||
|
||||
|
||||
class TaskQueueManager:
|
||||
|
||||
'''
|
||||
|
@ -59,10 +57,6 @@ class TaskQueueManager:
|
|||
# a special flag to help us exit cleanly
|
||||
self._terminated = False
|
||||
|
||||
# create and start the multiprocessing manager
|
||||
#self._manager = AnsibleManager()
|
||||
#self._manager.start()
|
||||
|
||||
# this dictionary is used to keep track of notified handlers
|
||||
self._notified_handlers = dict()
|
||||
|
||||
|
|
|
@ -273,7 +273,7 @@ class Base:
|
|||
if needle in self._attributes:
|
||||
return self._attributes[needle]
|
||||
|
||||
raise AttributeError("attribute not found: %s" % needle)
|
||||
raise AttributeError("attribute not found in %s: %s" % (self.__class__.__name__, needle))
|
||||
|
||||
def __getstate__(self):
|
||||
return self.serialize()
|
||||
|
|
|
@ -28,9 +28,9 @@ from ansible.playbook.taggable import Taggable
|
|||
|
||||
class Block(Base, Conditional, Taggable):
|
||||
|
||||
_block = FieldAttribute(isa='list')
|
||||
_rescue = FieldAttribute(isa='list')
|
||||
_always = FieldAttribute(isa='list')
|
||||
_block = FieldAttribute(isa='list', default=[])
|
||||
_rescue = FieldAttribute(isa='list', default=[])
|
||||
_always = FieldAttribute(isa='list', default=[])
|
||||
|
||||
# for future consideration? this would be functionally
|
||||
# similar to the 'else' clause for exceptions
|
||||
|
@ -41,6 +41,7 @@ class Block(Base, Conditional, Taggable):
|
|||
self._role = role
|
||||
self._task_include = task_include
|
||||
self._use_handlers = use_handlers
|
||||
self._dep_chain = []
|
||||
|
||||
super(Block, self).__init__()
|
||||
|
||||
|
@ -141,6 +142,7 @@ class Block(Base, Conditional, Taggable):
|
|||
def copy(self):
|
||||
new_me = super(Block, self).copy()
|
||||
new_me._use_handlers = self._use_handlers
|
||||
new_me._dep_chain = self._dep_chain[:]
|
||||
|
||||
new_me._parent_block = None
|
||||
if self._parent_block:
|
||||
|
@ -163,6 +165,7 @@ class Block(Base, Conditional, Taggable):
|
|||
'''
|
||||
|
||||
data = dict(when=self.when)
|
||||
data['dep_chain'] = self._dep_chain
|
||||
|
||||
if self._role is not None:
|
||||
data['role'] = self._role.serialize()
|
||||
|
@ -177,11 +180,11 @@ class Block(Base, Conditional, Taggable):
|
|||
serialize method
|
||||
'''
|
||||
|
||||
#from ansible.playbook.task_include import TaskInclude
|
||||
from ansible.playbook.task import Task
|
||||
|
||||
# unpack the when attribute, which is the only one we want
|
||||
self.when = data.get('when')
|
||||
self._dep_chain = data.get('dep_chain', [])
|
||||
|
||||
# if there was a serialized role, unpack it too
|
||||
role_data = data.get('role')
|
||||
|
@ -198,6 +201,10 @@ class Block(Base, Conditional, Taggable):
|
|||
self._task_include = ti
|
||||
|
||||
def evaluate_conditional(self, all_vars):
|
||||
if len(self._dep_chain):
|
||||
for dep in self._dep_chain:
|
||||
if not dep.evaluate_conditional(all_vars):
|
||||
return False
|
||||
if self._task_include is not None:
|
||||
if not self._task_include.evaluate_conditional(all_vars):
|
||||
return False
|
||||
|
@ -211,6 +218,9 @@ class Block(Base, Conditional, Taggable):
|
|||
|
||||
def evaluate_tags(self, only_tags, skip_tags, all_vars):
|
||||
result = False
|
||||
if len(self._dep_chain):
|
||||
for dep in self._dep_chain:
|
||||
result |= dep.evaluate_tags(only_tags=only_tags, skip_tags=skip_tags, all_vars=all_vars)
|
||||
if self._parent_block is not None:
|
||||
result |= self._parent_block.evaluate_tags(only_tags=only_tags, skip_tags=skip_tags, all_vars=all_vars)
|
||||
elif self._role is not None:
|
||||
|
@ -227,3 +237,6 @@ class Block(Base, Conditional, Taggable):
|
|||
if self._task_include:
|
||||
self._task_include.set_loader(loader)
|
||||
|
||||
for dep in self._dep_chain:
|
||||
dep.set_loader(loader)
|
||||
|
||||
|
|
|
@ -198,13 +198,13 @@ class Play(Base, Taggable):
|
|||
the parent role R last. This is done for all roles in the Play.
