added fping alarms and various aesthetic improvements to alarms

[netdata.git] / conf.d / health.d / memcached.conf

diff --git a/conf.d/health.d/memcached.conf b/conf.d/health.d/memcached.conf
index 9e795b157f1496818ea21723e746a50fd57d85bb..d248ef57a7204966a41ae9d2cf73f3dc917dddd3 100644 (file)
--- a/conf.d/health.d/memcached.conf
+++ b/conf.d/health.d/memcached.conf
@@ -2,9 +2,51 @@
 # make sure memcached is running
 
 template: memcached_last_collected_secs
-      on: memcached.connections
+      on: memcached.cache
     calc: $now - $last_collected_t
+   units: seconds ago
    every: 10s
-    warn: $this > ( 5 * $update_every)
-    crit: $this > (10 * $update_every)
+    warn: $this > (($status >= $WARNING)  ? ($update_every) : ( 5 * $update_every))
+    crit: $this > (($status == $CRITICAL) ? ($update_every) : (60 * $update_every))
+   delay: down 5m multiplier 1.5 max 1h
+    info: number of seconds since the last successful data collection
+      to: dba
 
+
+# detect if memcached cache is full
+
+template: memcached_cache_memory_usage
+      on: memcached.cache
+    calc: $used * 100 / ($used + $available)
+   units: %
+   every: 10s
+    warn: $this > (($status >= $WARNING)  ? (70) : (80))
+    crit: $this > (($status == $CRITICAL) ? (80) : (90))
+   delay: up 0 down 15m multiplier 1.5 max 1h
+    info: current cache memory usage
+      to: dba
+
+
+# find the rate memcached cache is filling
+
+template: cache_fill_rate
+      on: memcached.cache
+  lookup: min -10m at -50m unaligned of available
+    calc: ($this - $available) / (($now - $after) / 3600)
+   units: KB/hour
+   every: 1m
+    info: average rate the cache fills up (positive), or frees up (negative) space, for the last hour
+
+
+# find the hours remaining until memcached cache is full
+
+template: out_of_cache_space_time
+      on: memcached.cache
+    calc: ($cache_fill_rate > 0) ? ($available / $cache_fill_rate) : (inf)
+   units: hours
+   every: 10s
+    warn: $this > 0 and $this < (($status >= $WARNING)  ? (48) : (8))
+    crit: $this > 0 and $this < (($status == $CRITICAL) ? (24) : (2))
+   delay: down 15m multiplier 1.5 max 1h
+    info: estimated time the cache will run out of space, if the system continues to add data with the rate of the last hour
+      to: dba