Skip to content

Conversation

patrickfreed
Copy link
Contributor

RUST-812

This PR makes the following changes to hopefully eliminate the remaining flakiness in the in_window::load_balancing_test, which verifies that the operation-count tracking server selection algorithm successfully routes requests away from busy servers:

  • Add a 30 second local threshold config to the client, ensuring both mongoses are always in the latency window
  • Use SDAM monitoring events to ensure both mongoses are discovered before starting the test

From testing, it seems that this (in conjunction with skipping on auth + tls) eliminates the random failures, but we'll have to wait and see if they crop up again. Should these changes prove successful, I'll update the spec to require them so other drivers can avoid similar errors (DRIVERS-1926)

@patrickfreed patrickfreed marked this pull request as ready for review February 1, 2022 17:19
@patrickfreed
Copy link
Contributor Author

patrickfreed commented Feb 1, 2022

The lint errors appear unrelated + are fixed on master already

Copy link
Contributor

@abr-egn abr-egn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM!

@patrickfreed patrickfreed merged commit 015c198 into mongodb:master Feb 4, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants