dagpool – Dependency-Driven Greenthreads

Rationale

The dagpool module provides the DAGPool class, which addresses situations in which the value produced by one greenthread might be consumed by several others – while at the same time a consuming greenthread might depend on the output from several different greenthreads.

If you have a tree with strict many-to-one dependencies – each producer greenthread provides results to exactly one consumer, though a given consumer may depend on multiple producers – that could be addressed by recursively constructing a GreenPool of producers for each consumer, then waiting for all producers.

If you have a tree with strict one-to-many dependencies – each consumer greenthread depends on exactly one producer, though a given producer may provide results to multiple consumers – that could be addressed by causing each producer to finish by launching a GreenPool of consumers.

But when you have many-to-many dependencies, a tree doesn’t suffice. This is known as a Directed Acyclic Graph, or DAG.

You might consider sorting the greenthreads into dependency order (topological sort) and launching them in a GreenPool. But the concurrency of the GreenPool must be strictly constrained to ensure that no greenthread is launched before all its upstream producers have completed – and the appropriate pool size is data-dependent. Only a pool of size 1 (serializing all the greenthreads) guarantees that a topological sort will produce correct results.

Even if you do serialize all the greenthreads, how do you pass results from each producer to all its consumers, which might start at very different points in time?

One answer is to associate each greenthread with a distinct key, and store its result in a common dict. Then each consumer greenthread can identify its direct upstream producers by their keys, and find their results in that dict.

This is the essence of DAGPool.

A DAGPool instance owns a dict, and stores greenthread results in that dict. You spawn all greenthreads in the DAG, specifying for each its own key – the key with which its result will be stored on completion – plus the keys of the upstream producer greenthreads on whose results it directly depends.

Keys need only be unique within the DAGPool instance; they need not be UUIDs. A key can be any type that can be used as a dict key. String keys make it easier to reason about a DAGPool’s behavior, but are by no means required.

The DAGPool passes to each greenthread an iterable of (key, value) pairs. The key in each pair is the key of one of the greenthread’s specified upstream producers; the value is the value returned by that producer greenthread. Pairs are delivered in the order results become available; the consuming greenthread blocks until the next result can be delivered.

Tutorial

Example

Consider a couple of programs in some compiled language that depend on a set of precompiled libraries. Suppose every such build requires as input the specific set of library builds on which it directly depends.

a  zlib
| /  |
|/   |
b    c
|   /|
|  / |
| /  |
|/   |
d    e

We can’t run the build for program d until we have the build results for both b and c. We can’t run the build for library b until we have build results for a and zlib. We can, however, immediately run the builds for a and zlib.

So we can use a DAGPool instance to spawn greenthreads running a function such as this:

def builder(key, upstream):
    for libname, product in upstream:
        # ... configure build for 'key' to use 'product' for 'libname'
    # all upstream builds have completed
    # ... run build for 'key'
    return build_product_for_key

spawn all these greenthreads:

pool = DAGPool()
# the upstream producer keys passed to spawn() can be from any iterable,
# including a generator
pool.spawn("d", ("b", "c"), builder)
pool.spawn("e", ["c"], builder)
pool.spawn("b", ("a", "zlib"), builder)
pool.spawn("c", ["zlib"], builder)
pool.spawn("a", (), builder)

As with eventlet.spawn(), if you need to pass special build flags to some set of builds, these can be passed as either positional or keyword arguments:

def builder(key, upstream, cflags="", linkflags=""):
    ...

pool.spawn("d", ("b", "c"), builder, "-o2")
pool.spawn("e", ["c"], builder, linkflags="-pie")

However, if the arguments to each builder() call are uniform (as in the original example), you could alternatively build a dict of the dependencies and call spawn_many():

deps = dict(d=("b", "c"),
            e=["c"],
            b=("a", "zlib"),
            c=["zlib"],
            a=())
pool.spawn_many(deps, builder)

From outside the DAGPool, you can obtain the results for d and e (or in fact for any of the build greenthreads) in any of several ways.

pool.waitall() waits until the last of the spawned greenthreads has completed, and returns a dict containing results for all of them:

final = pool.waitall()
print("for d: {0}".format(final["d"]))
print("for e: {0}".format(final["e"]))

waitall() is an alias for wait() with no arguments:

final = pool.wait()
print("for d: {0}".format(final["d"]))
print("for e: {0}".format(final["e"]))

Or you can specifically wait for only the final programs:

final = pool.wait(["d", "e"])

The returned dict will contain only the specified keys. The keys may be passed into wait() from any iterable, including a generator.

