Rollout of pypi_pep_691_json Feature Flag

Everyone can contribute. Help move this issue forward while earning points, leveling up and collecting rewards.

Close this issue

Summary

This issue is to roll out the feature on production, that is currently behind the pypi_pep_691_json feature flag.

Owners

Most appropriate Slack channel to reach out to: #<slack-channel-of-dri-team>
Best individual to reach out to:

Expectations

What are we expecting to happen?

Clients sending Accept: application/vnd.pypi.simple.v1+json (or .latest+json) get PEP 691 JSON from the PyPI Simple API instead of HTML. Same file URLs, same sha256 hashes, same authorization, same forwarding.

Clients that do not send that header, including older pip, get byte-identical HTML to today.

The flag is read in two places only: lib/api/pypi_packages.rb:48 and ee/lib/ee/api/pypi_packages.rb:117. Upload, package file download, and the forwarded download route never consult it. Authorization runs before format selection, so the flag cannot change who can read what.

What can go wrong and how would we detect it?

Most failure modes are 200, not 5xx. This change reformats a response rather than performing an operation, so a wrong-but-well-formed body is a complete success server-side and fails only in the client.

Failure	Server-side signal
Presenter raises	5xx. Visible.
JSON valid but wrong (bad URL, missing hash)	None. 200, `pip install` fails client-side.
Wrong format selected for the client	None. 200 with an unparseable body.
Forwarded upstream sends malformed JSON	None. The 200 is written before streaming begins (`workhorse/internal/sendurl/sendurl.go:201`), so the client gets a truncated 200.
Forwarded upstream returns 5xx	Passed through with its own status. `ErrorResponseStatus` covers transport failures only.
Slow upstream	Connect and response-header timeouts are 10s each, mapped to 502. There is no idle-read timeout: https://gitlab.com/gitlab-org/gitlab/-/work_items/609164

A rise in 5xx is a real signal, but a flat 5xx rate is not evidence the JSON path is healthy. No metric distinguishes JSON from HTML, so the load-bearing check is the "Verify that the feature works" steps below: use a pip version that sends the PEP 691 Accept header. Otherwise a clean dashboard is indistinguishable from one where no client ever exercised the new path.

Note API::PypiPackages sets urgency :low, so the apdex target is 5 seconds (lib/gitlab/endpoint_attributes/config.rb:11). JSON-vs-HTML serialization differs by milliseconds and will not move apdex.

Rollback is clean: read-only path, no persisted state, and every variant is served Cache-Control: max-age=0, private, must-revalidate, so there is no shared cache to purge. Forward URLs already in lockfiles keep resolving either way, because the forward route does not read the flag.

Dashboards

SLI and latency for the affected endpoints: https://dashboards.gitlab.net/goto/bfwg4xsmtowlce?orgId=1

GET /api/:version/projects/:id/packages/pypi/simple
GET /api/:version/projects/:id/packages/pypi/simple/*package_name
GET /api/:version/groups/:id/-/packages/pypi/simple
GET /api/:version/groups/:id/-/packages/pypi/simple/*package_name
GET /api/:version/projects/:id/packages/pypi/forward/:package_name/*upstream_path

Check volume first. If gitlab_sli_rails_request_total is zero for these, the error and apdex ratios mean nothing.

Workhorse forwarding. No dashboard exists; query Grafana Explore against mimir-gitlab-gprd:
```
sum(rate(gitlab_workhorse_send_url_requests{status="request-failed"}[5m]))
  / sum(rate(gitlab_workhorse_send_url_requests[5m]))
```
A climbing gitlab_workhorse_send_url_open_requests is the signal that the missing idle-read timeout is being hit.
Kibana, for failures that never become a 5xx: search "SendURL: Copy response" and "SendURL: Do request" at https://log.gprd.gitlab.net

Rollout Steps

Note: Please make sure to run the chatops commands in the Slack channel that gets impacted by the command.

Rollout on non-production environments

Verify the MR with the feature flag is merged to master and has been deployed to non-production environments with /chatops run auto_deploy status <merge-commit-of-your-feature>

Deploy the feature flag at a percentage (recommended percentage: 50%) with /chatops run feature set pypi_pep_691_json <rollout-percentage> --actors --dev --pre --staging --staging-ref
Monitor that the error rates did not increase (repeat with a different percentage as necessary).

