@@ -24,3 +24,9 @@ Mail restore drills order after already queued application backups, integrity
 checks and database drills. Their runtime budget starts when the drill can run,
 so a long initial upload cannot exhaust the mail service's lock-wait budget.
 The existing shared lock still serializes later arrivals.
+
+Host integrity checks and database drills order after already queued backups
+and retries. They begin their own six-hour operational budget when runnable.
+This accounts for full SQLite validation on the persistent volume; observed
+random-read throughput was about 1 MiB/s during the initial 4.5 GiB snapshot
+set. Integrity and restore validation remain complete.
@@ -1,6 +1,10 @@
 [Unit]
+After=host-app-backup.service
 RequiresMountsFor=/var/lib/stacks-deployments
 
 [Service]
 ExecStart=
 ExecStart=/usr/local/sbin/host-backup.py --retry-failed --preserve-backups
+
+# Initial full snapshots and both-provider restores include multi-GiB databases.
+TimeoutStartSec=6h
@@ -4,3 +4,6 @@ RequiresMountsFor=/var/lib/stacks-deployments
 [Service]
 ExecStart=
 ExecStart=/usr/local/sbin/host-backup.py --preserve-backups
+
+# Initial full snapshots and both-provider restores include multi-GiB databases.
+TimeoutStartSec=6h
@@ -0,0 +1,8 @@
+[Unit]
+After=host-app-backup.service host-app-backup-retry.service
+RequiresMountsFor=/var/lib/stacks-deployments
+
+[Service]
+
+# Initial full snapshots and both-provider restores include multi-GiB databases.
+TimeoutStartSec=6h
@@ -0,0 +1,8 @@
+[Unit]
+After=host-app-backup.service host-app-backup-retry.service host-backup-check.service
+RequiresMountsFor=/var/lib/stacks-deployments
+
+[Service]
+
+# Initial full snapshots and both-provider restores include multi-GiB databases.
+TimeoutStartSec=6h