You can wait for any specified set of greenthreads; they need not be topologically last:

# returns as soon as both a and zlib have returned results, regardless of
# what else is still running
leaves = pool.wait(["a", "zlib"])

Suppose you want to wait specifically for just one of the final programs:

final = pool.wait(["d"])
dprog = final["d"]

The above wait() call will return as soon as greenthread d returns a result – regardless of whether greenthread e has finished.

__getitem()__ is shorthand for obtaining a single result:

# waits until greenthread d returns its result
dprog = pool["d"]

In contrast, get() returns immediately, whether or not a result is ready:

# returns immediately
if pool.get("d") is None:
    ...

Of course, your greenthread might not include an explicit return statement and hence might implicitly return None. You might have to test some other value.

# returns immediately
if pool.get("d", "notdone") == "notdone":
    ...

Suppose you want to process each of the final programs in some way (upload it?), but you don’t want to have to wait until they’ve both finished. You don’t have to poll get() calls – use wait_each():

for key, result in pool.wait_each(["d", "e"]):
    # key will be d or e, in completion order
    # process result...

As with wait(), if you omit the argument to wait_each(), it delivers results for all the greenthreads of which it’s aware:

for key, result in pool.wait_each():
    # key will be a, zlib, b, c, d, e, in whatever order each completes
    # process its result...

Introspection

Let’s say you have set up a DAGPool with the dependencies shown above. To your consternation, your waitall() call does not return! The DAGPool instance is stuck!

You could change waitall() to wait_each(), and print each key as it becomes available:

for key, result in pool.wait_each():
    print("got result for {0}".format(key))
    # ... process ...

Once the build for a has completed, this produces:

got result for a

and then stops. Hmm!

You can check the number of running greenthreads:

>>> print(pool.running())
4

and the number of waiting greenthreads:

>>> print(pool.waiting())
4

It’s often more informative to ask which greenthreads are still running:

>>> print(pool.running_keys())
('c', 'b', 'e', 'd')

but in this case, we already know a has completed.

We can ask for all available results:

>>> print(pool.keys())
('a',)
>>> print(pool.items())
(('a', result_from_a),)

The keys() and items() methods only return keys and items for which results are actually available, reflecting the underlying dict.

But what’s blocking the works? What are we waiting for?

>>> print(pool.waiting_for("d"))
set(['c', 'b'])

(waiting_for()’s optional argument is a single key.)

That doesn’t help much yet...

>>> print(pool.waiting_for("b"))
set(['zlib'])
>>> print(pool.waiting_for("zlib"))
KeyError: 'zlib'

Aha! We forgot to even include the zlib build when we were originally configuring this DAGPool!

(For non-interactive use, it would be more informative to omit waiting_for()’s argument. This usage returns a dict indicating, for each greenthread key, which other keys it’s waiting for.)

from pprint import pprint
pprint(pool.waiting_for())

{'b': set(['zlib']), 'c': set(['zlib']), 'd': set(['b', 'c']), 'e': set(['c'])}

In this case, a reasonable fix would be to spawn the zlib greenthread:

pool.spawn("zlib", (), builder)

Even if this is the last method call on this DAGPool instance, it should unblock all the rest of the DAGPool greenthreads.

Posting

If we happen to have zlib build results in hand already, though, we could instead post() that result instead of rebuilding the library:

pool.post("zlib", result_from_zlib)

This, too, should unblock the rest of the DAGPool greenthreads.

Preloading

If rebuilding takes nontrivial realtime, it might be useful to record partial results, so that in case of interruption you can restart from where you left off rather than having to rebuild everything prior to that point.

You could iteratively post() those prior results into a new DAGPool instance; alternatively you can preload the DAGPool from an existing dict:

pool = DAGPool(dict(a=result_from_a, zlib=result_from_zlib))

Any DAGPool greenthreads that depend on either a or zlib can immediately consume those results.

It also works to construct DAGPool with an iterable of (key, result) pairs.

