Desk

    • Friday, 2 October

    Four meetings and a release window. The atlas retry budget is the one thing to settle before Monday.

    4 meetings ledger ships 16:30 2 reviews waiting

    Numbers

    The week so far, counted at nine this morning.

    Unread mail126 need a reply
    Commits this week49across four repositories
    Open pull requests52 are waiting on you
    Focus time14.5 hMonday to this morning

    Mail

    Unread rows are bold. Sort by any column, or filter the list by sender, subject or label.

    Inbox, newest first
    FromSubjectLabelTime
    Dana WhitfieldRetry budget for atlas is too highReview
    Tom Keaneledger nightly build failed on step fourBuild
    Priya ShahNotes from the hummingbird design reviewReview
    Grace NolanDocs search finds nothing for quotaOps
    Sam OkaforCan you read the atlas migration plan?Review
    Lena HartmannStandup moves to 09:45 todayTeam
    Omar HaddadDisk alert on the build runnerOps
    Mei Chenledger 2.4 is outRelease
    Ruth AbelnDraft agenda for FridayTeam
    Carl Neumannhummingbird flakes in the queue testsBuild
    Yara SaidPlease read the docs style guideReview
    Nils Bergatlas error budget for SeptemberOps
    Clara FontaineHoliday cover for next weekTeam
    Joel Mbekiledger schema change needs a sign-offReview
    Three messages open in the demo. See all 40

    Next up

    Standup at 09:45, then the hummingbird design review at eleven. Two clear hours after that, before Dana at two.

    Today

    09:45
    Standup, ten minutes
    11:00
    hummingbird design review, with Priya and Carl
    14:00
    One to one with Dana
    16:30
    ledger release window, on call for an hour after

    By label

    1. Review
    2. Build
    3. Ops
    4. Team
    5. Release

    Waiting on you

    • atlas retry budgetDana, this morning09:14
    • atlas migration planSam, this morning07:48

    Code

    A year of commits, the last seven days by day, and what landed since Tuesday.

    1. Sat
    2. Sun
    3. Mon
    4. Tue
    5. Wed
    6. Thu
    7. Fri

    Open pull requests

    • Cap the retry budget at four attemptsatlas · opened ThuReview
    • Split the migration plan into three stepsatlas · opened TueReview
    • Wait two seconds between retriesatlas · opened WedDraft
    • Drop the old currency columnledger · opened TueApproved
    • Name the queue metrics after what they counthummingbird · opened MonDraft
    1. atlas

      Cap the retry budget at four attempts.

    2. docs

      Rewrite the quota page in plain words.

    3. hummingbird

      Revert the queue change that broke the tests.

    4. ledger

      Tag 2.4 and write the release note.

    5. atlas

      Split the migration plan into three steps.

    Week and notes

    What is left of the week, and somewhere to put a thought before it goes.

    • StandupEvery weekday, ten minutes09:45
    • hummingbird design reviewToday, with Priya and Carl11:00
    • ledger release windowToday, then on call until six16:30
    • On call startsSaturday morning, one week09:00
    • atlas migration walk-throughMonday, with Sam10:00
    • Team retroMonday, forty minutes15:00
  • All notes
    •

    Retry budget for atlas is too high

    From Dana Whitfield <dana.whitfield@example.com> · 09:14

    The gateway retries nine times before it gives up, so a slow shard turns into a flood. I would cap it at four and add a short wait between tries. Can you take it today? I can review this afternoon.

    •

    ledger nightly build failed on step four

    From Tom Keane <tom.keane@example.com> · 08:52

    Step four ran out of disk on the build runner. Omar is clearing it. Nothing is wrong with the code, so the 16:30 release can go ahead once the next run is green.

    •

    Notes from the hummingbird design review

    From Priya Shah <priya.shah@example.com> · 08:31

    Three things came out of it: the queue needs a dead letter path, the retry logic should live in one place, and we want a flag to turn the new path off. I wrote them up before this morning's session so we start from the same page.

    •

    • A. Reyes

    Signed in on this machine.

    •