From e338d15922aec307d90fb2e33007e9db758dafba Mon Sep 17 00:00:00 2001
From: Josue Zamudio <josuezamudio1@gmail.com>
Date: Mon, 28 Sep 2026 14:03:00 -0700
Subject: [PATCH] add shikane-watchdog user service to auto-recover shikane
 deadlocks

---
 .config/sway/scripts/shikane-watchdog.sh      | 49 +++++++++++++++++++
 .config/systemd/user/shikane-watchdog.service |  6 +++
 .config/systemd/user/shikane-watchdog.timer   | 10 ++++
 .../shikane-watchdog.timer                    |  1 +
 README.md                                     | 22 +++++++++
 5 files changed, 88 insertions(+)
 create mode 100755 .config/sway/scripts/shikane-watchdog.sh
 create mode 100644 .config/systemd/user/shikane-watchdog.service
 create mode 100644 .config/systemd/user/shikane-watchdog.timer
 create mode 120000 .config/systemd/user/timers.target.wants/shikane-watchdog.timer

diff --git a/.config/sway/scripts/shikane-watchdog.sh b/.config/sway/scripts/shikane-watchdog.sh
new file mode 100755
index 0000000..494b91c
--- /dev/null
+++ b/.config/sway/scripts/shikane-watchdog.sh
@@ -0,0 +1,49 @@
+#!/bin/sh
+# shikane-watchdog: auto-recover shikane from its deadlock bug.
+#
+# Bug (shikane 1.1.1): if a variant application is discarded mid-way
+# (e.g. by a failing DP atomic commit), the daemon gets stuck in
+# VariantInProgress/Discarded and keeps logging
+#   [WARN shikane::variant] Received invalid input AtomicChangeDone at state Discarded
+# `shikanectl reload` does not recover it; a service restart does
+# (re-applies the active profile in ~1s).
+#
+# A healthy daemon's last state-machine line is "VariantApplied".
+
+set -u
+
+unit=shikane
+stuck=0
+
+logs3=$(journalctl --user -u "$unit" -q --since "-3min" 2>/dev/null)
+[ -z "$logs3" ] && exit 0
+
+latest3=$(printf '%s\n' "$logs3" | grep 'daemon::state_machine' | tail -1)
+
+if [ -n "$latest3" ]; then
+    case "$latest3" in
+        *VariantApplied*)
+            : ;;  # healthy
+        *)
+            # InProgress/Discarded within the last 3 min: only a deadlock
+            # if the WARN signature is present (a legitimate slow transition
+            # has no WARN)
+            printf '%s\n' "$logs3" | grep -q 'invalid input' && stuck=1
+            ;;
+    esac
+else
+    # No state change in the last 3 min — check the last known state
+    last10=$(journalctl --user -u "$unit" -q --since "-10min" 2>/dev/null \
+        | grep 'daemon::state_machine' | tail -1)
+    case "$last10" in
+        '' | *VariantApplied*)
+            : ;;  # stable and applied, or nothing to check
+        *)
+            stuck=1 ;;  # stuck in-progress/discarded for >3 min
+    esac
+fi
+
+if [ "$stuck" -eq 1 ]; then
+    echo "shikane-watchdog: shikane stuck, restarting" >&2
+    systemctl --user restart "$unit"
+fi
diff --git a/.config/systemd/user/shikane-watchdog.service b/.config/systemd/user/shikane-watchdog.service
new file mode 100644
index 0000000..9252530
--- /dev/null
+++ b/.config/systemd/user/shikane-watchdog.service
@@ -0,0 +1,6 @@
+[Unit]
+Description=shikane watchdog - restart shikane if its state machine is stuck
+
+[Service]
+Type=oneshot
+ExecStart=/home/work/.config/sway/scripts/shikane-watchdog.sh
diff --git a/.config/systemd/user/shikane-watchdog.timer b/.config/systemd/user/shikane-watchdog.timer
new file mode 100644
index 0000000..220a5f2
--- /dev/null
+++ b/.config/systemd/user/shikane-watchdog.timer
@@ -0,0 +1,10 @@
+[Unit]
+Description=Run shikane watchdog every 2 minutes
+
+[Timer]
+OnBootSec=2min
+OnUnitInactiveSec=2min
+AccuracySec=5s
+
+[Install]
+WantedBy=timers.target
diff --git a/.config/systemd/user/timers.target.wants/shikane-watchdog.timer b/.config/systemd/user/timers.target.wants/shikane-watchdog.timer
new file mode 120000
index 0000000..eec543c
--- /dev/null
+++ b/.config/systemd/user/timers.target.wants/shikane-watchdog.timer
@@ -0,0 +1 @@
+/home/work/.config/systemd/user/shikane-watchdog.timer
\ No newline at end of file
diff --git a/README.md b/README.md
index 2c4fe1a..582ef7c 100644
--- a/README.md
+++ b/README.md
@@ -11,3 +11,25 @@ stow .
 ```
 If stow is not installed use ./install.sh from .dotfiles repo
 
+# User services
+
+## shikane-watchdog
+
+`.config/systemd/user/shikane-watchdog.timer` (every 2 min, user unit)
+runs `.config/sway/scripts/shikane-watchdog.sh`, which restarts the
+`shikane` user service if its state machine is stuck.
+
+Why: shikane 1.1.1 deadlocks when a profile application is discarded
+mid-way (e.g. by a failing DP atomic commit on the external display).
+The daemon then keeps logging
+`[WARN shikane::variant] Received invalid input AtomicChangeDone at state Discarded`
+and `shikanectl reload` no longer recovers it. The watchdog detects
+that signature in the journal and runs `systemctl --user restart shikane`
+(re-applies the active profile in ~1s). It never restarts a healthy
+or legitimately transitioning daemon.
+
+```console
+systemctl --user status shikane-watchdog.timer      # check it's active
+systemctl --user disable --now shikane-watchdog.timer  # disable
+```
+