From dfc4915ffb73b5b12ce50bcf6c1981afa9928f0a Mon Sep 17 00:00:00 2001
From: Ilya Mashchenko <ilya@netdata.cloud>
Date: Mon, 24 May 2021 14:36:12 +0300
Subject: health: make stocks alarms less sensitive (2) (#11153)

---
 health/health.d/disks.conf   | 10 ++--------
 health/health.d/net.conf     |  4 ++--
 health/health.d/ram.conf     |  8 ++++----
 health/health.d/web_log.conf |  4 ----
 4 files changed, 8 insertions(+), 18 deletions(-)

diff --git a/health/health.d/disks.conf b/health/health.d/disks.conf
index 60f8faed99..0b88106181 100644
--- a/health/health.d/disks.conf
+++ b/health/health.d/disks.conf
@@ -145,10 +145,7 @@ component: Disk
    lookup: average -10m unaligned
     units: %
     every: 1m
-    green: 90
-      red: 98
-     warn: $this > $green * (($status >= $WARNING)  ? (0.7) : (1))
-     crit: $this > $red   * (($status == $CRITICAL) ? (0.7) : (1))
+     warn: $this > 98 * (($status >= $WARNING)  ? (0.7) : (1))
     delay: down 15m multiplier 1.2 max 1h
      info: average percentage of time $family disk was busy over the last 10 minutes
        to: silent
@@ -170,10 +167,7 @@ component: Disk
    lookup: average -10m unaligned
     units: ms
     every: 1m
-    green: 2000
-      red: 5000
-     warn: $this > $green * (($status >= $WARNING)  ? (0.7) : (1))
-     crit: $this > $red   * (($status == $CRITICAL) ? (0.7) : (1))
+     warn: $this > 5000 * (($status >= $WARNING)  ? (0.7) : (1))
     delay: down 15m multiplier 1.2 max 1h
      info: average backlog size of the $family disk over the last 10 minutes
        to: silent
diff --git a/health/health.d/net.conf b/health/health.d/net.conf
index 04219e163a..db480031ce 100644
--- a/health/health.d/net.conf
+++ b/health/health.d/net.conf
@@ -96,7 +96,7 @@ component: Network
     hosts: *
  families: !net* !wl* *
    lookup: sum -10m unaligned absolute of received
-     calc: (($inbound_packets_dropped != nan AND $this > 1000) ? ($inbound_packets_dropped * 100 / $this) : (0))
+     calc: (($inbound_packets_dropped != nan AND $this > 10000) ? ($inbound_packets_dropped * 100 / $this) : (0))
     units: %
     every: 1m
      warn: $this >= 2
@@ -130,7 +130,7 @@ component: Network
     hosts: *
  families: wl*
    lookup: sum -10m unaligned absolute of received
-     calc: (($inbound_packets_dropped != nan AND $this > 1000) ? ($inbound_packets_dropped * 100 / $this) : (0))
+     calc: (($inbound_packets_dropped != nan AND $this > 10000) ? ($inbound_packets_dropped * 100 / $this) : (0))
     units: %
     every: 1m
      warn: $this >= 10
diff --git a/health/health.d/ram.conf b/health/health.d/ram.conf
index 0e3cc29faf..450c548171 100644
--- a/health/health.d/ram.conf
+++ b/health/health.d/ram.conf
@@ -50,12 +50,12 @@ component: Memory
       on: mem.oom_kill
       os: linux
    hosts: *
-  lookup: sum -1m unaligned
+  lookup: sum -30m unaligned
    units: kills
-   every: 10s
+   every: 5m
     warn: $this > 0
-   delay: down 5m
-    info: number of out of memory kills in the last minute
+   delay: down 10m
+    info: number of out of memory kills in the last 30 minutes
       to: sysadmin
 
 ## FreeBSD
diff --git a/health/health.d/web_log.conf b/health/health.d/web_log.conf
index 127c9a9c6a..c1237fa06d 100644
--- a/health/health.d/web_log.conf
+++ b/health/health.d/web_log.conf
@@ -66,7 +66,6 @@ component: Web log
     units: %
     every: 10s
      warn: ($1m_requests > 120) ? ($this > (($status >= $WARNING ) ? (  1 ) : ( 20 )) ) : ( 0 )
-     crit: ($1m_requests > 120) ? ($this > (($status == $CRITICAL) ? ( 20 ) : ( 30 )) ) : ( 0 )
     delay: up 2m down 15m multiplier 1.5 max 1h
      info: ratio of redirection HTTP requests over the last minute (3xx except 304)
        to: webmaster
@@ -82,7 +81,6 @@ component: Web log
     units: %
     every: 10s
      warn: ($1m_requests > 120) ? ($this > (($status >= $WARNING)  ? ( 10 ) : ( 30 )) ) : ( 0 )
-     crit: ($1m_requests > 120) ? ($this > (($status == $CRITICAL) ? ( 30 ) : ( 50 )) ) : ( 0 )
     delay: up 2m down 15m multiplier 1.5 max 1h
      info: ratio of client error HTTP requests over the last minute (4xx except 401)
        to: webmaster
@@ -335,7 +333,6 @@ component: Web log
     units: %
     every: 10s
      warn: ($web_log_1m_requests > 120) ? ($this > (($status >= $WARNING ) ? (  1 ) : ( 20 )) ) : ( 0 )
-     crit: ($web_log_1m_requests > 120) ? ($this > (($status == $CRITICAL) ? ( 20 ) : ( 30 )) ) : ( 0 )
     delay: up 2m down 15m multiplier 1.5 max 1h
      info: ratio of redirection HTTP requests over the last minute (3xx except 304)
        to: webmaster
@@ -351,7 +348,6 @@ component: Web log
     units: %
     every: 10s
      warn: ($web_log_1m_requests > 120) ? ($this > (($status >= $WARNING)  ? ( 10 ) : ( 30 )) ) : ( 0 )
-     crit: ($web_log_1m_requests > 120) ? ($this > (($status == $CRITICAL) ? ( 30 ) : ( 50 )) ) : ( 0 )
     delay: up 2m down 15m multiplier 1.5 max 1h
      info: ratio of client error HTTP requests over the last minute (4xx except 401)
        to: webmaster
-- 
cgit v1.2.3