Proposal: Reducing ticket load immediately before going into weeklong on-call rotation

This issue and the associated draft MR (gitlab-com/www-gitlab-com!88389 (closed)) aim to offer an improvement to how engineers prepare to participate in a weeklong on-call rotation.

Problem Statement

  • What is the problem?

When acting in one of the on-call rotations in Support, we want people to be alert and available. We acknowledge that:

Just being available for emergencies and outages causes stress, even if there are no pages.

Engineers may participate in the Support Response Crew and otherwise become Assigned to tickets in the week before their time in an on-call rotation.

  • Why is this a problem?

Working a full stack of Assigned tickets while also handling emergencies (or being available to do so) can be difficult. This can lead to an engineer who is tired and stressed from being on-call who then needs to send responses to Assigned tickets. In addition to the stress inherent in this situation, this can lead to delayed or degraded responses to the customer on the other end of the non-emergency tickets.

Proposal

On the Friday before their time in the week-long CMOC and Customer Emergencies rotations, engineers should:

  • unassign tickets 🎫 (an internal note with a summary and ideas on next steps must be present)
  • notify the Crew manager of each ticket

The Crew manager will be responsible for finding a new assignee.

With this approach, the engineer can work tickets as normal in the Crew during the week before they are on-call. See #3696 (closed) for discussion on handling being in Crew during the on-call week. Engineers may elect to engage in professional development activities during their time awaiting pages: these are less stressful activities that can be interrupted readily.

DRI

Brie Carranza (@bcarranza) will act as the DRI for this issue.

Required Resources

  • This change requires continued support from management and the GitLab Support team at large.
  • This change requires gitlab-com/www-gitlab-com!88389 (closed) to be merged.

Potential Roadblocks/Things to consider

  • This adds an additional task for the manager working in the Crew each Friday.
  • People should aim to do this on Friday morning to give the manager as much time as possible to find an alternate assignee. This enables the engineer going on-call and the person picking up the ticket to 🍐 pair before the engineer goes on-call to discuss nuanced tickets.
  • How should this change impact folks in the US Federal on-call rotation?

Desired Outcome

  • What does success look like?

    • Engineers in on-call rotations don't have to spend time finding ticket assignees while they are actively on-call.
    • Engineers in an on-call rotation the following week can fully perform Crew duties knowing that there is a built-in hand-off for any tickets the pick up during that week before they are on-call.
  • How do we measure success?

    • SSAT
    • Feedback from SEs
  • Where would future feedback go?

The feedback section of this issue or another issue in the support-team-meta issue tracker. See the thread in #support_team-chat.

Related Issues/MRs/Epics/Tickets

gitlab-com/www-gitlab-com!88389 (closed)

🌐 Relevant Handbook Links

  • Support Response Crew
  • On-call in GitLab Support

👂 Feedback

Feedback about this change once it goes live, if any. Alternately, create a separate issue and link to it here.

Edited Aug 17, 2021 by Brie Carranza
Assignee Loading
Time tracking Loading