OpenStack-like notification generator.

It works like this: In OpenStack an operation is a series of notifications connected by a common request_id. For example, to create an instance we would have the following events:

compute.run_instance.start scheduler.run_instance.start scheduler.run_instance.scheduled scheduler.run_instance.end compute.instance.update compute.instance.update compute.instance.update compute.instance.update compute.instance.update compute.instance.update compute.run_instance.end

But, since this is a large system, we could have lots of these operations going on currently. Each operation takes time to perform. So the notifications generated by these operations are interleaved.

This library simulates these many concurrent operations. You tell the EventGenerator how many new operations to start per minute and it will. Note that you'll get a lot more notificationis than the operations/second (since the operations persist over time in the future).

The library tries to avoid bad sequences (like doing an update on a deleted instance), but it can happen in some race conditions.

You can generate in real-time by passing in a real datetime or you can generate the events as fast as possible by incrementing a starting time by the tick amount. See the examples below.

To generate events in real-time ...

g = notigen.EventGenerator(100)  # Number of operations per minute
now = datetime.datetime.utcnow()
start = now
nevents = 0
while nevents < 10000:
    e = g.generate(now)
    if e:
        nevents += len(e)

    now = datetime.datetime.utcnow()

To generate events as fast as possible ...

g = notigen.EventGenerator(100)  # Not really relevant
now = datetime.datetime.utcnow()
start = now
nevents = 0
while nevents < 10000:
    e = g.generate(now)
    if e:
        nevents += len(e)

    now = g.move_to_next_tick(now)

Datetimes are returned as strings as OpenStack would.