Summary

The "failure read" and "failure write" tests (see failure.yml) are flaky for a number of drivers and add little benefit. They basically assert that reads or writes during a failover will fail if retryable reads and retryable writes are not enabled. That test does show that a failover can cause read/write failures, but otherwise does not assert a behavior that anyone actually wants. Asking a driver team to "make the driver fail more often" doesn't make sense, either. Considering the noise it creates, it would be better to remove it.

Motivation

Who is the affected end user?

Drivers devs.

How does this affect the end user?

They are annoyed by intermittent failure notifications.

How likely is it that this problem or use case will occur?

The failure is intermittent. Fails maybe 5% of the time for the Ruby driver and less often for others.

If the problem does occur, what are the consequences and how severe are they?

People get notifications about a failure that they can't do anything about.

Is this issue urgent?

No.

Is this ticket required by a downstream team?

No.

Is this ticket only for tests?

Yes.

Acceptance Criteria

The "failure read" and "failure write" Astrolabe tests are less flaky or gone.