Logs were broken because ADD_ROUTE was being sent *after* messages
started flowing from the new child. That's an explicit non-goal of the
design, so fix it.
Ansible has no blocking services running, or really any service that
would have an outsized benefit from multiple IO waiters. Probably we
only need 1, but let's start with 2 just in case.
When creating a context using Router.method(via=somechild),
unidirectional mode was set on the new child correctly, however if the
child were to call Router.method(), due to a typing mistake the new
child would start without it.
This doesn't impact the Ansible extension, as only forked tasks are
started directly by children, and they are not responsible for routing
messages.
Add test so it can't happen again.
The previous method of spinning up a transient thread to import the
service pool in a child context could deadlock with use of the importer
on the main thread. Therefore wake the main thread to handle import for
us, and use a regular Receiver to buffer messages to the stub, which is
inherited rather than replaced by the real service pool.
Previously given something like:
l = mitogen.core.Latch()
l.put(1)
l.put(2)
s = mitogen.select.Select([l], oneshot=False)
assert 1 == s.get(block=False)
assert 2 == s.get(block=False)
The second call would throw TimeoutError, because Select.add() only
queued the receiver/latch once if it was non-empty, rather than once for
each item as should happen.
Its functionality was duplicated by _on_broker_exit() somewhere along
the way, and nothing has referred to it in a long time. I have no idea
how this happened.
Merge its docstring into _on_broker_exit() and delete it, remove the
Router "shutdown" signal after confirming it has no users, and move all
the Router-originated error messages together in a block at the top of
the class.
Already covered by router_test.AddHandlerTest.test_dead_message_sent_at_shutdown
Stream.set_protocol() was updated to break the reference on the previous
protocol, to encourage a crash should an old protocol continue operating
after it's not supposed to be active any more.
That broke DelimitedProtocol's protocol switching functionality.
Given a message sent on "ssh.foo" to "mypkg.mymod", instead of logging
it to "mitogen.ctx.ssh.foo" in the master process, with the message
prefixed with the original logger name, instead log it to
"mypkg.mymod.[ssh.foo]", permitting normal logging package filtering
features to work as they usually do.
This also helps tidy up logging output a little bit.
Given:
- Broker asleep in poll()
- thread B calling Latch.put()
Previously,
- B takes lock,
- B wakes socket by dropping GIL and writing to it
- Broker wakes from poll(), acquires GIL only to find Latch._lock is held
- Broker drops GIL, sleeps on futex() for _lock
- B wakes, acquires GIL, releases _lock
- Broker wakes from futex(), acquires lock
Now,
- B takes lock, updates state, releases lock
- B wakes socket by droppping GIL and writing to it
- Broker wakes from poll(), acquires GIL and _lock
- Everyone lives happily ever after.
Given:
- thread A asleep in Latch._get_sleep()
- thread B calling Latch.put()
Previously,
- B takes lock,
- B wakes socket by dropping GIL and writing to it
- A wakes from poll(), acquires GIL only to find Latch._lock is held
- A drops GIL, sleeps on futex() for _lock
- B wakes, acquires GIL, releases _lock
- A wakes from futex(), acquires lock
Now,
- B takes lock, updates state, releases lock
- B wakes socket by droppping GIL and writing to it
- A wakes from poll(), acquires GIL and _lock
- Everyone lives happily ever after.
Move all details of broker/router setup out of connection.py, instead
deferring it to a WorkerModel class exported by process.py via
get_worker_model(). The running strategy can override the configured
worker model via _get_worker_model().
ClassicWorkerModel is installed by default, which implements the
extension's existing process model.
Add optional support for the third party setproctitle module, so
children have pretty names in ps output.
Add optional support for per-CPU multiplexers to classic runs.
Now it's possible for stream.protocol to not refer to MitogenProtocol,
move the signal handler to a MitogenProtocol subclass instead.
Fixes a crash where CTRL+C during child bootstrap would print
AttributeError.
