diff --git a/tests/ghost/recipe_meta.py b/tests/ghost/recipe_meta.py
index dccdac5..d5a5861 100644
--- a/tests/ghost/recipe_meta.py
+++ b/tests/ghost/recipe_meta.py
@@ -8,15 +8,14 @@
 # mysqldump pre-hook; P4 (ops.py + test_{backup,restore,upgrade}.py) seeds a `ci_marker` row there.
 HEALTH_PATH = "/"  # Ghost serves a themed site HTML at root (200)
 HEALTH_OK = (200,)
-DEPLOY_TIMEOUT = 2400  # subprocess timeout for `abra app deploy`
+DEPLOY_TIMEOUT = 900  # subprocess timeout for `abra app deploy`
 HTTP_TIMEOUT = 900
 
 # Ghost's first-boot does a full schema migration (dozens of tables) against a fresh MySQL `ghost`
-# DB. On cc-ci's slow single node this takes ~6min, during which the recipe healthcheck
-# (start_period 1m → ~5min grace) marks the still-booting task unhealthy and swarm kills it; the
-# NEXT task finds the schema already created and boots fast → converges. But the first task's
-# migration + the early MySQL-not-ready (`exit 2`) app restarts can eat ~18min, so the default 1200s
-# convergence wait timed out right as it was converging. Bump to 2400s (matched to DEPLOY_TIMEOUT) so
-# the post-migration fast-boot task has room to converge within one deploy (the volume persists
-# across the in-deploy task restarts). Documented as heavy-recipe cold-boot fragility in DECISIONS.
-EXTRA_ENV = {"TIMEOUT": "2400"}
+# DB. The migration must finish within the recipe healthcheck grace (start_period 1m + 10×30s ≈ 6min)
+# — otherwise swarm kills the still-migrating task, which leaves a stale `migrations_lock` row and
+# every later task then refuses to boot (`MigrationsAreLockedError` deadlock). On the cc-ci node with
+# 4 dedicated vCPU the migration completes well inside that grace and the app converges in a few
+# minutes, so 900s is an ample-but-bounded budget (fails fast if the deadlock ever recurs, rather
+# than a long blackout). See DECISIONS (ghost MySQL cold-boot).
+EXTRA_ENV = {"TIMEOUT": "900"}