Skip to content

Evaluate idx_security_policy_dismissals_project_findings_uuids index performance

Everyone can contribute. Help move this issue forward while earning points, leveling up and collecting rewards.

We need to evaluate the performance of the GIN index on security_findings_uuids in the security_policy_dismissals table when we have more production data. This follows up on this discussion.

During the implementation of MR !207638, we added a query that uses the GIN index on security_findings_uuids:

SELECT DISTINCT unnest(security_findings_uuids) FROM "security_policy_dismissals" WHERE "security_policy_dismissals"."project_id" = 74981975 AND (security_findings_uuids && ARRAY['65b9693b-49c4-56e9-a9e3-2ed49b86b1a4', 'f4129fe1-7401-5d67-b36f-1382ca8fe06c']::text[])

At the time of implementation, the security_policy_dismissals table had only 21 records in production, making it difficult to properly evaluate index performance.

  • Initial query plan: https://console.postgres.ai/shared/bfb3baa5-c377-43c9-975f-8c26ed5617bf
  • Table growth expectation: Moderate growth as it's used for dismissing policies in warn mode during MR evaluation
Edited by 🤖 GitLab Bot 🤖