Summary

Ensure server selection does not bias towards overloaded an server.

Motivation

The server ingress rate limiter plans to reject excess operations quickly when overloaded. In certain cases this will be problematic for our power of 2 random choices server selection algorithm based on operationCount (implemented in SPEC-1555) because it relies on the assumption that request latency will go up during overload. However, in this case request latency will go down which can lead to a lower operationCount on the overloaded server. The end result will be increased error rates as new requests are bias towards the already overloaded server.

Who is the affected end user?

Customers.

How does this affect the end user?

Potential for higher error rates during overload.

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

Likely during partial overload, eg when only 1 out of 3 mongoses are overloaded.

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

Higher error rates during overload. Longer time to recovery.

Is this issue urgent?

Does this ticket have a required timeline? What is it?

Is this ticket required by a downstream team?

Needed by e.g. Atlas, Shell, Compass?

Is this ticket only for tests?

No.

Acceptance Criteria

What specific requirements must be met to consider the design phase complete?