Exception Propagation

But what if we spawn a zlib build that fails? Suppose the zlib greenthread terminates with an exception? In that case none of b, c, d or e can proceed! Nor do we want to wait forever for them.

dprog = pool["d"]
eventlet.dagpool.PropagateError: PropagateError(d): PropagateError: PropagateError(c): PropagateError: PropagateError(zlib): OriginalError

DAGPool provides a PropagateError exception specifically to wrap such failures. If a DAGPool greenthread terminates with an Exception subclass, the DAGPool wraps that exception in a PropagateError instance whose key attribute is the key of the failing greenthread and whose exc attribute is the exception that terminated it. This PropagateError is stored as the result from that greenthread.

Attempting to consume the result from a greenthread for which a PropagateError was stored raises that PropagateError.

pool["zlib"]
eventlet.dagpool.PropagateError: PropagateError(zlib): OriginalError

Thus, when greenthread c attempts to consume the result from zlib, the PropagateError for zlib is raised. Unless the builder function for greenthread c handles that PropagateError exception, that greenthread will itself terminate. That PropagateError will be wrapped in another PropagateError whose key attribute is c and whose exc attribute is the PropagateError for zlib.

Similarly, when greenthread d attempts to consume the result from c, the PropagateError for c is raised. This in turn is wrapped in a PropagateError whose key is d and whose exc is the PropagateError for c.

When someone attempts to consume the result from d, as shown above, the PropagateError for d is raised.

You can programmatically chase the failure path to determine the original failure if desired:

orig_err = err
key = "unknown"
while isinstance(orig_err, PropagateError):
    key = orig_err.key
    orig_err = orig_err.exc

Scanning for Success / Exceptions

Exception propagation means that we neither perform useless builds nor wait for results that will never arrive.

However, it does make it difficult to obtain partial results for builds that did succeed.

For that you can call wait_each_success():

for key, result in pool.wait_each_success():
    print("{0} succeeded".format(key))
    # ... process result ...

a succeeded

Another problem is that although five different greenthreads failed in the example, we only see one chain of failures. You can enumerate the bad news with wait_each_exception():

for key, err in pool.wait_each_exception():
    print("{0} failed with {1}".format(key, err.exc.__class__.__name__))

c failed with PropagateError
b failed with PropagateError
e failed with PropagateError
d failed with PropagateError
zlib failed with OriginalError

wait_each_exception() yields each PropagateError wrapper as if it were the result, rather than raising it as an exception.

Notice that we print err.exc.__class__.__name__ because err.__class__.__name__ is always PropagateError.

Both wait_each_success() and wait_each_exception() can accept an iterable of keys to report:

for key, result in pool.wait_each_success(["d", "e"]):
    print("{0} succeeded".format(key))

(no output)

for key, err in pool.wait_each_exception(["d", "e"]):
    print("{0} failed with {1}".format(key, err.exc.__class__.__name__))

e failed with PropagateError
d failed with PropagateError

Both wait_each_success() and wait_each_exception() must wait until the greenthreads for all specified keys (or all keys) have terminated, one way or the other, because of course we can’t know until then how to categorize each.

Module Contents

exception eventlet.dagpool.Collision

DAGPool raises Collision when you try to launch two greenthreads with the same key, or post() a result for a key corresponding to a greenthread, or post() twice for the same key. As with KeyError, str(collision) names the key in question.

class eventlet.dagpool.DAGPool(preload={})

A DAGPool is a pool that constrains greenthreads, not by max concurrency, but by data dependencies.

This is a way to implement general DAG dependencies. A simple dependency tree (flowing in either direction) can straightforwardly be implemented using recursion and (e.g.) GreenThread.imap(). What gets complicated is when a given node depends on several other nodes as well as contributing to several other nodes.

With DAGPool, you concurrently launch all applicable greenthreads; each will proceed as soon as it has all required inputs. The DAG is implicit in which items are required by each greenthread.

Each greenthread is launched in a DAGPool with a key: any value that can serve as a Python dict key. The caller also specifies an iterable of other keys on which this greenthread depends. This iterable may be empty.

The greenthread callable must accept (key, results), where:

key
is its own key
results
is an iterable of (key, value) pairs.