|
||||
'''
|
||||
|
||||
task_list = []
|
||||
block_list = []
|
||||
|
||||
if len(self.roles) > 0:
|
||||
for r in self.roles:
|
||||
task_list.extend(r.compile())
|
||||
block_list.extend(r.compile())
|
||||
|
||||
return task_list
|
||||
return block_list
|
||||
|
||||
def compile(self):
|
||||
'''
|
||||
|
@ -213,14 +213,14 @@ class Play(Base, Taggable):
|
|||
tasks specified in the play.
|
||||
'''
|
||||
|
||||
task_list = []
|
||||
block_list = []
|
||||
|
||||
task_list.extend(compile_block_list(self.pre_tasks))
|
||||
task_list.extend(self._compile_roles())
|
||||
task_list.extend(compile_block_list(self.tasks))
|
||||
task_list.extend(compile_block_list(self.post_tasks))
|
||||
block_list.extend(self.pre_tasks)
|
||||
block_list.extend(self._compile_roles())
|
||||
block_list.extend(self.tasks)
|
||||
block_list.extend(self.post_tasks)
|
||||
|
||||
return task_list
|
||||
return block_list
|
||||
|
||||
def get_vars(self):
|
||||
return self.vars.copy()
|
||||
|
|
|
@ -304,25 +304,25 @@ class Role(Base, Conditional, Taggable):
|
|||
can correctly take their parent's tags/conditionals into account.
|
||||
'''
|
||||
|
||||
task_list = []
|
||||
block_list = []
|
||||
|
||||
# update the dependency chain here
|
||||
new_dep_chain = dep_chain + [self]
|
||||
|
||||
deps = self.get_direct_dependencies()
|
||||
for dep in deps:
|
||||
dep_tasks = dep.compile(dep_chain=new_dep_chain)
|
||||
for dep_task in dep_tasks:
|
||||
dep_blocks = dep.compile(dep_chain=new_dep_chain)
|
||||
for dep_block in dep_blocks:
|
||||
# since we're modifying the task, and need it to be unique,
|
||||
# we make a copy of it here and assign the dependency chain
|
||||
# to the copy, then append the copy to the task list.
|
||||
new_dep_task = dep_task.copy()
|
||||
new_dep_task._dep_chain = new_dep_chain
|
||||
task_list.append(new_dep_task)
|
||||
new_dep_block = dep_block.copy()
|
||||
new_dep_block._dep_chain = new_dep_chain
|
||||
block_list.append(new_dep_block)
|
||||
|
||||
task_list.extend(compile_block_list(self._task_blocks))
|
||||
block_list.extend(self._task_blocks)
|
||||
|
||||
return task_list
|
||||
return block_list
|
||||
|
||||
def serialize(self, include_deps=True):
|
||||
res = super(Role, self).serialize()
|
||||
|
|
|
@ -102,7 +102,6 @@ class Task(Base, Conditional, Taggable):
|
|||
self._block = block
|
||||
self._role = role
|
||||
self._task_include = task_include
|
||||
self._dep_chain = []
|
||||
|
||||
super(Task, self).__init__()
|
||||
|
||||
|
@ -226,7 +225,6 @@ class Task(Base, Conditional, Taggable):
|
|||
|
||||
def copy(self):
|
||||
new_me = super(Task, self).copy()
|
||||
new_me._dep_chain = self._dep_chain[:]
|
||||
|
||||
new_me._block = None
|
||||
if self._block:
|
||||
|
@ -244,7 +242,6 @@ class Task(Base, Conditional, Taggable):
|
|||
|
||||
def serialize(self):
|
||||
data = super(Task, self).serialize()
|
||||
data['dep_chain'] = self._dep_chain
|
||||
|
||||
if self._block:
|
||||
data['block'] = self._block.serialize()
|
||||
|
@ -263,7 +260,6 @@ class Task(Base, Conditional, Taggable):
|
|||
#from ansible.playbook.task_include import TaskInclude
|
||||
|
||||
block_data = data.get('block')
|
||||
self._dep_chain = data.get('dep_chain', [])
|
||||
|
||||
if block_data:
|
||||
b = Block()
|
||||
|
@ -289,10 +285,6 @@ class Task(Base, Conditional, Taggable):
|
|||
super(Task, self).deserialize(data)
|
||||
|
||||
def evaluate_conditional(self, all_vars):
|
||||
if len(self._dep_chain):
|
||||
for dep in self._dep_chain:
|
||||
if not dep.evaluate_conditional(all_vars):
|
||||
return False
|
||||
if self._block is not None:
|
||||
if not self._block.evaluate_conditional(all_vars):
|
||||
return False
|
||||
|
@ -303,9 +295,6 @@ class Task(Base, Conditional, Taggable):
|
|||
|
||||
def evaluate_tags(self, only_tags, skip_tags, all_vars):
|
||||
result = False
|
||||
if len(self._dep_chain):
|
||||
for dep in self._dep_chain:
|
||||
result |= dep.evaluate_tags(only_tags=only_tags, skip_tags=skip_tags, all_vars=all_vars)
|
||||
if self._block is not None:
|
||||
result |= self._block.evaluate_tags(only_tags=only_tags, skip_tags=skip_tags, all_vars=all_vars)
|
||||
return result | super(Task, self).evaluate_tags(only_tags=only_tags, skip_tags=skip_tags, all_vars=all_vars)
|
||||
|
@ -324,5 +313,3 @@ class Task(Base, Conditional, Taggable):
|
|||
if self._task_include:
|
||||
self._task_include.set_loader(loader)
|
||||
|
||||
for dep in self._dep_chain:
|
||||
dep.set_loader(loader)
|
||||
|
|
|
@ -68,8 +68,8 @@ class StrategyBase:
|
|||
num_failed = len(self._tqm._failed_hosts)
|
||||
num_unreachable = len(self._tqm._unreachable_hosts)
|
||||
|
||||
debug("running the cleanup portion of the play")
|
||||
result &= self.cleanup(iterator, connection_info)
|
||||
#debug("running the cleanup portion of the play")
|
||||
#result &= self.cleanup(iterator, connection_info)
|
||||
debug("running handlers")
|
||||
result &= self.run_handlers(iterator, connection_info)
|
||||
|
||||
|
@ -131,6 +131,7 @@ class StrategyBase:
|
|||
if result[0] == 'host_task_failed':
|
||||
if not task.ignore_errors:
|
||||
debug("marking %s as failed" % host.get_name())
|
||||
iterator.mark_host_failed(host)
|
||||
self._tqm._failed_hosts[host.get_name()] = True
|
||||
self._callback.runner_on_failed(task, task_result)
|
||||
elif result[0] == 'host_unreachable':
|
||||
|
