diff --git a/.github/workflows/e2e.yml b/.github/workflows/e2e.yml
index 089ab5c..f30d3da 100644
--- a/.github/workflows/e2e.yml
+++ b/.github/workflows/e2e.yml
@@ -12,14 +12,15 @@
 #              release's binary, images and plugin from that release's assets; skipped until
 #              a release exists
 #   upgrade    the newest published release through its own installer and its own assets,
-#              then this commit's assets on top of it; skipped until a release exists
+#              rows of every kind seeded into its database, then this commit's assets on
+#              top of it; skipped until a release exists
 #   source     a full install built on the host from this checkout (FELIS_SKIP_FETCH):
 #              the fallback a release without assets, or an unsupported one, takes. It builds
 #              everything with Docker, so it runs by hand and weekly only
 #
 # This runs on pushes that touch what gets installed, by hand, and weekly (a moving
 # upstream: apt mirrors, k3s's install script and release assets, Adoptium).
-# deploy/e2e_check.sh holds the assertions.
+# deploy/e2e_check.sh holds the assertions, deploy/e2e_seed.sh the upgrade's seed and its check.
 name: e2e
 
 on:
@@ -256,13 +257,25 @@ jobs:
           sudo bash deploy/e2e_check.sh release
           bash deploy/e2e_release.sh check-own release.log
 
+      # A fresh install's database holds only what its migrations wrote: the seed gives the
+      # upgrade's move and its pending migrations existing rows to carry.
+      - name: Seed the release's database
+        if: steps.release.outputs.tag != ''
+        run: sudo bash deploy/e2e_seed.sh seed
+
       - name: Upgrade to this commit
         if: steps.release.outputs.tag != ''
         run: sudo FELIS_ARTIFACT_DIR="$GITHUB_WORKSPACE/dist" FELIS_INSTALL_MODE=full bash deploy/bootstrap.sh 2>&1 | tee upgrade.log
 
+      # Both checks run and report: the seeded rows go last, after the restore drill has
+      # also put them back from a bundle.
       - name: Check the upgrade
         if: steps.release.outputs.tag != ''
-        run: sudo bash deploy/e2e_check.sh upgrade
+        run: |
+          rc=0
+          sudo bash deploy/e2e_check.sh upgrade || rc=1
+          sudo bash deploy/e2e_seed.sh check || rc=1
+          exit "$rc"
 
       - name: Diagnostics
         if: failure()
diff --git a/deploy/e2e_seed.sh b/deploy/e2e_seed.sh
new file mode 100644
index 0000000..5712a20
--- /dev/null
+++ b/deploy/e2e_seed.sh
@@ -0,0 +1,208 @@
+#!/bin/bash
+# Seeds the database of the release the e2e upgrade job installed, and checks after the
+# upgrade that every seeded row came through unchanged. The job runs it around the upgrade:
+#
+#   sudo bash deploy/e2e_seed.sh seed    # after the release installed
+#   sudo bash deploy/e2e_seed.sh check   # after this commit ran over it
+#
+# A fresh install's database holds only the rows its migrations write, so without the seed
+# the upgrade moves and migrates an almost empty database: the move's row-count comparison
+# compares zeros and the pending migrations never meet an existing row. The seed writes the
+# columns the oldest release has (v0.1.0), so it applies to every release since, and the
+# check reads the same columns back: a migration that rewrites one of them on purpose
+# updates the snapshot query here.
+#
+# Every row is inert. Nothing in it is a credential anyone holds: the session and the
+# challenge carry the hash of bytes nobody kept and are spent already. Nothing is due for
+# the reaper or the operator, and the retention sweep keeps spent rows for 30 days after
+# they were spent. servers stays empty: its rows mirror MinecraftServer objects the
+# operator and the reaper reconcile, and a row without one would not stay as written.
+set -euo pipefail
+
+phase="${1:?usage: e2e_seed.sh seed|check}"
+export KUBECONFIG=/etc/rancher/k3s/k3s.yaml
+KUBECTL=(/usr/local/bin/k3s kubectl)
+STATE_DIR=/var/tmp/felis-e2e-seed
+PG_MOVED_MARKER=/var/lib/felis/postgres-moved
+TABLES=(users account_links quotas sessions webauthn_challenges platform_settings audit_logs
+  world_backups image_builds image_submissions account_migrations op_login_requests)
+fails=0
+
+pass() { printf 'PASS %s\n' "$*"; }
+fail() { printf 'FAIL %s\n' "$*"; fails=$((fails + 1)); }
+
+# db_where says where the platform's database lives: in felis-postgres once the install
+# has it, else in the host PostgreSQL a release before it ran.
+db_where() {
+  if "${KUBECTL[@]}" -n felis get deploy/felis-postgres >/dev/null 2>&1; then
+    echo pod
+  else
+    echo host
+  fi
+}
+
+# db_psql runs the SQL on its stdin in the felis database at <where>.
+db_psql() { # where
+  case "$1" in
+    pod)
+      "${KUBECTL[@]}" -n felis exec -i deploy/felis-postgres -c postgres -- \
+        psql -X -q -At -v ON_ERROR_STOP=1 -U felis -d felis
+      ;;
+    # From /, which the postgres user can always enter: psql warns about any other cwd.
+    host) (cd / && runuser -u postgres -- psql -X -q -At -v ON_ERROR_STOP=1 -d felis) ;;
+  esac
+}
+
+# snapshot prints every seeded row, one table after the other, each in key order. The host
+# server's time zone is the host's and the pod's is UTC, so timestamps print in UTC.
+snapshot() { # where
+  db_psql "$1" <<'EOF'
+SET TimeZone = 'UTC';
+SELECT 'users ' || row(id, username, email, role, email_verified, disabled, deleted_at, created_at, updated_at)::text
+  FROM users WHERE id LIKE 'e2e-seed-%' ORDER BY id;
+SELECT 'account_links ' || row(user_id, mc_uuid, auth_source, verified_at)::text
+  FROM account_links WHERE user_id LIKE 'e2e-seed-%' ORDER BY mc_uuid;
+SELECT 'quotas ' || row(user_id, max_servers, max_cpu_milli, max_memory_mb, max_storage_gb, updated_by, updated_at)::text
+  FROM quotas WHERE user_id LIKE 'e2e-seed-%' ORDER BY user_id;
+SELECT 'sessions ' || row(token_hash, user_id, created_at, expires_at, revoked_at)::text
+  FROM sessions WHERE user_id LIKE 'e2e-seed-%' ORDER BY token_hash;
+SELECT 'webauthn_challenges ' || row(id, user_id, purpose, session_data, expires_at, consumed_at, created_at)::text
+  FROM webauthn_challenges WHERE id LIKE 'e2e-seed-%' ORDER BY id;
+SELECT 'platform_settings ' || row(key, value, updated_at)::text
+  FROM platform_settings WHERE key = 'e2e_seed';
+SELECT 'audit_logs ' || row(id, actor, source, action, server_name, request_id, payload, created_at)::text
+  FROM audit_logs WHERE actor = 'e2e-seed-staff' ORDER BY id;
+SELECT 'world_backups ' || row(id, server_name, former_owner, backup_ref, size_bytes, reason, status, created_at, expires_at, deleted_at)::text
+  FROM world_backups WHERE id LIKE 'e2e-seed-%' ORDER BY id;
+SELECT 'image_builds ' || row(id, image_ref, status, dockerfile, context_ref, base_image, requested_by, job_name, log_ref, error, created_at, finished_at)::text
+  FROM image_builds WHERE id LIKE 'e2e-seed-%' ORDER BY id;
+SELECT 'image_submissions ' || row(id, submitted_by, display_name, context_ref, status, image_ref, build_id, reviewed_by, reject_reason, created_at, reviewed_at)::text
+  FROM image_submissions WHERE id LIKE 'e2e-seed-%' ORDER BY id;
+SELECT 'account_migrations ' || row(id, source_user_id, target_user_id, state, confirm_factor, confirmed_at, code_hash, code_expires_at, redeemed_at, created_at, updated_at)::text
+  FROM account_migrations WHERE id LIKE 'e2e-seed-%' ORDER BY id;
+SELECT 'op_login_requests ' || row(id, user_id, email, expires_at, created_at, consumed_at, approved_at, approved_by)::text
+  FROM op_login_requests WHERE id LIKE 'e2e-seed-%' ORDER BY id;
+EOF
+}
+
+# seed writes the rows in one transaction and keeps where they went and what they read as.
+# Text carries quotes, backslashes, control characters and non-ASCII, and the bytea a NUL,
+# so a dump or restore that mangles any of them shows in the check. Rows that the retention
+# sweep would judge by age are dated now.
+seed() {
+  local where before t
+  where="$(db_where)"
+  if ! db_psql "$where" <<'EOF'; then
+BEGIN;
+INSERT INTO users (id, username, email, role, email_verified, disabled, created_at, updated_at) VALUES
+  ('e2e-seed-player', 'e2e_seed_player', 'Seed.Player+e2e@example.com', 'user', true, false,
+   '2026-01-02 03:04:05.678901+00', '2026-01-02 03:04:05.678901+00'),
+  ('e2e-seed-staff', 'e2e_seed_staff', 'staff@example.com', 'admin', false, true,
+   '2026-01-03 00:00:00+00', '2026-01-03 00:00:00+00');
+INSERT INTO account_links (user_id, mc_uuid, auth_source, verified_at) VALUES
+  ('e2e-seed-player', '5eed0000-e2e0-4000-8000-000000000001', 'thirdparty', '2026-01-02 04:00:00+00');
+INSERT INTO quotas (user_id, max_servers, max_cpu_milli, max_memory_mb, max_storage_gb, updated_by, updated_at) VALUES
+  ('e2e-seed-player', 2, 4000, 8192, NULL, 'e2e-seed-staff', '2026-01-04 00:00:00+00');
+INSERT INTO sessions (token_hash, user_id, created_at, expires_at, revoked_at) VALUES
+  (encode(sha256(convert_to(gen_random_uuid()::text, 'UTF8')), 'hex'), 'e2e-seed-player',
+   now(), now() + interval '30 days', now());
+INSERT INTO webauthn_challenges (id, user_id, purpose, session_data, expires_at, consumed_at, created_at) VALUES
+  ('e2e-seed-challenge', 'e2e-seed-player', 'passkey_register', '\x00ff0a0d5c27'::bytea,
+   now() + interval '5 minutes', now(), now());
+INSERT INTO platform_settings (key, value, updated_at) VALUES
+  ('e2e_seed', '{"text": "quote '' dq \" backslash \\ tab\t newline\n 猫 🐱", "n": 1.50, "list": [1, null, true, {"k": "v"}]}',
+   '2026-01-05 00:00:00+00');
+INSERT INTO audit_logs (actor, source, action, server_name, request_id, payload, created_at) VALUES
+  ('e2e-seed-staff', 'panel', 'e2e.seed', NULL, 'e2e-seed-req-1', '{"reason": "种子 \"quoted\"", "ids": [1, 2, 3]}', now()),
+  ('e2e-seed-staff', 'cli', 'e2e.seed', 'e2e-seed-world', 'e2e-seed-req-2', NULL, now());
+INSERT INTO world_backups (id, server_name, former_owner, backup_ref, size_bytes, reason, status, created_at, expires_at, deleted_at) VALUES
+  ('e2e-seed-backup', 'e2e-seed-world', 'e2e-seed-player', 'e2e-seed/world.tar.zst', 123456789012, 'manual', 'deleted',
+   '2026-01-06 00:00:00+00', '2026-04-06 00:00:00+00', '2026-04-07 00:00:00+00');
+INSERT INTO image_builds (id, image_ref, status, dockerfile, context_ref, requested_by, error, created_at, finished_at) VALUES
+  ('e2e-seed-build', 'registry.felis.svc:5000/user-uploads/e2e-seed:latest', 'failed',
+   E'FROM scratch\nLABEL note="e2e seed"\n', 'e2e-seed/context', 'e2e-seed-staff', 'e2e seed: never built',
+   '2026-01-07 00:00:00+00', '2026-01-07 00:01:00+00');
+INSERT INTO image_submissions (id, submitted_by, display_name, context_ref, status, reviewed_by, reject_reason, created_at, reviewed_at) VALUES
+  ('e2e-seed-submission', 'e2e-seed-player', '种子整合包 e2e', 'e2e-seed/submission', 'rejected', 'e2e-seed-staff', 'e2e seed',
+   '2026-01-08 00:00:00+00', '2026-01-08 01:00:00+00');
+INSERT INTO account_migrations (id, source_user_id, target_user_id, state, confirm_factor, confirmed_at, redeemed_at, created_at, updated_at) VALUES
+  ('e2e-seed-migration', 'e2e-seed-staff', 'e2e-seed-player', 'redeemed', 'email_otp', '2026-01-09 00:00:00+00',
+   '2026-01-09 00:05:00+00', '2026-01-09 00:00:00+00', '2026-01-09 00:05:00+00');
+INSERT INTO op_login_requests (id, user_id, email, expires_at, created_at, consumed_at, approved_at, approved_by) VALUES
+  ('e2e-seed-oplogin', 'e2e-seed-staff', 'staff@example.com', now() + interval '10 minutes', now(), now(), now(), 'e2e-seed-staff');
+COMMIT;
+EOF
+    fail "seed the release's database (${where})"
+    return
+  fi
+  mkdir -p "$STATE_DIR"
+  printf '%s\n' "$where" > "${STATE_DIR}/where"
+  if ! before="$(snapshot "$where")"; then
+    fail "read the seeded rows back"
+    return
+  fi
+  printf '%s\n' "$before" > "${STATE_DIR}/before"
+  for t in "${TABLES[@]}"; do
+    if grep -q "^${t} " <<<"$before"; then
+      pass "seeded ${t} (${where})"
+    else
+      fail "seeded ${t} (${where}): the snapshot has no row of it"
+    fi
+  done
+}
+
+# check_upgrade reads the seeded rows from felis-postgres and compares them with what the
+# release's database held. A database the upgrade moved off the host also left the marker.
+check_upgrade() {
+  local where after seq
+  if [ ! -s "${STATE_DIR}/before" ]; then
+    fail "the seed step left its snapshot in ${STATE_DIR}/before"
+    return
+  fi
+  where="$(db_where)"
+  if [ "$where" != pod ]; then
+    fail "the database lives in felis-postgres after the upgrade"
+    return
+  fi
+  if [ "$(cat "${STATE_DIR}/where")" = host ]; then
+    if [ -f "$PG_MOVED_MARKER" ]; then
+      pass "the upgrade moved the seeded host database into felis-postgres"
+    else
+      fail "the upgrade moved the seeded host database into felis-postgres: no ${PG_MOVED_MARKER}"
+    fi
+  fi
+  if ! after="$(snapshot pod)"; then
+    fail "read the seeded rows from felis-postgres"
+    return
+  fi
+  if [ "$after" = "$(cat "${STATE_DIR}/before")" ]; then
+    pass "every seeded row came through the upgrade unchanged"
+  else
+    fail "the seeded rows changed across the upgrade (< release, > felis-postgres):"
+    diff "${STATE_DIR}/before" <(printf '%s\n' "$after") || true
+  fi
+  # A restore that loses a sequence's position hands out ids that are taken: the next
+  # audit row would fail on its primary key.
