As recently there was back-and-forth with this hardcoded value
(0.001 -> 0.01 -> 0.005), obviousely the optimal value for it depends on
Ansible usage scanario and is better to be configurable.
This patch adds a new config option in DEFAULT section,
`internal_poll_interval`, with default of 0.001 corresponding to the
value hardcoded in Ansible v2.1.
This config option is then used instead of hardcoded values where
needed.
Related GH issue: 14219
which got lost in recent big 'performance improvements' merge by @jimi-c.
I had made a previous PR to fix this, then @bcoca had committed an
improved fix. Now it's lost again.
cf: d2b3b2c03e (lost here)
cf: 25e9b5788b (previous fix)
Earlier PR #14849
Earlier issue #14843
Please note that jimi-c broke this last time as well ... seeing a
pattern here.
We couldn't copy to_unicode, to_bytes, to_str into module_utils because
of licensing. So once created it we had two sets of functions that did
the same things but had different implementations. To remedy that, this
change removes the ansible.utils.unicode versions of those functions.
* dynamic role_include
* more fixes for dynamic include roles
* set play yfrom iterator when dynamic
* changes from jimi-c
* avoid modules that break ad hoc
TODO: should really be a config
for `VariableManager._get_magic_variables()`.
This saves a lot of time re-iterating the nearly always constant global
list of groups and their members.
Generate once and cache, and invalidate cache in case `add_host:` or
`group_by:` are used.
This feature also cleans up and extends the meta subsystem:
* Allows for some meta actions (noop, clear_facts, clear_host_errors,
and end_play) to operate on a per-host basis, meaning they can work
with the free strategy as expected.
* Allows for conditionals on meta tasks.
* Fixes a bug where (for the linear strategy) metas were not treated
as a run_once task, meaning every host in inventory would run the
meta task.
Fixes#1476
* attempt #11 to role_include
* fixes from jimi-c
* do not override load_data, move all to load
* removed debugging
* implemented tasks_from parameter, must break cache
* fixed issue with cache and tasks_from
* make resolution of from_tasks prioritize literal
* avoid role dependency dedupe when include_role
* fixed role deps and handlers are now loaded
* simplified code, enabled k=v parsing
used example from jimi-c
* load role defaults for task when include_role
* fixed issue with from_Tasks overriding all subdirs
* corrected priority order of main candidates
* made tasks_from a more generic interface to roles
* fix block inheritance and handler order
* allow vars: clause into included role
* pull vars already processed vs from raw data
* fix from jimi-c blocks i broke
* added back append for dynamic includes
* only allow for basename in from parameter
* fix for docs when no default
* fixed notes
* added include_role to changelog
Copying the TaskInclude task (which is the parent) before loading the blocks
makes the code much more simple and clean, and fixes a bug introduced during
the performance improvement changes (and specifically the change which moved
things to a single-parent model).
Fixes#17064
Since we introduced static includes in 2.1, this broke the functionality
where a notify could be sent to a named include statement, triggering all
handlers contained within the include. This patch fixes that by adding a
search through the parents of a handler for any TaskIncludes which match.
Fixes#15915
Instead of immediately returning a failed code (indicating a break in
the play execution), we internally 'or' that failure code with the result
(now an integer flag instead of a boolean) so that we can properly handle
the rescue/always portions of blocks and still remember that the break
condition was hit.
Fixes#16937
* when including statically, make sure that all parents were also included
statically (issue #16990)
* properly resolve nested static include paths
* print a message when a file is statically included
Fixes#16990
Due to the way we load plugins, internally to Python there can be issues when
the debug strategy is loaded after the linear strategy. To work around this,
we're changing the import line for the linear strategy to avoid the problem.
Related to #16825
2e003adb added the ability for tasks using any_errors_fatal to fail
when there were unreachable hosts. However that patch used the running
unreachable hosts data rather than the results from the current task,
which causes failures when any run_once or BYPASS_HOST_LOOP task is hit
after an unreachable host causes a failure. This patch corrects that by
using the current set of results to determine if any hosts were
unreachable during the last task only.
Fixesansible/ansible-modules-core#4160
* Instead of rebuilding the handler list all over the place, we now
compile the handlers at the point the play is post-validated so that
the view of the play in the PlayIterator contains the definitive list
* Assign the dep_chain to the handlers as they're compiling, just as we
do for regular tasks
* Clean up the logic used to find a given handler, which is greatly
simplified by the above changes
Fixes#15418
Due to the fact that roles may be instantiated with different sets of
params (multiple inclusions of the same role or via role dependencies),
simply tracking notified handlers by name does not work. This patch
changes the way we track handler notifications by using the handler
object itself instead of just the name, allowing for multiple internal
instances. Normally this would be bad, but we also modify the way we
search for handlers by first looking at the notifying tasks dependency
chain (ensuring that roles find their own handlers first) and then at
the main list of handlers, using the first match it finds.
This patch also modifies the way we setup the internal list of handlers,
which should allow us to correctly identify if a notified handler exists
more easily.
Fixes#15084
This allows the PlaybookExecutor to receive more information regarding
what happened internal to the TaskQueueManager and strategy, to determine
things like whether or not the play iteration should stop.
Fixes#15523
With some earlier changes, continuing to forward failed hosts on
to the iterator with each TQM run() call was causing plays with
max_fail_pct set to fail, as hosts which failed in previous plays
were counting those old failures against the % calculation.
Also changed the linear strategy's calculation to use the internal
failed list, rather than the iterator, as this now represents the
hosts failed during the current run only.
