Skip to content

Disable Sendfile interface for serving Sidekiq Web assets

Stan Hu requested to merge sh-fix-sidekiq-admin-cng into master

What does this MR do and why?

Sidekiq bundles its own CSS and JavaScript assets in the gem, and registers these via Rack::Static (https://github.com/mperham/sidekiq/blob/ab90830e85e02304ac39dc29f8684c75564726de/lib/sidekiq/web.rb#L149-L157). By default, Rack::Sendfile will use the Sendfile interface if available to offload the sending to Workhorse (https://github.com/rack/rack/blob/ab41dccfe287b7d2589778308cb297eb039e88c6/lib/rack/sendfile.rb#L127). However, Workhorse might not have these assets, as they are bundled with the installed Ruby gem.

Cloud Native GitLab worked around this issue by routing /admin/sidekiq directly to the webservice container, but this doesn't work if the NGINX ingress controller isn't used. Also, users have reported that the ordering of the rules may matter when another load balancer is used (gitlab-org/charts/gitlab#2731 (closed)).

To avoid these kinds of issues leading to 404s, we now disable the use of Sendfile for admin/sidekiq routes. These assets are small and served infrequently, so it should be fine to do this.

Relates to https://gitlab.com/gitlab-com/gl-infra/infrastructure/-/issues/14150

How to set up and validate locally

  1. Check out this branch.
  2. Restart GDK gdk restart
  3. Watch Workhorse logs (e.g. gdk tail gitlab-workhorse | grep sidekiq).
  4. In your browser, click on Inspect. Click Disable cache to make sure your browser loads assets again.
  5. Visit /admin/sidekiq with this branch.

Before

You'll see logs like this:

{"correlation_id":"01FF8WNQS663173EE0HX2GQ7ED","file":"/Users/stanhu/.rbenv/versions/2.7.4/lib/ruby/gems/2.7.0/gems/sidekiq-5.2.9/web/assets/javascripts/application.js","level":"info","method":"GET","msg":"Send file","time":"2021-09-10T15:18:58-07:00","uri":"/admin/sidekiq/javascripts/application.js"}

The Send file shows that Workhorse is actually sending these files.

After

These logs are gone.

MR acceptance checklist

These checklists encourage us to confirm any changes have been analyzed to reduce risks in quality, performance, reliability, security, and maintainability.

Quality

  • Quality checklist confirmed
  1. I have self-reviewed this MR per code review guidelines.
  2. For the code that that this change impacts, I believe that the automated tests (Testing Guide) validate functionality that is highly important to users (including consideration of all test levels). If the existing automated tests do not cover this functionality, I have added the necessary additional tests or I have added an issue to describe the automation testing gap and linked it to this MR.
  3. I have considered the technical aspects of the impact of this change on both gitlab.com hosted customers and self-hosted customers.
  4. I have considered the impact of this change on the front-end, back-end, and database portions of the system where appropriate and applied frontend, backend and database labels accordingly.
  5. I have tested this MR in all supported browsers, or determiend that this testing is not needed.
  6. I have confirmed that this change is backwards compatible across updates, or I have decided that this does not apply.
  7. I have properly separated EE content from FOSS, or this MR is FOSS only. (Where should EE code go?)
  8. If I am introducing a new expectation for existing data, I have confirmed that existing data meets this expectation or I have made this expectation optional rather than required.

Performance, reliability, and availability

  • Performance, reliability, and availability checklist confirmed
  1. I am confident that this MR does not harm performance, or I have asked a reviewer to help assess the performance impact. (Merge request performance guidelines)
  2. I have added information for database reviewers in the MR description, or I have decided that it is unnecessary. (Does this MR have database-related changes?)
  3. I have considered the availability and reliability risks of this change. I have also considered the scalability risk based on future predicted growth
  4. I have considered the performance, reliability and availability impacts of this change on large customers who may have significantly more data than the average customer.

Documentation

  • Documentation checklist confirmed
  1. I have included changelog trailers, or I have decided that they are not needed. (Does this MR need a changelog?)
  2. I have added/updated documentation, or I have decided that documentation changes are not needed for this MR. (Is documentation required?)

Security

  • Security checklist confirmed
  1. I have confirmed that if this MR contains changes to processing or storing of credentials or tokens, authorization, and authentication methods, or other items described in the security review guidelines, I have added the label security and I have @-mentioned @gitlab-com/gl-security/appsec.

Deployment

  • Deployment checklist confirmed
  1. I have considered using a feature flag for this change because the change may be high risk. If I decided to use a feature flag, I plan to test the change in staging before I test it in production, and I have considered rolling it out to a subset of production customers before doing rolling it out to all customers. When to use a feature flag
  2. I have informed the Infrastructure department of a default setting or new setting change per definition of done, or decided that this is not needed.
Edited by Andy Schoenen

Merge request reports