+  seq="$(db_psql pod <<<"SELECT coalesce(pg_sequence_last_value(pg_get_serial_sequence('audit_logs', 'id')::regclass), 0) >= (SELECT max(id) FROM audit_logs)")" || seq=error
+  if [ "$seq" = t ]; then
+    pass "audit_logs' id sequence is past every seeded id"
+  else
+    fail "audit_logs' id sequence is past every seeded id (${seq})"
+  fi
+}
+
+case "$phase" in
+  seed) seed ;;
+  check) check_upgrade ;;
+  *)
+    echo "usage: e2e_seed.sh seed|check" >&2
+    exit 2
+    ;;
+esac
+
+if [ "$fails" -eq 0 ]; then
+  echo "ALL PASS (seed ${phase})"
+  exit 0
+fi
+echo "${fails} FAILED (seed ${phase})"
+exit 1
diff --git a/docs/operations.md b/docs/operations.md
index 49298e5..c583355 100644
--- a/docs/operations.md
+++ b/docs/operations.md
@@ -21,7 +21,7 @@ release pins by digest, with its data on the host in `/var/lib/felis/postgres`.
 | OS family | Package manager | Architectures | Status |
 |---|---|---|---|
 | CentOS Stream 9 (firewalld active, SELinux enforcing) | dnf | aarch64 | **[VM-VERIFIED]** fresh install from release assets and its rerun, upgrade from v0.1.0 (moving the database off the host PostgreSQL 13 into felis-postgres), uninstall and reinstall |
-| Ubuntu 24.04 LTS | apt | x86_64 | **[CI]** fresh install and same-commit rerun from the pushed commit's release assets; the README's one-line install as a new host runs it (the newest release's own assets); upgrade from the newest release, installed from its assets by its own installer, onto them; the on-host build weekly |
+| Ubuntu 24.04 LTS | apt | x86_64 | **[CI]** fresh install and same-commit rerun from the pushed commit's release assets; the README's one-line install as a new host runs it (the newest release's own assets); upgrade from the newest release, installed from its assets by its own installer and seeded with rows in twelve tables, onto them, every seeded row read back unchanged; the on-host build weekly |
 | RHEL / Rocky / Alma 9, Fedora | dnf | x86_64, aarch64 | [CODE-ONLY] same code path as CentOS Stream |
 | Debian 12, other Ubuntu releases | apt | x86_64, aarch64 | [CODE-ONLY] |
 | openSUSE Leap / Tumbleweed | zypper | x86_64, aarch64 | [CODE-ONLY] |
@@ -570,7 +570,7 @@ instead, scale felis-postgres to 0, move the new major's directory out of
 `/var/lib/felis/postgres`, move `postgres-18.old` back as `/var/lib/felis/postgres/18`, and
 rerun the older release's installer.
 
-### The database's move into k3s [VM-VERIFIED]
+### The database's move into k3s [VM-VERIFIED] [CI]
 
 Releases before the move ran the database on a PostgreSQL the installer installed on the
 host. The first rerun of a release with felis-postgres moves it, once:
@@ -587,6 +587,11 @@ host. The first rerun of a release with felis-postgres moves it, once:
    holds other databases keeps running; its `felis` copy is then reachable over loopback
    only.
 
+The e2e upgrade job seeds the newest release's database with users, links, sessions, audit
+rows, backups, builds and the rest (`deploy/e2e_seed.sh`), upgrades, and checks that
+felis-postgres holds every seeded row with the same values after the pending migrations.
+While that release is v0.1.0, the upgrade is this move.
+
 From then on the host config points at felis-postgres (`127.0.0.1:15432`, and
 `deployment = "felis/felis-postgres"`, through which `felis db` runs `pg_dump`, `psql`
 and `pg_restore` inside the pod) and the pods at `felis-postgres.felis.svc:5432`.