Reworked gluster status based on heal summary

checkmk/custom-files/local/gluster

 #!/bin/bash
 
-# gluster agent plugin
+# we parse the output of gluster volume heal [vol] info split-brain
+# which looks like that:
 #
-# * determine all volumes
-# * determie all bricks for this volume
-# * discover brick information - yields output like that:
+# Brick t3web-00.it.hs-hannover.de:/srv/gluster/fileadmin
+# Status: Connected
+# Number of entries in split-brain: 0
 #
-# Status of volume: fileadmin
-# ------------------------------------------------------------------------------
-# Brick                : Brick t3web-00.it.hs-hannover.de:/srv/gluster/fileadmin
-# TCP Port             : 49152
-# RDMA Port            : 0
-# Online               : Y
-# Pid                  : 609
-# File System          : xfs
-# Device               : /dev/sdd1
-# Mount Options        : rw,relatime,attr2,inode64,noquota
-# Inode Size           : N/A
-# Disk Space Free      : 108.5GB
-# Total Disk Space     : 120.0GB
-# Inode Count          : 62914048
-# Free Inodes          : 62693863
-#
-# We currently only utilize the Online: Y information.
+# Brick t3web-01.it.hs-hannover.de:/srv/gluster/fileadmin
+# Status: Connected
+# Number of entries in split-brain: 0
 
 hostname=$(hostname)
+entries_warn=500
+entries_crit=1000
+entries_min=0
+entries_max=2000
 
 # iterate over all available volumns
 for volume in $(gluster volume list)
 do
-  # find all bricks for this volume
-	for brick in $(gluster volume info $volume | grep -E '^Brick[0-9]: ' | sed -e 's/Brick[0-9]: //g')
-	do
-		# skip if this is not your own brick!
-		if [[ "$brick" != *"$hostname"* ]]; then
+  gluster volume heal $volume info summary > /tmp/gluster_heal_summary
+  # check heal info summary
+  while read -r brickline; do
+    read -r statusline
+    read -r entriesline
+    # we do not use the 3 lines below - may use them for more detailed metrics
+    read -r entriespendingline
+    read -r entriessplitbrainline
+    read -r entrieshealingline
+    # bricks are separated by empty line
+    read -r emptyline
+    # skip chunks that are not ours
+    if [[ "$brickline" != *"$hostname"* ]]
+    then
       continue
     fi
-		# grab the full status of this brick
-		brick_status=$(gluster volume status $volume $brick detail)
-		online=$(echo "$brick_status" | grep -E '^Online' | cut -d ':' -f2 | tr -d '[:space:]')
-		case "$online" in
-		 Y) brick_status=0 ;; # Online means OK
-		 N) brick_status=1 ;; # Not online is WARN
-		 *) brick_status=3 ;; # everything else we don't know...
+    # parse lines of the last brick, since it is located on this gluster cluster node
+    brick=$(echo "$brickline" | cut -f2 -d' ')
+    heal_status=$(echo "$statusline" | cut -d ':' -f2 | tr -d '[:space:]')
+    entries=$(echo "$entriesline" | cut -d ':' -f2 | tr -d '[:space:]')
+    case "$heal_status" in
+     Connected) chmk_status=0 ;; # Online means OK
+             *) chmk_status=3 ;; # everything else we don't know...
     esac
-		echo "$brick_status $brick - Online status of volume '$volume' is: '$online'"
-	done
+    metrics="healing_entries=$entries;$entries_warn;$entries_crit;$entries_min;$entries_max"
+    msg="Online status of volume '$volume' brick '$brick' is: '$heal_status'"
+    echo "$chmk_status gluster_volume:$volume $metrics $msg"
+  done < /tmp/gluster_heal_summary
 done