Skip to content

Draft: Add guidelines regarding the preferred data read per query execution

Gregory Havenga requested to merge ghavenga-query-quidelines-update into master

What does this MR do and why?

Update our database guidelines relating to the quantity of data read per query execution to prevent endangering the operation of GitLab instances by over-contending database resources.

The 50MB limit proposed is arbitrarily decided as a reasonable upper bound, but this could be altered.

This kind of problem is becoming more common as namespace traversal queries usually force a substantial project table read, in addition to the information read per project to fufill the need, such as !130582 (comment 1553941765).

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 Gregory Havenga

Merge request reports