[twk3]

Owner: @twk3

Linear: ENG-2184

Changes

• docs/eks/upgrade-1.1.md, "Upgrading to Chart 1.1": a page for chart 1.0.x → 1.1.x (images 2026-10-14-00x → 2026-11-10-00x), laid out like upgrade-1.0.md:
  • a What Changes table;
  • AI Queue Worker (feat: add the ai-queue-worker deployment [ENG-2181] #74): AI tasks move from webhooks to the ai-queue-worker deployment, off by default. An install that ran AI on webhooks loses AI at this upgrade, so the section lists what to change in the same upgrade. That step was only in the quickstart;
  • Pre-Upgrade Index Check (feat: check what an upgrade's index builds cost before any pod rolls [ENG-1881] #72): what the hook does, that this upgrade passes it on most installs, and the ServiceAccount-rename workaround, linking to the full docs in upgrading.md;
  • the six indexes the scheduler builds and the five migrations it runs, each with the time measured in the rehearsal, and when the check can stop the upgrade: only a large orchestrationSession collection, about 18 GB at the default rate against the 600-second threshold;
  • what an operator does about the duplicate-key warnings from the API key and record key hash backfills;
  • rolling back to the 2026-10-14 images builds and drops nothing.
• docs/eks/upgrading.md: an Upgrade Guides by Version table near the top, linking the 1.1 and 1.0 guides. Before You Begin points to it instead of naming the 1.0 guide.
• docs/README.md lists the 1.1 guide, and the quickstart's upgrade paragraph points to the table.
• CHANGELOG.md: four [Unreleased] Upgrade Notes, for the AI worker, the index check, the index builds and the key hashes, linking to the 1.1 page.

How the numbers were measured

The full results are in the comment on ENG-2184.

• The rehearsal ran on Docker Compose, not the chart. The scheduler image and its startup tasks are the same under both, so the times carry over.
• Setup: a 2026-10-14-002 install seeded with 3.1M instances (41 GB), 310k runs and 124k orchestrationSession, upgraded to images built from release/2026-11-10-003.
• index-plan --gate, the command feat: check what an upgrade's index builds cost before any pod rolls [ENG-1881] #72's hook runs, exited 0 with a 25-second estimate. The scheduler finished startup in 18 s.
• Starting the 2026-10-14-002 scheduler on the upgraded database built and dropped nothing.
• I did not run helm upgrade itself.

🤖 Generated with Claude Code

---

^{Open in [code]smith · Need help on this PR? Tag @codesmith-bot with what you need.}

[baz-reviewer]

Baz Summary

Document the Chart 1.1 upgrade flow, including AI worker migration, MongoDB index checks, measured startup migrations, key-hash handling, and rollback behavior. Update the upgrade navigation, quickstart, README, and changelog so operators can find version-specific guidance and prepare safely.

Topics

Topic	Details
Upgrade Navigation	Organize version-specific upgrade documentation and direct users from the general guide, quickstart, and documentation index to the appropriate chart upgrade instructions. Modified files (3) docs/README.md docs/eks/quickstart.md docs/eks/upgrading.md Latest Contributors(2) User Commit Date dj@currents.dev docs: move the 1.1 upg... November 12, 2026 agoldis@users.noreply.... feat: add the ai-queue... November 12, 2026
Chart 1.1 Upgrade	Guide operators through the Chart 1.0-to-1.1 upgrade, including ai-queue-worker configuration, the currents-index-check pre-upgrade gate, measured MongoDB indexes and migrations, duplicate-key remediation, and rollback expectations. Modified files (2) CHANGELOG.md docs/eks/upgrade-1.1.md Latest Contributors(1) User Commit Date dj@currents.dev docs: add the ai-queue... November 12, 2026

---

Merger   ✅ Ready to Merge

MERGE: the documentation deliverable is present, CI ran, and there are no unresolved discussions.

---

  _{Activate Skill Maintainer to keep your skills up to date}

  _{This PR would have been improved with Baz Planner - Try it now}