## Summary This issue is to roll out [the feature](https://gitlab.com/gitlab-org/gitlab/-/work_items/598413) on production, that is currently behind the `code_completion_v4_endpoint` feature flag on Gitlab. ## Owners * Most appropriate Slack channel to reach out to: `#g_ai_coding` * Best individual to reach out to: @mikewronski ## Expectations ### What are we expecting to happen? All code completions through Gitlab.com API will go to AI Gateway v4 `/v4/code/suggestions` instead of `/v2/code/completions` . There is unlikely to be any noticeable change in completion behaviour as the v4 endpoint is now on parity with v2 completions with slight variations in underlying implementation. ### What can go wrong and how would we detect it? * Code completions may not work correctly, may be delayed or output may be of lower quality * Monitor error rates in the [Code Suggestions telemetry dashboard](https://log.gprd.gitlab.net/app/dashboards#/view/6c947f80-7c07-11ed-9f43-e3784d7fe3ca?_g=%28refreshInterval:%28pause:!t,value:0%29,time:%28from:now-6h,to:now%29%29) ## Rollout Steps ### Rollout on non-production environments * [ ] Enable the feature flag on staging: `/chatops run feature set code_completion_v4_endpoint true --dev --pre --staging --staging-ref` * [ ] Verify that the feature works as expected. ### Specific rollout on production * [ ] `/chatops gitlab run feature set --feature-group=gitlab_team_members code_completion_v4_endpoint true` * [ ] `/chatops run feature set --project=gitlab-org/gitlab code_completion_v4_endpoint true` * [ ] Verify that the feature works. ### Global rollout on production * [ ] Incrementally roll out the feature on production. * [ ] After the feature has been 100% enabled, wait for at least one day. ### Release the feature * [ ] Create a merge request to remove the `code_completion_v4_endpoint` feature flag. * [ ] Close this rollout issue. ## Rollback Steps * [ ] `/chatops run feature set code_completion_v4_endpoint false`