machines/stellio: persist the journal

Storage=volatile left nothing behind for the 2026-09-23/24 reboots: the
last persisted entry is journald stopping during the 2026-09-22 switch,
and /sys/fs/pstore is empty too. Keep the eMMC-wear intent with a 256M
cap and a 1 min sync interval matching the ext4 commit=60.

machines/stellio/configuration.nix

@@ -35,20 +35,22 @@
     clan.core.settings.machine.description = "Flint-2 router";
 
     # Root is ext4 on the Flint 2's eMMC: keep steady-state writes off it.
-    # Logs live in RAM (64M; the router needs them for pppd/kea debugging,
+    # The journal is persistent but bounded (256M) and synced once a minute
-    # unlike sirius which drops them), /tmp is tmpfs, no core dumps, and
+    # (the router needs logs for pppd/kea debugging and, after the Sep 2026
-    # ext4 skips atime updates and flushes its journal every 60 s instead
+    # unlogged reboots, for post-mortem; volatile storage left nothing).
-    # of 5 s so the small periodic writes (Kea leases, timesyncd clock,
+    # /tmp is tmpfs, no core dumps, and ext4 skips atime updates and
-    # speedtest textfile) coalesce. A power cut loses at most a minute of
+    # flushes its journal every 60 s instead of 5 s so the small periodic
-    # that state, none of which matters on a gateway.
+    # writes (Kea leases, timesyncd clock, speedtest textfile, journal)
+    # coalesce. A power cut loses at most a minute of that state.
     fileSystems."/".options = [
       "noatime"
       "commit=60"
     ];
     boot.tmp.useTmpfs = true;
     services.journald.settings.Journal = {
-      Storage = "volatile";
+      Storage = "persistent";
-      RuntimeMaxUse = "64M";
+      SystemMaxUse = "256M";
+      SyncIntervalSec = "1m";
     };
     systemd.coredump.enable = false;