A newly-launched DAGPool greenthread is entered immediately, and can perform any necessary setup work. At some point it will iterate over the (key, value) pairs from the passed ‘results’ iterable. Doing so blocks the greenthread until a value is available for each of the keys specified in its initial dependencies iterable. These (key, value) pairs are delivered in chronological order, not the order in which they are initially specified: each value will be delivered as soon as it becomes available.

The value returned by a DAGPool greenthread becomes the value for its key, which unblocks any other greenthreads waiting on that key.

If a DAGPool greenthread terminates with an exception instead of returning a value, attempting to retrieve the value raises PropagateError, which binds the key of the original greenthread and the original exception. Unless the greenthread attempting to retrieve the value handles PropagateError, that exception will in turn be wrapped in a PropagateError of its own, and so forth. The code that ultimately handles PropagateError can follow the chain of PropagateError.exc attributes to discover the flow of that exception through the DAG of greenthreads.

External greenthreads may also interact with a DAGPool. See wait_each(), waitall(), post().

It is not recommended to constrain external DAGPool producer greenthreads in a GreenPool: it may be hard to provably avoid deadlock.

__init__(preload={})

DAGPool can be prepopulated with an initial dict or iterable of (key, value) pairs. These (key, value) pairs are of course immediately available for any greenthread that depends on any of those keys.

__getitem__(key)

__getitem__(key) (aka dagpool[key]) blocks until key has a value, then delivers that value.

get(key, default=None)

get() returns the value for key. If key does not yet have a value, get() returns default.

items()

Return a snapshot tuple of currently-available (key, value) pairs.

keys()

Return a snapshot tuple of keys for which we currently have values.

kill(key)

Kill the greenthread that was spawned with the specified key.

If no such greenthread was spawned, raise KeyError.

post(key, value, replace=False)

post(key, value) stores the passed value for the passed key. It then causes each greenthread blocked on its results iterable, or on wait_each(keys), to check for new values. A waiting greenthread might not literally resume on every single post() of a relevant key, but the first post() of a relevant key ensures that it will resume eventually, and when it does it will catch up with all relevant post() calls.

Calling post(key, value) when there is a running greenthread with that same key raises Collision. If you must post(key, value) instead of letting the greenthread run to completion, you must first call kill(key).

The DAGPool implicitly post()s the return value from each of its greenthreads. But a greenthread may explicitly post() a value for its own key, which will cause its return value to be discarded.

Calling post(key, value, replace=False) (the default replace) when a value for that key has already been posted, by any means, raises Collision.

Calling post(key, value, replace=True) when a value for that key has already been posted, by any means, replaces the previously-stored value. However, that may make it complicated to reason about the behavior of greenthreads waiting on that key.

After a post(key, value1) followed by post(key, value2, replace=True), it is unspecified which pending wait_each([key...]) calls (or greenthreads iterating over results involving that key) will observe value1 versus value2. It is guaranteed that subsequent wait_each([key...]) calls (or greenthreads spawned after that point) will observe value2.

A successful call to post(key, PropagateError(key, ExceptionSubclass)) ensures that any subsequent attempt to retrieve that key’s value will raise that PropagateError instance.

running()

Return number of running DAGPool greenthreads. This includes greenthreads blocked while iterating through their results iterable, that is, greenthreads waiting on values from other keys.

running_keys()

Return keys for running DAGPool greenthreads. This includes greenthreads blocked while iterating through their results iterable, that is, greenthreads waiting on values from other keys.

spawn(key, depends, function, *args, **kwds)

Launch the passed function(key, results, ...) as a greenthread, passing it:

  • the specified key
  • an iterable of (key, value) pairs
  • whatever other positional args or keywords you specify.

Iterating over the results iterable behaves like calling wait_each(depends).

Returning from function() behaves like post(key, return_value).

If function() terminates with an exception, that exception is wrapped in PropagateError with the greenthread’s key and (effectively) posted as the value for that key. Attempting to retrieve that value will raise that PropagateError.

Thus, if the greenthread with key ‘a’ terminates with an exception, and greenthread ‘b’ depends on ‘a’, when greenthread ‘b’ attempts to iterate through its results argument, it will encounter PropagateError. So by default, an uncaught exception will propagate through all the downstream dependencies.

If you pass spawn() a key already passed to spawn() or post(), spawn() raises Collision.

spawn_many(depends, function, *args, **kwds)

spawn_many() accepts a single function whose parameters are the same as for spawn().

