Sha256: dc0964fd58a14fbcc65d145f4fc061c0352b0b4f772f4b01e4adf1efa10edefb
Contents?: true
Size: 1.91 KB
Versions: 36
Compression:
Stored size: 1.91 KB
Contents
= How to profile a Decidim app The development_app includes a bunch of gems that profile the application. Run the following command in the decidim root's folder: [source,bash] ---- bundle exec rake development_app ---- and then move into it and boot the server [source,bash] ---- cd development_app bundle exec rails s ---- == Bullet Bullet detects N+1 queries and suggests how to fix them, although it doesn't catch them all. It's currently configured in `config/initializers/bullet.rb` to log in the regular rails log and also in its own `log/bullet.log`. You'll now see entries like the following: [source,bash] ---- user: xxx GET / USE eager loading detected Decidim::Comments::Comment => [:author] Add to your query: .includes([:author]) Call stack ---- It also warns you when there's an unnecessary eager load. More details: https://github.com/flyerhzm/bullet == Rack-mini-profiler This gem can analyze memory, database, and call stack with flamegraphs. It will show up in development on the top left corner and it gives you all sorts of profiling insights about that page. It'll tell you where the response time was spend on in the call stack. This gem is further enhanced with the `flamegraph`, `stackprof` and `memory_profiler` gems which provide more detailed analysis. Try out by appending `?pp=flamegraph`, `?pp=profile-gc` or `?pp=analyze-memory` to the URL. You can read more about these options at https://github.com/MiniProfiler/rack-mini-profiler#flamegraphs. More details: https://github.com/MiniProfiler/rack-mini-profiler == Profiling best practices You need to take the insights of these gems with a grain of salt though, if you run this in development. Rails' development settings have nothing to do with a production set up where classes are not reloaded and assets are precompiled and served from a web server. Therefore, you should mimic these settings as much as possible if you want your findings to be realistic.
Version data entries
36 entries across 36 versions & 1 rubygems