From 025261e03e7ac3e9c58ed18ba9180051361debfe Mon Sep 17 00:00:00 2001
From: askalf <263217947+askalf@users.noreply.github.com>
Date: Thu, 24 Sep 2026 22:51:39 -0400
Subject: [PATCH] ci: drop the review kick; forge dispatches reviews from
 webhooks

---
 .github/workflows/review-kick.yml | 59 -------------------------------
 1 file changed, 59 deletions(-)
 delete mode 100644 .github/workflows/review-kick.yml

diff --git a/.github/workflows/review-kick.yml b/.github/workflows/review-kick.yml
deleted file mode 100644
index e9ac653..0000000
--- a/.github/workflows/review-kick.yml
+++ /dev/null
@@ -1,59 +0,0 @@
-name: Review kick
-
-# Wakes the fleet's review dispatcher as soon as a PR head changes, instead of
-# leaving the PR to wait for the dispatcher's timer. The dispatcher is
-# idempotent, so a kick never files a second review; it only moves the next
-# run to now.
-#
-# pull_request_target, so the job always runs this file as it is on main: a
-# PR cannot edit what runs on the self-hosted runner. It never checks out or
-# runs PR code. Fork PRs are not reviewed by the fleet and are skipped, as are
-# drafts and Dependabot.
-
-on:
-  pull_request_target:
-    types: [opened, synchronize, reopened, ready_for_review]
-
-permissions: {}
-
-concurrency:
-  group: review-kick
-  cancel-in-progress: false
-
-jobs:
-  kick:
-    if: >-
-      github.event.pull_request.draft == false &&
-      github.event.pull_request.head.repo.full_name == github.repository &&
-      github.event.pull_request.user.login != 'dependabot[bot]'
-    runs-on: [self-hosted, amnesia]
-    timeout-minutes: 2
-    steps:
-      - name: Queue review-dispatch for this head
-        run: |
-          if ! systemctl cat review-dispatch.service >/dev/null 2>&1; then
-            echo "::error::review-dispatch.service is not installed on this runner's host. Move this job to a runner on the dispatcher's host."
-            exit 1
-          fi
-          # The dispatcher is a oneshot that lists the open PRs when it starts, so
-          # a run already in flight may not have seen this head. A transient unit
-          # waits for that run to end, then starts another; kicks that arrive while
-          # it waits share it. After 15 minutes the unit gives up and the
-          # dispatcher's own timer covers the head.
-          if systemctl is-active --quiet review-dispatch-requeue.service; then
-            echo "a requeue is already waiting on the active dispatcher run; this head rides it"
-            exit 0
-          fi
-          if ! systemctl is-active --quiet review-dispatch.service; then
-            systemctl start --no-block review-dispatch.service
-            echo "review-dispatch started"
-            exit 0
-          fi
-          # The inner script is single-quoted on purpose: it runs in the unit, not here.
-          # shellcheck disable=SC2016
-          if systemd-run --unit=review-dispatch-requeue --quiet --collect /bin/bash -c \
-              'd=$((SECONDS + 900)); while systemctl is-active --quiet review-dispatch.service; do [ "$SECONDS" -ge "$d" ] && exit 0; sleep 5; done; systemctl start --no-block review-dispatch.service'; then
-            echo "dispatcher active; the requeue unit starts it when the current run ends"
-          else
-            echo "requeue unit could not be created (one is probably still being collected); this head rides the next run"
-          fi