During early initialization under hackbench, it is possible for Broker
to be in LogHandler._send() while the main thread has already destroyed
_buffer. So we must synchronize them, but only while the handler is
corked.
Fix a race where if Stream.on_receive() detects disconnect, it calls
Stream.on_disconnect(), which fires Stream 'disconnect' event, whereas
if BufferedWriter.on_transmit() detects disconnect, it called
Protocol.on_disconnect(), which did not fire the Stream 'disconnect'
event.
Since mitogen.parent listens on Stream's 'disconnect' event to reap
children, this was causing a very difficult to trigger test failure.
Triggered after <1000 runs on a Xeon E5530 with hyperthreading using
hackbench running at the same priority:
$ hackbench -s 1048576 -l 100000000000 -g 4
To ensure a test process can successfully recreate an Ansible
MuxProcess, reset fork-inherited globals during disconnection.
There is basically no good place for this. Per the comments on #91, it
would be far better if the context's identity was tied to its router,
rather than some global variable.
Split Stream into many, many classes
* mitogen.parent.Connection: Handles connection setup logic only.
* Maintain references to stdout and stderr streams.
* Manages TimerList timer to cancel connection attempt after
deadline
* Blocking setup code replaced by async equivalents running on the
broker
* mitogen.parent.Options: Tracks connection-specific options. This
keeps the connection class small, but more importantly, it is
generic to the future desire to build and execute command lines
without starting a full connection.
* mitogen.core.Protocol: Handles program behaviour relating to events
on a stream. Protocol performs no IO of its own, instead deferring
it to Stream and Side. This makes testing much easier, and means
libssh can reimplement Stream and Side to reuse MitogenProtocol
* mitogen.core.MitogenProtocol: Guts of the old Mitogen stream
implementtion
* mitogen.core.BufferedWriter: Guts of the old Mitogen buffered
transmit implementation, made generic
* mitogen.core.DelineatedProtocol: Guts of the old IoLogger, knows how
to split up input and pass it on to a
on_line_received()/on_partial_line_received() callback.
* mitogen.parent.BootstrapProtocol: Asynchronous equivalent of the old
blocking connect code. Waits for various prompts (MITO001 etc) and
writes the bootstrap using a BufferedWriter. On success, switches
the stream to MitogenProtocol.
* mitogen.core.Message: move encoding parts of MitogenProtocol out to
Message (where it belongs) and write a bunch of new tests for
pickling.
* The bizarre Stream.construct() is gone now, Option.__init__ is its
own constructor. Should fix many LGTM errors.
* Update all connection methods: Every connection method is updated to
use async logic, defining protocols as required to handle interactive
prompts like in SSH or su. Add new real integration tests for at least
doas and su.
* Eliminate manual fd management: File descriptors are trapped in file
objects at their point of origin, and Side is updated to use file
objects rather than raw descriptors. This eliminates a whole class of
bugs where unrelated FDs could be closed by the wrong component. Now
an FD's open/closed status is fused to it everywhere in the library.
* Halve file descriptor usage: now FD open/close state is tracked by
its file object, we don't need to duplicate FDs everywhere so that
receive/transmit side can be closed independently. Instead both sides
back on to the same file object. Closes#26, Closes#470.
* Remove most uses of dup/dup2: Closes#256. File descriptors are
trapped in a common file object and shared among classes. The
remaining few uses for dup/dup2 are as close to minimal as possible.
* Introduce mitogen.parent.Process: uniform interface for subprocesses
created either via mitogen.fork or the subprocess module. Remove all
the crap where we steal a pid from subprocess guts. Now we use
subprocess to manage its processes as it should be. Closes#169 by
using the new Timers facility to poll for a slow-to-exit subprocess.
* Fix su password race: Closes#363. DelineatedProtocol naturally
retries partially received lines, preventing the cause of the original
race.
* Delete old blocking IO utility functions
iter_read()/write_all()/discard_until().
Closes#26Closes#147Closes#169Closes#256Closes#363Closes#419Closes#470
David Wilson