@ -151,26 +152,25 @@ class StrategyBase:
|
|||
# lookup the role in the ROLE_CACHE to make sure we're dealing
|
||||
# with the correct object and mark it as executed
|
||||
for (entry, role_obj) in ROLE_CACHE[task_result._task._role._role_name].iteritems():
|
||||
#hashed_entry = frozenset(task_result._task._role._role_params.iteritems())
|
||||
hashed_entry = hash_params(task_result._task._role._role_params)
|
||||
if entry == hashed_entry :
|
||||
role_obj._had_task_run = True
|
||||
|
||||
elif result[0] == 'include':
|
||||
host = result[1]
|
||||
task = result[2]
|
||||
include_file = result[3]
|
||||
include_vars = result[4]
|
||||
|
||||
if isinstance(task, Handler):
|
||||
# FIXME: figure out how to make includes work for handlers
|
||||
pass
|
||||
else:
|
||||
original_task = iterator.get_original_task(task)
|
||||
if original_task._role:
|
||||
include_file = self._loader.path_dwim_relative(original_task._role._role_path, 'tasks', include_file)
|
||||
new_tasks = self._load_included_file(original_task, include_file, include_vars)
|
||||
iterator.add_tasks(host, new_tasks)
|
||||
#elif result[0] == 'include':
|
||||
# host = result[1]
|
||||
# task = result[2]
|
||||
# include_file = result[3]
|
||||
# include_vars = result[4]
|
||||
#
|
||||
# if isinstance(task, Handler):
|
||||
# # FIXME: figure out how to make includes work for handlers
|
||||
# pass
|
||||
# else:
|
||||
# original_task = iterator.get_original_task(host, task)
|
||||
# if original_task and original_task._role:
|
||||
# include_file = self._loader.path_dwim_relative(original_task._role._role_path, 'tasks', include_file)
|
||||
# new_tasks = self._load_included_file(original_task, include_file, include_vars)
|
||||
# iterator.add_tasks(host, new_tasks)
|
||||
|
||||
elif result[0] == 'add_host':
|
||||
task_result = result[1]
|
||||
|
@ -314,6 +314,8 @@ class StrategyBase:
|
|||
|
||||
|
||||
task_list = compile_block_list(block_list)
|
||||
|
||||
# set the vars for this task from those specified as params to the include
|
||||
for t in task_list:
|
||||
t.vars = include_vars.copy()
|
||||
|
||||
|
@ -355,10 +357,12 @@ class StrategyBase:
|
|||
iterator.mark_host_failed(host)
|
||||
del self._tqm._failed_hosts[host_name]
|
||||
|
||||
if host_name not in self._tqm._unreachable_hosts and iterator.get_next_task_for_host(host, peek=True):
|
||||
if host_name in self._blocked_hosts:
|
||||
work_to_do = True
|
||||
# check to see if this host is blocked (still executing a previous task)
|
||||
if not host_name in self._blocked_hosts:
|
||||
continue
|
||||
elif iterator.get_next_task_for_host(host, peek=True) and host_name not in self._tqm._unreachable_hosts:
|
||||
work_to_do = True
|
||||
|
||||
# pop the task, mark the host blocked, and queue it
|
||||
self._blocked_hosts[host_name] = True
|
||||
task = iterator.get_next_task_for_host(host)
|
||||
|
@ -367,6 +371,7 @@ class StrategyBase:
|
|||
self._queue_task(host, task, task_vars, connection_info)
|
||||
|
||||
self._process_pending_results(iterator)
|
||||
time.sleep(0.01)
|
||||
|
||||
# no more work, wait until the queue is drained
|
||||
self._wait_on_pending_results(iterator)
|
||||
|
|
|
@ -20,11 +20,96 @@ from __future__ import (absolute_import, division, print_function)
|
|||
__metaclass__ = type
|
||||
|
||||
from ansible.errors import AnsibleError
|
||||
from ansible.executor.play_iterator import PlayIterator
|
||||
from ansible.playbook.task import Task
|
||||
from ansible.plugins.strategies import StrategyBase
|
||||
from ansible.utils.debug import debug
|
||||
|
||||
class StrategyModule(StrategyBase):
|
||||
|
||||
def _get_next_task_lockstep(self, hosts, iterator):
|
||||
'''
|
||||
Returns a list of (host, task) tuples, where the task may
|
||||
be a noop task to keep the iterator in lock step across
|
||||
all hosts.
