Migrate sre-oncall (handover) to Woodhouse

The codebase at https://ops.gitlab.net/gitlab-com/gl-infra/sre-oncall-handover-automation/ should be moved to Woodhouse. A large portion of the logic already exists in https://gitlab.com/gitlab-com/gl-infra/oncall-robot-assistant, which is stubbed for migration here - https://gitlab.com/gitlab-com/gl-infra/infrastructure/-/issues/11776.

The incoming/outgoing Engineers On-Call (EOCs) rely on the /sre-oncall slash command to semi-automatically create an incident before a shift change.

Rather than require a human run the command, we can easily leverage the PagerDuty schedule to detect a shift change (or use the set schedule of every 8 hours) to lookup the incoming/outgoing EOCs and create the issue.

To do so, we'll need to maintain a map of SREs Slack and PagerDuty user IDs.

If this hasn't already been done in the old codebase, we should ensure that SREs who have already finished their shift can still create a handover issue. We could use a slash-command parameter or a modal to allow SREs to override the current shift and choose a previous shift.

Definition of Done

  • the issue creation is automatically created 1-hour before shift change
    • the incoming and outgoing on-call engineers are assigned the issue
    • a reminder message with a link to the handover 1-hour before the shift-change is sent via direct message to both the incoming and outgoing EOCs
  • a "shift-change" notification is posted to the #incident-management room at the handover time with a link to the issue
  • the /sre-oncall slash command is deprecated
  • the Slack/GitLab integration with the "On Call Handover" app is deprecated
Edited by Devin Sylva