The difference is that spawn_many() accepts a dependency dict depends. A new greenthread is spawned for each key in the dict. That dict key’s value should be an iterable of other keys on which this greenthread depends.

If the depends dict contains any key already passed to spawn() or post(), spawn_many() raises Collision. It is indeterminate how many of the other keys in depends will have successfully spawned greenthreads.

wait(keys=<object object>)

keys is an optional iterable of keys. If you omit the argument, it waits for all the keys from preload data, from post() calls and from spawn() calls: in other words, all the keys of which this DAGPool is aware.

wait() blocks the calling greenthread until all of the relevant keys have values. wait() returns a dict whose keys are the relevant keys, and whose values come from the preload data, from values returned by DAGPool greenthreads or from post() calls.

If a DAGPool greenthread terminates with an exception, wait() will raise PropagateError wrapping that exception. If more than one greenthread terminates with an exception, it is indeterminate which one wait() will raise.

If an external greenthread posts a PropagateError instance, wait() will raise that PropagateError. If more than one greenthread posts PropagateError, it is indeterminate which one wait() will raise.

See also wait_each_success(), wait_each_exception().

wait_each(keys=<object object>)

keys is an optional iterable of keys. If you omit the argument, it waits for all the keys from preload data, from post() calls and from spawn() calls: in other words, all the keys of which this DAGPool is aware.

wait_each() is a generator producing (key, value) pairs as a value becomes available for each requested key. wait_each() blocks the calling greenthread until the next value becomes available. If the DAGPool was prepopulated with values for any of the relevant keys, of course those can be delivered immediately without waiting.

Delivery order is intentionally decoupled from the initial sequence of keys: each value is delivered as soon as it becomes available. If multiple keys are available at the same time, wait_each() delivers each of the ready ones in arbitrary order before blocking again.

The DAGPool does not distinguish between a value returned by one of its own greenthreads and one provided by a post() call or preload data.

The wait_each() generator terminates (raises StopIteration) when all specified keys have been delivered. Thus, typical usage might be:

for key, value in dagpool.wait_each(keys):
    # process this ready key and value
# continue processing now that we've gotten values for all keys

By implication, if you pass wait_each() an empty iterable of keys, it returns immediately without yielding anything.

If the value to be delivered is a PropagateError exception object, the generator raises that PropagateError instead of yielding it.

See also wait_each_success(), wait_each_exception().

wait_each_exception(keys=<object object>)

wait_each_exception() filters results so that only exceptions are yielded. Not every provided (or defaulted) key will necessarily be represented, though naturally the generator will not finish until all have completed.

Unlike other DAGPool methods, wait_each_exception() simply yields PropagateError instances as values rather than raising them.

In all other respects, wait_each_exception() behaves like wait_each().

wait_each_success(keys=<object object>)

wait_each_success() filters results so that only success values are yielded. In other words, unlike wait_each(), wait_each_success() will not raise PropagateError. Not every provided (or defaulted) key will necessarily be represented, though naturally the generator will not finish until all have completed.

In all other respects, wait_each_success() behaves like wait_each().

waitall()

waitall() blocks the calling greenthread until there is a value for every DAGPool greenthread launched by spawn(). It returns a dict containing all preload data, all data from post() and all values returned by spawned greenthreads.

See also wait().

waiting()

Return number of waiting DAGPool greenthreads, that is, greenthreads still waiting on values from other keys. This explicitly does not include external greenthreads waiting on wait(), waitall(), wait_each().

waiting_for(key=<object object>)

waiting_for(key) returns a set() of the keys for which the DAGPool greenthread spawned with that key is still waiting. If you pass a key for which no greenthread was spawned, waiting_for() raises KeyError.

waiting_for() without argument returns a dict. Its keys are the keys of DAGPool greenthreads still waiting on one or more values. In the returned dict, the value of each such key is the set of other keys for which that greenthread is still waiting.

This method allows diagnosing a “hung” DAGPool. If certain greenthreads are making no progress, it’s possible that they are waiting on keys for which there is no greenthread and no post() data.

exception eventlet.dagpool.PropagateError(key, exc)

When a DAGPool greenthread terminates with an exception instead of returning a result, attempting to retrieve its value raises PropagateError.

Attributes:

key
the key of the greenthread which raised the exception
exc
the exception object raised by the greenthread