|
||||
'''
|
||||
|
||||
noop_task = Task()
|
||||
noop_task.action = 'meta'
|
||||
noop_task.args['_raw_params'] = 'noop'
|
||||
noop_task.set_loader(iterator._play._loader)
|
||||
|
||||
host_tasks = {}
|
||||
for host in hosts:
|
||||
host_tasks[host.name] = iterator.get_next_task_for_host(host, peek=True)
|
||||
|
||||
num_setups = 0
|
||||
num_tasks = 0
|
||||
num_rescue = 0
|
||||
num_always = 0
|
||||
|
||||
lowest_cur_block = len(iterator._blocks)
|
||||
|
||||
for (k, v) in host_tasks.iteritems():
|
||||
(s, t) = v
|
||||
if s.cur_block < lowest_cur_block and s.run_state != PlayIterator.ITERATING_COMPLETE:
|
||||
lowest_cur_block = s.cur_block
|
||||
|
||||
if s.run_state == PlayIterator.ITERATING_SETUP:
|
||||
num_setups += 1
|
||||
elif s.run_state == PlayIterator.ITERATING_TASKS:
|
||||
num_tasks += 1
|
||||
elif s.run_state == PlayIterator.ITERATING_RESCUE:
|
||||
num_rescue += 1
|
||||
elif s.run_state == PlayIterator.ITERATING_ALWAYS:
|
||||
num_always += 1
|
||||
|
||||
def _advance_selected_hosts(hosts, cur_block, cur_state):
|
||||
'''
|
||||
This helper returns the task for all hosts in the requested
|
||||
state, otherwise they get a noop dummy task. This also advances
|
||||
the state of the host, since the given states are determined
|
||||
while using peek=True.
|
||||
'''
|
||||
# we return the values in the order they were originally
|
||||
# specified in the given hosts array
|
||||
rvals = []
|
||||
for host in hosts:
|
||||
(s, t) = host_tasks[host.name]
|
||||
if s.run_state == cur_state and s.cur_block == cur_block:
|
||||
new_t = iterator.get_next_task_for_host(host)
|
||||
#if new_t != t:
|
||||
# raise AnsibleError("iterator error, wtf?")
|
||||
rvals.append((host, t))
|
||||
else:
|
||||
rvals.append((host, noop_task))
|
||||
return rvals
|
||||
|
||||
# if any hosts are in ITERATING_SETUP, return the setup task
|
||||
# while all other hosts get a noop
|
||||
if num_setups:
|
||||
return _advance_selected_hosts(hosts, lowest_cur_block, PlayIterator.ITERATING_SETUP)
|
||||
|
||||
# if any hosts are in ITERATING_TASKS, return the next normal
|
||||
# task for these hosts, while all other hosts get a noop
|
||||
if num_tasks:
|
||||
return _advance_selected_hosts(hosts, lowest_cur_block, PlayIterator.ITERATING_TASKS)
|
||||
|
||||
# if any hosts are in ITERATING_RESCUE, return the next rescue
|
||||
# task for these hosts, while all other hosts get a noop
|
||||
if num_rescue:
|
||||
return _advance_selected_hosts(hosts, lowest_cur_block, PlayIterator.ITERATING_RESCUE)
|
||||
|
||||
# if any hosts are in ITERATING_ALWAYS, return the next always
|
||||
# task for these hosts, while all other hosts get a noop
|
||||
if num_always:
|
||||
return _advance_selected_hosts(hosts, lowest_cur_block, PlayIterator.ITERATING_ALWAYS)
|
||||
|
||||
# at this point, everything must be ITERATING_COMPLETE, so we
|
||||
# return None for all hosts in the list
|
||||
return [(host, None) for host in hosts]
|
||||
|
||||
|
||||
def run(self, iterator, connection_info):
|
||||
'''
|
||||
The linear strategy is simple - get the next task and queue
|
||||
|
@ -40,6 +125,7 @@ class StrategyModule(StrategyBase):
|
|||
|
||||
try:
|
||||
debug("getting the remaining hosts for this loop")
|
||||
self._tqm._failed_hosts = iterator.get_failed_hosts()
|
||||
hosts_left = self.get_hosts_remaining(iterator._play)
|
||||
debug("done getting the remaining hosts for this loop")
|
||||
if len(hosts_left) == 0:
|
||||
|
@ -51,11 +137,13 @@ class StrategyModule(StrategyBase):
|
|||
# queue up this task for each host in the inventory
|
||||
callback_sent = False
|
||||
work_to_do = False
|
||||
for host in hosts_left:
|
||||
while True:
|
||||
task = iterator.get_next_task_for_host(host)
|
||||
|
||||
host_tasks = self._get_next_task_lockstep(hosts_left, iterator)
|
||||
for (host, task) in host_tasks:
|
||||
if not task:
|
||||
break
|
||||
continue
|
||||
|
||||
work_to_do = True
|
||||
|
||||
debug("getting variables")
|
||||
task_vars = self._variable_manager.get_vars(loader=self._loader, play=iterator._play, host=host, task=task)
|
||||
|
@ -74,17 +162,14 @@ class StrategyModule(StrategyBase):
|
|||
debug("'%s' failed tag evaluation" % task)
|
||||
continue
|
||||
|
||||
break
|
||||
|
||||
if not task:
|
||||
continue
|
||||
|
||||
work_to_do = True
|
||||
if task.action == 'meta':
|
||||
# meta tasks store their args in the _raw_params field of args,
|
||||
# since they do not use k=v pairs, so get that
|
||||
meta_action = task.args.get('_raw_params')
|
||||
if meta_action == 'flush_handlers':
|
||||
if meta_action == 'noop':
|
||||
# FIXME: issue a callback for the noop here?
|
||||
continue
|
||||
elif meta_action == 'flush_handlers':
|
||||
self.run_handlers(iterator, connection_info)
|
||||
else:
|
||||
raise AnsibleError("invalid meta action requested: %s" % meta_action, obj=task._ds)
|
||||
|
@ -100,6 +185,11 @@ class StrategyModule(StrategyBase):
|
|||
|
||||
debug("done queuing things up, now waiting for results queue to drain")
|
||||
self._wait_on_pending_results(iterator)
|
||||
|
||||
# FIXME: MAKE PENDING RESULTS RETURN RESULTS PROCESSED AND USE THEM
|
||||
# TO TAKE ACTION, ie. FOR INCLUDE STATEMENTS TO PRESERVE THE
|
||||
# LOCK STEP OPERATION
|
||||
|
||||
debug("results queue empty")
|
||||
except (IOError, EOFError), e:
|
||||
debug("got IOError/EOFError in task loop: %s" % e)
|
||||
|
|
4
v2/samples/include.yml
Normal file
4
v2/samples/include.yml
Normal file
|
@ -0,0 +1,4 @@
|
|||
- debug: msg="this is the include, a=={{a}}"
|
||||
- debug: msg="this is the second debug in the include"
|
||||
- debug: msg="this is the third debug in the include, and a is still {{a}}"
|
||||
|
3
v2/samples/localhosts
Normal file
3
v2/samples/localhosts
Normal file
|
@ -0,0 +1,3 @@
|
|||
l1 ansible_ssh_host=127.0.0.1
|
||||
l2 ansible_ssh_host=127.0.0.2
|
||||
l3 ansible_ssh_host=127.0.0.3
|
0
v2/samples/roles/common/meta/main.yml
Normal file
0
v2/samples/roles/common/meta/main.yml
Normal file
1
v2/samples/roles/common/tasks/main.yml
Normal file
1
v2/samples/roles/common/tasks/main.yml
Normal file
|
@ -0,0 +1 @@
|
|||
- debug: msg="this is a task from the common role"
|
2
v2/samples/roles/role_a/meta/main.yml
Normal file
2
v2/samples/roles/role_a/meta/main.yml
Normal file
|
@ -0,0 +1,2 @@
|
|||
dependencies:
|
||||
- common
|
1
v2/samples/roles/role_a/tasks/main.yml
