diff options
Diffstat (limited to 'doc/development/performance.md')
| -rw-r--r-- | doc/development/performance.md | 48 |
1 files changed, 44 insertions, 4 deletions
diff --git a/doc/development/performance.md b/doc/development/performance.md index 3b59393bae6..e20633a05f6 100644 --- a/doc/development/performance.md +++ b/doc/development/performance.md @@ -1,3 +1,9 @@ +--- +stage: none +group: unassigned +info: To determine the technical writer assigned to the Stage/Group associated with this page, see https://about.gitlab.com/handbook/engineering/ux/technical-writing/#designated-technical-writers +--- + # Performance Guidelines This document describes various guidelines to follow to ensure good and @@ -334,15 +340,49 @@ These results can also be placed into a PostgreSQL database by setting the `RSPEC_PROFILING_POSTGRES_URL` variable. This is used to profile the test suite when running in the CI environment. -We store these results also when running CI jobs on the default branch on -`gitlab.com`. Statistics of these profiling data are [available -online](https://gitlab-org.gitlab.io/rspec_profiling_stats/). For example, -you can find which tests take longest to run or which execute the most +We store these results also when running nightly scheduled CI jobs on the +default branch on `gitlab.com`. Statistics of these profiling data are +[available online](https://gitlab-org.gitlab.io/rspec_profiling_stats/). For +example, you can find which tests take longest to run or which execute the most queries. This can be handy for optimizing our tests or identifying performance issues in our code. ## Memory profiling +We can use two approaches, often in combination, to track down memory issues: + +- Leaving the code intact and wrapping a profiler around it. +- Monitor memory usage of the process while disabling/enabling different parts of the code we suspect could be problematic. + +### Using Memory Profiler + +We can use `memory_profiler` for profiling. + +The [`memory_profiler`](https://github.com/SamSaffron/memory_profiler) gem is already present in GitLab's `Gemfile`, +you just need to require it: + +```ruby +require 'sidekiq/testing' + +report = MemoryProfiler.report do + # Code you want to profile +end + +output = File.open('/tmp/profile.txt','w') +report.pretty_print(output) +``` + +The report breaks down 2 key concepts: + +- Retained: long lived memory use and object count retained due to the execution of the code block. +- Allocated: all object allocation and memory allocation during code block. + +As a general rule, **retained** will always be smaller than or equal to allocated. + +The actual RSS cost will always be slightly higher as MRI heaps are not squashed to size and memory fragments. + +### Rbtrace + One of the reasons of the increased memory footprint could be Ruby memory fragmentation. To diagnose it, you can visualize Ruby heap as described in [this post by Aaron Patterson](https://tenderlovemaking.com/2017/09/27/visualizing-your-ruby-heap.html). |