Enable the feature globally on non-production environments with /chatops run feature set pypi_pep_691_json true --dev --pre --staging --staging-ref
Verify that the feature works as expected. The best environment to validate the feature in is staging-canary as this is the first environment deployed to. Make sure you are configured to use canary.
If the feature flag causes end-to-end tests to fail, disable the feature flag on staging to avoid blocking deployments.
- See #e2e-run-staging Slack channel and look for the following messages:
  - test kicked off: Feature flag pypi_pep_691_json has been set to true on **gstg**
  - test result: This pipeline was triggered due to toggling of pypi_pep_691_json feature flag

If you encounter end-to-end test failures and are unable to diagnose them, you may reach out to the #s_developer_experience Slack channel for assistance. Note that end-to-end test failures on staging-ref don't block deployments.

Before production rollout

If the change is significant and you wanted to announce in #whats-happening-at-gitlab, it best to do it before rollout to gitlab-org/gitlab-com.

Specific rollout on production

For visibility, all /chatops commands that target production must be executed in the #production Slack channel and cross-posted (with the command results) to the responsible team's Slack channel.

Ensure that the feature MRs have been deployed to both production and canary with /chatops run auto_deploy status <merge-commit-of-your-feature>
Depending on the type of actor you are using, pick one of these options:
- For project-actor: /chatops run feature set --project=gitlab-org/gitlab,gitlab-org/gitlab-foss,gitlab-com/www-gitlab-com pypi_pep_691_json true
- For group-actor: /chatops run feature set --group=gitlab-org,gitlab-com pypi_pep_691_json true
- For user-actor: /chatops run feature set --user=<gitlab-username-of-dri> pypi_pep_691_json true
- For all internal users: /chatops run feature set --feature-group=gitlab_team_members pypi_pep_691_json true
Verify that the feature works for the specific actors.

Preparation before global rollout

Set a milestone to this rollout issue to signal for enabling and removing the feature flag when it is stable.
Check if the feature flag change needs to be accompanied with a change management issue. Cross link the issue here if it does.
Ensure that you or a representative in development can be available for at least 2 hours after feature flag updates in production. If a different developer will be covering, or an exception is needed, please inform the oncall SRE by using the @sre-oncall Slack alias.
Ensure that documentation exists for the feature, and the version history text has been updated.
Ensure that any breaking changes have been announced following the release post process to ensure GitLab customers are aware.
Notify the #support_gitlab-com Slack channel and your team channel (more guidance when this is necessary in the dev docs).

Global rollout on production

For visibility, all /chatops commands that target production must be executed in the #production Slack channel and cross-posted (with the command results) to the responsible team's Slack channel.

Incrementally roll out the feature on production.
- Example: /chatops run feature set pypi_pep_691_json <rollout-percentage> --actors.
- Between every step wait for at least 15 minutes and monitor the appropriate graphs on https://dashboards.gitlab.net.
After the feature has been 100% enabled, wait for at least one day before releasing the feature.

(Optional) Release the feature with the feature flag

WARNING: This approach has the downside that it makes it difficult for us to clean up the flag. For example, on-premise users could disable the feature on their GitLab instance. But when you remove the flag at some point, they suddenly see the feature as enabled and they can't roll it back to the previous behavior. To avoid this potential breaking change, use this approach only for urgent matters.

See instructions if you're sure about enabling the feature globally through the feature flag definition

If you're still unsure whether the feature is deemed stable but want to release it in the current milestone, you can change the default state of the feature flag to be enabled. To do so, follow these steps:

Release the feature

After the feature has been deemed stable, the clean up should be done as soon as possible to permanently enable the feature and reduce complexity in the codebase.

You can either create a follow-up issue for Feature Flag Cleanup or use the checklist below in this same issue.

Create a merge request to remove the pypi_pep_691_json feature flag. Ask for review/approval/merge as usual. The MR should include the following changes:
- Remove all references to the feature flag from the codebase.
- Remove the YAML definitions for the feature from the repository.
Ensure that the cleanup MR has been included in the release package. If the merge request was deployed before the monthly release was tagged, the feature can be officially announced in a release blog post: /chatops run release check <merge-request-url> <milestone>
Close the feature issue to indicate the feature will be released in the current milestone.
Once the cleanup MR has been deployed to production, clean up the feature flag from all environments by running these chatops command in #production channel: /chatops run feature delete pypi_pep_691_json --dev --pre --staging --staging-ref --production
Close this rollout issue.

Rollback Steps

This feature can be disabled on production by running the following Chatops command:

/chatops run feature set pypi_pep_691_json false

Disable the feature flag on non-production environments:

/chatops run feature set pypi_pep_691_json false --dev --pre --staging --staging-ref

Delete feature flag from all environments:

/chatops run feature delete pypi_pep_691_json --dev --pre --staging --staging-ref --production

Edited Aug 27, 2026 by 🤖 GitLab Bot 🤖