fix(tests): fix flaky test #1972
Conversation
There was a problem hiding this comment.
Very interesting, thanks for it!
Maybe this will fix the next: #1826
| var messages1: seq[WakuMessage] = @[] | ||
| var messages2: seq[WakuMessage] = @[] |
There was a problem hiding this comment.
I think that is equivalent:
| var messages1: seq[WakuMessage] = @[] | |
| var messages2: seq[WakuMessage] = @[] | |
| var messages1 = @[WakuMessage] | |
| var messages2 = @[WakuMessage] |
|
You can find the image built from this PR at |
| await nodes[0].publish(pubsubTopics[0], message1) | ||
| var message = WakuMessage(payload: ("Payload_" & $i).toBytes(), contentTopic: contentTopics[0]) | ||
| doAssert(nodes[0].wakuRlnRelay.appendRLNProof(message, epochTime())) | ||
| messages1.add(message) |
There was a problem hiding this comment.
I am not sure what the difference is - does appendRLNProof somehow "cache" the proof when called in the same epoch? Because it seems like the only difference is in when does the "await" happen - here vs. a bit later, but the proof generation does not really change. Or what am I missing?
There was a problem hiding this comment.
mm as it was before:
- await generate proof
- await publish message
- await generate proof
- await publish message
the problem is that generate proof takes a bit of time. If this time grows > Epoch, then the second published message wont be considered spam, since enough time passed so you are not rate limited. This was the reason the test were failing. I was expecting only 1 message to arrive, but 2 arrives, since the second was sent too late, hence skipping the rate limit.
There was a problem hiding this comment.
So would it make sense to wrap this in an async proc and execute the proof generation in parallel and then just await all futures? I feel like it's highly probable that if the proof generation takes a lot of time, we could still hit the "too slow" issue this way (even though with lower probability)
There was a problem hiding this comment.
you are very right, but found an easier solution. "paralellizing" the proof generation would imply more chanes, since its not async, buttt just realized we can use the same epoch time for all of them. so with this it shouldn't matter that proof generation takes 1 hour, since all messages are using the same epoch.
| await nodes[0].publish(pubsubTopics[0], message1) | ||
| var message = WakuMessage(payload: ("Payload_" & $i).toBytes(), contentTopic: contentTopics[0]) | ||
| doAssert(nodes[0].wakuRlnRelay.appendRLNProof(message, epochTime())) | ||
| messages1.add(message) |
testing rln-relay is applied in all rln pubsub/content topics.