Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  • How to deal with QCA-DS PRs which operate on a different board

  • JC came to last Friday’s iteration planning, but nobody else was there.

  • How to deal with issues as they arise mid-iteration (triage)?

  • Expectations around updating board outside of standup

2025_02_06 retrospective

How did this go? Ideas for improvement?

JM – Feeling pretty good. Enjoing daily check ins. Still some friction but far less. Think we’re getting better at handling fast-breaking/spontaneous stuff. One thing I did this cycle was breaking a big task into subtasks mid-iteration. It technically “broke the rules” but it contributed a lot to me feeling productive/staying on task this week.

JC – Probably a learning opportunity in how I put together my epic, felt like I was stuck. When things were moving I liked the process a lot more. It was helpful to keep on task and know my goals. (later) re: breaking the rules, we decided I should put together some of my scripts for managing/analyzing kubernetes utilization. We did this mid-iteration, but posted in the zenhub channel.

MT – Felt like this was a step backwards, I don’t think my productivity was utilized. Was mostly waiting on others for reviews/prioritization/feedback. There were more times than I’d prefer that I couldn’t execute on things. So felt like my throughput was lower in the past two weeks. Feel there’s a lot of room to improve on how we track things on ZenHub - Multi-day megatickets with shifting goals defeat the purpose of tracking work. I think it’s easy for the board to drift away from capturing true state, and I think this happened this iteration, instead tasks were driven by my github/slack/email inbox. I think I’m at my best when I can look at ZenHub feed and know what I need to do, and that things were in these other channels hurt productivity.

LW – How do you think the standups function in unblocking/slowing your productivity? Were reviews not getting picked up or people were picking them but not actioning them?

MT – Yes, that contributed but a larger part might have been that I didn’t pick enough work at the beginning of the iteration. Also several blockages had their roots in a lack of feedback from external (to our org) stakeholders.

LW – Regarding tasks in email/slack, were those because zenhub was depleted, or because new tasks weren’t added quickly enough.

MT - The latter, if I have too little on my daily list, I’m more prone to distraction.

LW – Like MT, thought this was a step backwards. I have a lot of the megatickets with shifting goals. So as they shifted the board didn’t reflect what I was actually doing. So I kinda solved this by opening new sub-tickets and added them directly to current iteration. But then others just sat there for a while. I think it will be difficult to fix when tickets can’t be updated mid-iteration. Not sure about more graceful ways to fix this than changing things mid-iteration. Otherwise I found the daily check-ins helpful. I’m unclear on when things get significant enough to get tracked on ZenHub (ex emails and slack), since sometimes my whole day will get sunk into those. So being able to track those on ZH would be helpful.

MT – This had kinda come up when I was doing iterations w/ ZB and JE last year. We set a goal by setting an 80-20 goal of having MOST work tracked on board, but allowing 20ish% to be untracked comms. So I think we’re best off if not everything is tracked, just a majority.

JW – I think me taking a few days off showed how this can kinda be a fragile system. I did find a need to go outside the system and review a bunch of stuff unassigned to me to unblock Matt, but it was nice to have a list of the blocking items conveniently. Agree with the idea of changing board state mid-iteration. On weds my check in with MT made me realize that there were a bunch of easy PRs that could be quickly reviewed, and I went “outside the system” to clear them quickly without being assigned.

Procedural changes:

  • Breaking into subtasks mid-iterations

    • Anyone can break big tasks into subtasks mid-iteration

  • Triaging new issues mid-iteration

    • We’ll make an “urgent” column that will be reviewed first in each standup

  • Expedited review without assignment

    • This is allowed, just self-assign and do it. But big/nontrivial PRs will still need assignment in standup.

  • Expectations around updating board outside standups

    • This needs further iteration/discussion at future meetings. (rollover)

  • One week iterations? (rollover)

(lots of discussion)

LW: How about we stick to two-week iterations but once a week (just the science team?) does some more shuffling than would happen in a standup

JW: I think sub-task breakdown should be allow mid-iteration, with some hope that metadata (like epochs?) is mapped into the new subtasks

MT – I’m strongly in favor of breaking big tasks down into subtasks mid-iteration.

LW – Rule could be “break it down and tell people at the next standup”

JW – Task breakdown could be done in 1:1s with leads, that way it’s kinda kosher to be adding new items to backlogs/icebox/new epics

MT – Hard to reason around this being a rule without clear guidelines on when board can be updated. Kinda touches on late-breaking issues.

JW – We could have an “urgent” column where late-breaking stuff goes, and we can immediately assign people and action it, but then we discuss it in the next standup

JM – Could each have a permanent “chaos” ticket where we assign an estimate when we get sucked into crazy stuff. Then we can communicate how much time goes into craziness.

JC – In this case, I’d create a ticket in new issues, and tell LW, and have her move it and post in the zenhub channel.

MT – I’m up to try the “alert/urgent” column. Not a perfect idea but it will be a starting point for iteration. Will let us use the current system in place, and highlight blockingness/corral stakeholders.

JM – The review column series could be done without. Not sure adding more columns is a solution, since we’re already not checking many.

JW – I’m in favor of simplifying review columns

LW – I do think we need a better notification system for urgent fires. And they could go before all other updates in standup.