When using run_once, there is only one dict of facts so passing that
to the VariableManager results in the fact cache containing the same
dictionary reference for all hosts in inventory. This patch fixes that
by making sure we pass a copy of the facts dict to VariableManager.
Fixes#14279
In the free strategy, we mark a host as blocked when it has work to do
(the PlayIterator returns a task) to prevent multiple tasks from being sent
to the host. However, we check for role duplicates after setting the blocked
flag, but were not clearing that when the task was skipped leading to an
infinite loop. This patch corrects that by clearing the blocked flag when
the task is skipped.
Fixes#15681
* Don't filter hosts remaining based on their failed state. Instead rely
on the PlayIterator to return None/ITERATING_COMPLETE when the host is
failed.
* In the free strategy, make sure we wait outside the host loop for all
pending results to be processed.
* Use the internal _set_failed_state() instead of manually setting things
when a failed child state is hit
Fixes#15623
This was reinitialized every time we forked before so we weren't sharing
the same Locks. It also was not accounting for modules which were
directly invoked by an action plugin instead of going through the
strategy plguins.
Previously we were first checking the fail/run state of the child
state for tasks/rescue/always portions of the block. Instead we are now
always recursively iterating over the child state and then evaluating
whether the child state is failed or complete before changing the failed/
run state within the current block.
Fixes#14324
If add_host is performed with hostname existing in inventory, but
not yet accessed and put in inventory cache, additional host with
same hostname and different uuid is created, causing patterns to
misbehave.
This makes our recursive, ast.parse performance measures as fast as
pre-ziploader baseline.
Since this unittest isn't testing that the returned module data is
correct we don't need to worry about os.rename not having any module
data. Should devise a separate test for the module and caching code
* Remove workaround for fixed bug.
The bug where PluginLoader required objects to directly inherit from
base_classes has been fixed. Remove workaround from this strategy
plugin Also switched to using super so that we don't have to modify
all of hte code anytime something like that happens.
* These should be to_uniocde because they're being sent to display()
`ansible_failed_task`:
Contains the task data, essentially a serialized view of the Task() object.
`ansible_failed_result`:
Contains the result of the task failure, which is identical in function
to registering the result. Doing so automatically like this saves the user
from having to register every result in a block and then trying to figure
out which result had the failure
Similar to the way try/except/finally work, these variables will not be
available in the `always` portion of a block unless there is a corresponding
`rescue` first.
Fixes#12341
main_q is not used anywhere in the codebase.
It is created in TaskQueueManager._initialize_processes, bundled with rslt_q
into TaskQueueManger._workers, later unwrapped in StrategyBase but not used.
This queue is closed in TaskQueueManger._cleanup_processes.
Historically, it is passed as a init parameter into WorkerProcess,
introduced in 62d7956, but this behavior is changed in 120b9a7.
Signed-off-by: 夏恺(Xia Kai) <xiaket@gmail.com>
- now workers passes queue to task_executor so it can send back events per item and on retry attempt
- updated result class to pass along events to strategy
- base strategy updated to forward new events to callback
- callbacks now remove 'items' on final result but process them directly when invoked per item
- new callback method to deal with retry attempt messages (also now obeys nolog)
- updated tests to match new signature of task_executor
fixes#14558fixes#14072
* Fix the way task_include fields were created and copied
* Have blocks get_dep_chain() look at task_include's blocks for proper
dep chain inheritance
* Fix the way task_include fields are copied to prevent a recursive
degradation
Fixes#14460
* Make sure dep chains are checked recursively for nested blocks
* Fixing iterator is_failed() check to make sure we're not in a
rescue block before returning True
* Use is_failed() to test whether a host should be added to the TQM
failed_hosts list
* Use is_failed() when compiling the list of hosts left to iterate
over in both the linear and free strategies
Fixes#14222
* Raise an error if the action is using BYPASS_HOST_LOOP, to prevent
unexpected behavior from those actions
* Show a warning regarding tasks marked as run_once, as the free strategy
does not yet support that behavior
* Minor tweak to linear strategies run_once code to make sure we don't
raise an error if an action isn't found
* Fixed a bug in PlayIterator when ITERATING_ALWAYS, where the block
was advanced but the incorrect data structure elements were cleared
* Cleaned up the logic of is_failed() in PlayIterator
* Fixed a bug in the free strategy which had not been updated to use
the base strategy _execute_meta() method
* Stopped strategies from using is_failed() to determine if tasks should
still be fetched for a host
Fixes#14040
Instead of bombing out of the strategy, we now properly mark hosts failed
so that the play iterator can handle block rescue/always properly.
Fixes#14024
* Added additional methods to the iterator code to assess host failures
while also taking into account the block rescue/always states
* Fixed bugs in the free strategy, where results were not always being
processed after being collected
* Added some prettier printing to the state output from iterator
Fixes#13699
Also fixes a bug where we were passing an incorrect number of parameters to
_do_handler_run() when processing an include file in a handler task/block.
Fixes#13560
James Cammarata
Brian Coca
Pavlo Shchelokovskyy
Brian Coca
Tobias Wolf
Toshio Kuratomi
Michael Scherer
jctanner
Remi Ferrand
Andrey Shestakov
Vincent Roy
Toshio Kuratomi
Nikolay Sokolov
Kishin Yagami
Brian Coca
qqo
夏恺(Xia Kai)
Felix Fontein
Selivanov Pavel
Tomasz Kontusz