Skip to content

Cache last commit id for path

What does this MR do?

To speed up the Repository#last_commit_for_path, caching the last commit id for path.

Benchmark

Script

require "benchmark/ips"

Benchmark.ips do |x|
  x.config(time: 5, warmup: 2)

  project = Project.find_with_namespace("gitlab-org/gitlab-ce")
  repository = project.repository
  commit = repository.commit("master")

  x.report("cache_last_commit_id_for_path") do
    repository.cache_last_commit_id_for_path(commit.id, nil)
  end

  x.report("last_commit_id_for_path") do
    repository.last_commit_id_for_path(commit.id, nil)
  end

  x.compare!
end

Result

cache_last_commit_id_for_path
                          2.449k (±16.8%) i/s -     12.105k
last_commit_id_for_path
                        110.703  (± 5.4%) i/s -    561.000

Comparison:
cache_last_commit_id_for_path:     2449.2 i/s
last_commit_id_for_path:      110.7 i/s - 22.12x slower

Are there points in the code the reviewer needs to double check?

  • Including the commit id in cache key, so it is not necessary to clear the cache manually.
  • Cache data increase by this MR.

Why was this MR needed?

Showing repository files is very slow.

The Repository#last_commit_for_path is called by the number of contents in RefsController#logs_tree.

Does this MR meet the acceptance criteria?

Merge request reports

Loading