From ba47e197056601c7db3ce8db0db277d155e8118a Mon Sep 17 00:00:00 2001
From: thiagoftsm <thiagoftsm@gmail.com>
Date: Tue, 26 May 2020 11:57:56 +0000
Subject: New alarms (exporting and Backend) (#9075)

New alarms for exporting and backend.
---
 health/health.d/exporting.conf | 34 ++++++++++++++++++++++++++++++++++
 1 file changed, 34 insertions(+)
 create mode 100644 health/health.d/exporting.conf

(limited to 'health/health.d/exporting.conf')

diff --git a/health/health.d/exporting.conf b/health/health.d/exporting.conf
new file mode 100644
index 0000000000..506cb0cf75
--- /dev/null
+++ b/health/health.d/exporting.conf
@@ -0,0 +1,34 @@
+
+template: exporting_last_buffering
+families: *
+      on: exporting_data_size
+    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 * $update_every))
+   delay: down 5m multiplier 1.5 max 1h
+    info: number of seconds since the last successful buffering of exporting data
+      to: dba
+
+template: exporting_metrics_sent
+families: *
+      on: exporting_data_size
+   units: %
+    calc: abs($sent) * 100 / abs($buffered)
+   every: 10s
+    warn: $this != 100
+   delay: down 5m multiplier 1.5 max 1h
+    info: percentage of metrics sent to the external database server
+      to: dba
+
+template: exporting_metrics_lost
+families: *
+      on: exporting_data_size
+   units: metrics
+    calc: abs($lost)
+   every: 10s
+    crit: ($this != 0) || ($status == $CRITICAL && abs($sent) == 0)
+   delay: down 5m multiplier 1.5 max 1h
+    info: number of metrics lost due to repeating failures to contact the external database server
+      to: dba
-- 
cgit v1.2.3