Turn on/off enclosure slot fault LED even when disk isn't present

Previously when a drive faulted, the statechange-led.sh script would lookup
the drive's LED sysfs entry in /sys/block/sd*/device/enclosure_device, and
turn it on.  During testing we noticed that if you pulled out a drive, or if
the drive was so badly broken that it no longer appeared to Linux, that the
/sys/block/sd* path would be removed, and the script could not lookup the
LED entry.

To fix this, this patch looks up the disks's more persistent
"/sys/class/enclosure/X:X:X:X/Slot N" LED sysfs path at pool import.  It then
passes that path to the statechange-led script to use, rather than having the
script look it up on the fly.  This allows the script to turn on/off the slot
LEDs even when the drive is missing.

Closes #5309 
Closes #2375 

1 files changed, 55 insertions, 44 deletions

diff --git a/cmd/zed/zed.d/statechange-led.sh b/cmd/zed/zed.d/statechange-led.sh
index ca911d2b9..af1a14a9a 100755
--- a/cmd/zed/zed.d/statechange-led.sh
+++ b/cmd/zed/zed.d/statechange-led.sh
@@ -2,8 +2,13 @@
 #
 # Turn off/on the VDEV's enclosure fault LEDs when the pool's state changes.
 #
-# Turn LED on if the VDEV becomes faulted/degraded, and turn it back off when
-# it's healthy again.  This requires that your enclosure be supported by the
+# Turn LED on if the VDEV becomes faulted or degraded, and turn it back off
+# when it's online again.  It will also turn on the LED (or keep it on) if
+# the drive becomes unavailable, unless the drive was in was a previously
+# online state (online->unavail is a normal state transition during an
+# autoreplace).
+#
+# This script requires that your enclosure be supported by the
 # Linux SCSI enclosure services (ses) driver.  The script will do nothing
 # if you have no enclosure, or if your enclosure isn't supported.
 #
@@ -13,76 +18,82 @@
 #   0: enclosure led successfully set
 #   1: enclosure leds not not available
 #   2: enclosure leds administratively disabled
-#   3: ZED built without libdevmapper
+#   3: ZED didn't pass enclosure sysfs path
+#   4: Enclosure sysfs path doesn't exist
 
 [ -f "${ZED_ZEDLET_DIR}/zed.rc" ] && . "${ZED_ZEDLET_DIR}/zed.rc"
 . "${ZED_ZEDLET_DIR}/zed-functions.sh"
 
-# ZEVENT_VDEV_UPATH will not be present if ZFS is not built with libdevmapper
-[ -n "${ZEVENT_VDEV_UPATH}" ] || exit 3
+if [ ! -d /sys/class/enclosure ] ; then
+	exit 1
+fi
 
 if [ "${ZED_USE_ENCLOSURE_LEDS}" != "1" ] ; then
 	exit 2
 fi
 
-if [ ! -d /sys/class/enclosure ] ; then
-	exit 1
-fi
+[ -n "${ZEVENT_VDEV_ENC_SYSFS_PATH}" ] || exit 3
+
+[ -e "${ZEVENT_VDEV_ENC_SYSFS_PATH}/fault" ] || exit 4
 
 # Turn on/off enclosure LEDs
 function led
 {
-	name=$1
+	file="$1/fault"
 	val=$2
 
 	# We want to check the current state first, since writing to the
 	# 'fault' entry always always causes a SES command, even if the
 	# current state is already what you want.
-	if [ -e /sys/block/$name/device/enclosure_device*/fault ] ; then
-		# We have to do some monkey business to deal with spaces in
-		# enclosure_device names.  I've seen horrible things like this: 
-		#
-		# '/sys/block/sdfw/device/enclosure_device:SLOT 43 41  /fault'
-		#
-		# ...so escape all spaces.
-		file=`ls /sys/block/$name/device/enclosure_device*/fault | sed 's/\s/\\ /g'`
-
-		current=`cat "$file"`
+	current=$(cat "${file}")
 
-		# On some enclosures if you write 1 to fault, and read it back,
-		# it will return 2.  Treat all non-zero values as 1 for
-		# simplicity.
-		if [ "$current" != "0" ] ; then
-			current=1
-		fi
+	# On some enclosures if you write 1 to fault, and read it back,
+	# it will return 2.  Treat all non-zero values as 1 for
+	# simplicity.
+	if [ "$current" != "0" ] ; then
+		current=1
+	fi
 
-		if [ "$current" != "$val" ] ; then
-			# Set the value twice.  I've seen enclosures that were
-			# flakey about setting it the first time.
-			echo $val > "$file"
-			echo $val > "$file"
-		fi
+	if [ "$current" != "$val" ] ; then
+		# Set the value twice.  I've seen enclosures that were
+		# flakey about setting it the first time.
+		echo "$val" > "$file"
+		echo "$val" > "$file"
 	fi
 }
 
 # Decide whether to turn on/off an LED based on the state
 # Pass in path name and fault string ("ONLINE"/"FAULTED"/"DEGRADED"...etc)
+#
+# We only turn on LEDs when a drive becomes FAULTED, DEGRADED, or UNAVAIL and
+# only turn it on when it comes back ONLINE.  All other states are ignored, and
+# keep the previous LED state.  
 function process {
-	# path=/dev/sda, fault=
-
-	path=$1
+	path="$1"
 	fault=$2
-	name=`basename $path`
-
-	if [ -z "$name" ] ; then
-		return
-	fi
-
+	prev=$3
 	if [ "$fault" == "FAULTED" ] || [ "$fault" == "DEGRADED" ] ; then
-		led $name 1
-	else
-		led $name 0
+		led "$path" 1
+	elif [ "$fault" == "UNAVAIL" ] && [ "$prev" != "ONLINE" ] ; then
+		# For the most part, UNAVAIL should turn on the LED.  However,
+		# during an autoreplace, we see our new drive go online,
+		# followed by our "old" drive going ONLINE->UNAVAIL. Since the
+		# "old" drive has the same slot information, we want to ignore
+		# the ONLINE->UNAVAIL event.
+		#
+		# 	NAME             STATE     READ WRITE CKSUM
+		#	mypool3          DEGRADED     0     0     0
+		#	  mirror-0       DEGRADED     0     0     0
+		#	    A1           ONLINE       0     0     0
+		#	    A2           ONLINE       0   880     0
+		#	    replacing-3  UNAVAIL      0     0     0
+		#	      old        UNAVAIL      0 2.93K     0  corrupted data
+		#	      A3         ONLINE       0     0   156  (resilvering)
+		led "$path" 1
+	elif [ "$fault" == "ONLINE" ] ; then
+		led "$path" 0
 	fi
 }
 
-process "$ZEVENT_VDEV_UPATH" "$ZEVENT_VDEV_STATE_STR"
+process "$ZEVENT_VDEV_ENC_SYSFS_PATH" "$ZEVENT_VDEV_STATE_STR" \
+    "$ZEVENT_VDEV_LASTSTATE_STR"