fixed load/users getting distributed to missing worker

[radhakrishnaakamat]

fixed load/users getting distributed to the missing workers.
I am new to open source commit and python.
Please let me know if anything needs to be changed.
fixes #2008 #2010

[cyberw]

Can you add a unit test? You can override/shorten the timeout in the tests to reduce the time they take by using pytests mock feature (there should be some examples among the tests)

[radhakrishnaakamat]

It asserts to Count of active workers in dispatch.. Here the missing worker should be removed after heartdead interval.. Added comment to code and added more sensible message in assert

…

On Sat, 12 Feb, 2022, 19:46 Lars Holmberg, ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In locust/test/test_dispatch.py <#2010 (comment)>: > + weight = 1 + + class User3(User): + weight = 1 + + user_classes = [User1, User2, User3] + + worker_nodes = [WorkerNode(str(i + 1)) for i in range(3)] + worker_nodes[1].heartbeat = HEARTBEAT_DEAD - 1 + + users_dispatcher = UsersDispatcher(worker_nodes=worker_nodes, user_classes=user_classes) + + users_dispatcher.new_dispatch(target_user_count=9, spawn_rate=3) + users_dispatcher._wait_between_dispatch = 0 + users_dispatcher.prepare_rebalance() + list(users_dispatcher) what does this line do? :) — Reply to this email directly, view it on GitHub <#2010 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJXIJ6SAZXB7U6XXLQS2LHLU2ZTUVANCNFSM5OFSJBIA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[cyberw]

import from runners instead?

[radhakrishnaakamat]

I tried that, but was getting error during initialization.
Will try the reverse tomorrow..

Could you help me with any resource on how to setup debugging in distributed mode.. That would be really helpful for me to contribute in a better way.. I am resorting to print statement for debugging(This is my first python project)

[cyberw]

Sorry, I dont have much to contribute with debugging distributed. You know how to run an individual unit test, right?

[radhakrishnaakamat]

Yes, initially used to run tox.. later found out the way to run individually.

[cyberw]

Ignore the python 3.6 failure, that test is just flaky I think...

[mboutet]

Hi @radhakrishnaakamat, can you link this PR to the corresponding issue #2008?

[radhakrishnaakamat]

@mboutet thanks for the pointers in #2008, I was able to fix it in the right places(or at least I feel :-P ) in a better way.
@cyberw could you kindly review it once again, I have updated the tests and tox worked fine too. Thanks!!

[cyberw]

👍👍