Skip to content

Speed up usage_data_metrics_spec

Sylvester Chin requested to merge sc1-fix-slow-usage_data_metrics_spec into master

What does this MR do and why?

This MR adds stubs for an expensive metric aggregation which is not required as the spec checks for the keys/schema of the uncached_data rather than its individual key's value.

Due to !122840 (merged), there are more metric events to be computed. This increases the spec runtime. Many jobs like https://gitlab.com/gitlab-org/gitlab/-/jobs/4664900896 have timed out.

The feature flag use_metric_definitions_for_events_list can be disabled to resolve the problem but this is not feasible since it will be removed eventually to use the new logic path.

In the before flamegraph, bulk of the calls came from lib/gitlab/usage/metrics/instrumentations/aggregated_metric.rb and lib/gitlab/usage/metrics/instrumentations/redis_hll_metric.rb.

Screenshots or screen recordings

Screenshots are required for UI changes, and strongly recommended for all other merge requests.

Before After
2023-07-17-9p49pVx3tL.flamegraph.svg 2023-07-17-w2MAOxsVnp.flamegraph.svg

How to set up and validate locally

  1. Run time bin/rspec spec/lib/gitlab/usage_data_metrics_spec.rb on master branch and this branch.

MR acceptance checklist

This checklist encourages us to confirm any changes have been analyzed to reduce risks in quality, performance, reliability, security, and maintainability.

Edited by Sylvester Chin

Merge request reports

Loading