diff --git a/ansible/roles/backup_store/defaults/main.yml b/ansible/roles/backup_store/defaults/main.yml
index e6d17a3..9032d55 100644
--- a/ansible/roles/backup_store/defaults/main.yml
+++ b/ansible/roles/backup_store/defaults/main.yml
@@ -22,7 +22,10 @@ backup_store_check_push_token: ""
 # source's dump timestamp out of the artefact filename, so running it more often
 # catches "the source stopped dumping" within hours rather than a day, and lets
 # the Gatus heartbeat be 7h instead of 30h.
-backup_store_check_on_calendar: "*-*-* 05:30:00,11:30:00,17:30:00,23:30:00"
+# 05:30 then every 6h. Note the syntax: a comma-separated list of full
+# times ("05:30:00,11:30:00,...") is NOT valid and systemd rejects the unit
+# with "bad unit file setting" - validated with `systemd-analyze calendar`.
+backup_store_check_on_calendar: "*-*-* 05/6:30:00"
 
 # An artefact older than this is stale. Sources dump daily at 02:00-02:30 and the
 # pull is at 04:00, so 26h tolerates exactly one missed night before alarming.
diff --git a/ansible/roles/healthcheck/templates/checks/disk-usage.sh.j2 b/ansible/roles/healthcheck/templates/checks/disk-usage.sh.j2
index b0e76ce..86fd3a6 100644
--- a/ansible/roles/healthcheck/templates/checks/disk-usage.sh.j2
+++ b/ansible/roles/healthcheck/templates/checks/disk-usage.sh.j2
@@ -1,19 +1,35 @@
-    # Every real filesystem must be under the threshold. tmpfs, devtmpfs,
-    # squashfs and overlay are excluded: they are either RAM, read-only, or
-    # container layers, and none of them fills up in a way an operator can act on.
+    # Every real filesystem must be under the threshold. The exclusions are all
+    # pseudo-filesystems whose "usage" is not a disk-space fact an operator can
+    # act on: tmpfs and devtmpfs are RAM, squashfs is read-only, overlay is a
+    # container layer, and efivarfs is UEFI variable storage - which sits near
+    # 90% full on a perfectly healthy machine and would page every single day.
+    #
+    # NOTE: no -P. `df -P --output=...` fails with "options -P and --output are
+    # mutually exclusive", and an earlier version of this check had both - so df
+    # errored, the loop read nothing, and it reported "max 0% on /" and exited 0
+    # on every host regardless of real usage. A disk check that always says 0%
+    # is worse than no disk check, so the guard below treats "no filesystems
+    # read" as a failure rather than as health.
     local threshold={{ healthcheck_disk_threshold }}
-    local worst=0 worst_mount="" over=""
+    local worst=0 worst_mount="" over="" seen=0
 
     while read -r pct mount; do
         pct=${pct%\%}
         [ -z "$pct" ] && continue
+        case "$pct" in (*[!0-9]*) continue;; esac
+        seen=$((seen + 1))
         if [ "$pct" -gt "$worst" ]; then worst=$pct; worst_mount=$mount; fi
         if [ "$pct" -ge "$threshold" ]; then over="${over}${over:+, }${mount} ${pct}%"; fi
-    done < <(df -P -x tmpfs -x devtmpfs -x squashfs -x overlay --output=pcent,target 2>/dev/null | tail -n +2)
+    done < <(df -x tmpfs -x devtmpfs -x squashfs -x overlay -x efivarfs -x ramfs --output=pcent,target 2>/dev/null | tail -n +2)
+
+    if [ "$seen" -eq 0 ]; then
+        MESSAGE="df returned no filesystems - the check itself is broken"
+        return 1
+    fi
 
     if [ -n "$over" ]; then
         MESSAGE="over ${threshold}%: ${over}"
         return 1
     fi
-    MESSAGE="max ${worst}% on ${worst_mount:-/}"
+    MESSAGE="max ${worst}% on ${worst_mount:-/} (${seen} filesystems)"
     return 0