Normal file
1
v2/samples/roles/role_a/tasks/main.yml
Normal file
|
@ -0,0 +1 @@
|
|||
- debug: msg="this is a task from role A"
|
2
v2/samples/roles/role_b/meta/main.yml
Normal file
2
v2/samples/roles/role_b/meta/main.yml
Normal file
|
@ -0,0 +1,2 @@
|
|||
dependencies:
|
||||
- common
|
1
v2/samples/roles/role_b/tasks/main.yml
Normal file
1
v2/samples/roles/role_b/tasks/main.yml
Normal file
|
@ -0,0 +1 @@
|
|||
- debug: msg="this is a task from role B"
|
2
v2/samples/roles/test_role/meta/main.yml
Normal file
2
v2/samples/roles/test_role/meta/main.yml
Normal file
|
@ -0,0 +1,2 @@
|
|||
dependencies:
|
||||
- test_role_dep
|
|
@ -1 +1,3 @@
|
|||
- debug: msg="here we are in the role, foo={{foo}}"
|
||||
- fail:
|
||||
when: foo != "bar"
|
||||
|
|
1
v2/samples/roles/test_role_dep/tasks/main.yml
Normal file
1
v2/samples/roles/test_role_dep/tasks/main.yml
Normal file
|
@ -0,0 +1 @@
|
|||
- debug: msg="here we are in the role dependency"
|
|
@ -1,17 +1,20 @@
|
|||
- hosts: localhost
|
||||
- hosts: all
|
||||
connection: local
|
||||
gather_facts: no
|
||||
gather_facts: yes
|
||||
tasks:
|
||||
- block:
|
||||
- command: /bin/false
|
||||
- debug: msg="you shouldn't see me"
|
||||
- debug: msg="this is the first task"
|
||||
- fail:
|
||||
when: inventory_hostname == "l2"
|
||||
- debug: msg="only l1 and l3 should see me"
|
||||
rescue:
|
||||
- debug: msg="this is the rescue"
|
||||
- command: /bin/false
|
||||
- debug: msg="you shouldn't see this either"
|
||||
- debug: msg="no host should see this run"
|
||||
always:
|
||||
- debug: msg="this is the always block, it will always be seen"
|
||||
when: foo|default('') != "some value"
|
||||
tags:
|
||||
- foo
|
||||
- bar
|
||||
- debug: msg="you should only see l1 and l3 run this"
|
||||
|
|
8
v2/samples/test_blocks_of_blocks.yml
Normal file
8
v2/samples/test_blocks_of_blocks.yml
Normal file
|
@ -0,0 +1,8 @@
|
|||
- hosts: localhost
|
||||
gather_facts: no
|
||||
tasks:
|
||||
- block:
|
||||
- block:
|
||||
- block:
|
||||
- block:
|
||||
- debug: msg="are we there yet?"
|
26
v2/samples/test_include.yml
Normal file
26
v2/samples/test_include.yml
Normal file
|
@ -0,0 +1,26 @@
|
|||
- hosts: localhost
|
||||
gather_facts: no
|
||||
tasks:
|
||||
- block:
|
||||
- include: include.yml
|
||||
when: 1 == 2
|
||||
- include: include.yml a=1
|
||||
when: 1 == 1
|
||||
notify: foo
|
||||
- include: include.yml a={{item}}
|
||||
with_items:
|
||||
- foo
|
||||
- bar
|
||||
- bam
|
||||
- fail:
|
||||
|
||||
#rescue:
|
||||
#- include: include.yml a=rescue
|
||||
|
||||
always:
|
||||
- include: include.yml a=always
|
||||
|
||||
handlers:
|
||||
- name: foo
|
||||
include: include.yml a="this is a handler"
|
||||
|
|
@ -1,4 +1,5 @@
|
|||
- hosts: ubuntu1404
|
||||
- hosts: localhost
|
||||
connection: local
|
||||
gather_facts: no
|
||||
vars:
|
||||
foo: "BAD!!"
|
||||
|
|
6
v2/samples/test_roles_complex.yml
Normal file
6
v2/samples/test_roles_complex.yml
Normal file
|
@ -0,0 +1,6 @@
|
|||
- hosts: localhost
|
||||
gather_facts: no
|
||||
roles:
|
||||
- { role: role_a, tags: A, when: skip != "A" }
|
||||
- { role: role_b, tags: B, when: skip != "B" }
|
||||
|
Loading…
Reference in a new issue