1 files changed, 80 insertions, 0 deletions
diff --git a/health/health.d/riakkv.conf b/health/health.d/riakkv.conf
new file mode 100644
index 000000000..745302778
--- /dev/null
+++ b/health/health.d/riakkv.conf
@@ -0,0 +1,80 @@
+# Ensure that Riak is running.  template: riak_last_collected_secs
+template: riak_last_collected_secs
+      on: riak.kv.throughput
+    calc: $now - $last_collected_t
+   units: seconds ago
+   every: 10s
+    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
+
+# Warn if a list keys operation is running.
+template: riak_list_keys_active
+      on: riak.core.fsm_active
+    calc: $list_fsm_active
+   units: state machines
+   every: 10s
+    warn: $list_fsm_active > 0
+    info: number of currently running list keys finite state machines
+      to: dba
+
+
+## Timing healthchecks
+# KV GET
+template: 1h_kv_get_mean_latency
+      on: riak.kv.latency.get
+    calc: $node_get_fsm_time_mean
+  lookup: average -1h unaligned of time
+   every: 30s
+   units: ms
+    info: mean average KV GET latency over the last hour
+
+template: riak_kv_get_slow
+      on: riak.kv.latency.get
+    calc: $mean
+  lookup: average -3m unaligned of time
+   units: ms
+   every: 10s
+    warn: ($this > ($1h_kv_get_mean_latency * 2) )
+    crit: ($this > ($1h_kv_get_mean_latency * 3) )
+    info: average KV GET time over the last 3 minutes, compared to the average over the last hour
+   delay: down 5m multiplier 1.5 max 1h
+      to: dba
+
+# KV PUT
+template: 1h_kv_put_mean_latency
+      on: riak.kv.latency.put
+    calc: $node_put_fsm_time_mean
+  lookup: average -1h unaligned of time
+   every: 30s
+   units: ms
+    info: mean average KV PUT latency over the last hour
+
+template: riak_kv_put_slow
+      on: riak.kv.latency.put
+    calc: $mean
+  lookup: average -3m unaligned of time
+   units: ms
+   every: 10s
+    warn: ($this > ($1h_kv_put_mean_latency * 2) )
+    crit: ($this > ($1h_kv_put_mean_latency * 3) )
+    info: average KV PUT time over the last 3 minutes, compared to the average over the last hour
+   delay: down 5m multiplier 1.5 max 1h
+      to: dba
+
+
+## VM healthchecks
+
+# Default Erlang VM process limit: 262144
+# On systems observed, this is < 2000, but may grow depending on load.
+template: riak_vm_high_process_count
+      on: riak.vm
+    calc: $sys_process_count
+   units: processes
+   every: 10s
+    warn: $this > 10000
+    crit: $this > 100000
+    info: number of processes running in the Erlang VM (the default limit on ERTS 10.2.4 is 262144)
+      to: dba