diff options
Diffstat (limited to 'health/health.d/cgroups.conf')
| -rw-r--r-- | health/health.d/cgroups.conf | 37 |
1 files changed, 37 insertions, 0 deletions
diff --git a/health/health.d/cgroups.conf b/health/health.d/cgroups.conf index 45b34806c..aa416c795 100644 --- a/health/health.d/cgroups.conf +++ b/health/health.d/cgroups.conf @@ -32,3 +32,40 @@ component: Memory delay: down 15m multiplier 1.5 max 1h info: cgroup memory utilization to: sysadmin + +# ----------------------------------------------------------------------------- +# check for packet storms + +# 1. calculate the rate packets are received in 1m: 1m_received_packets_rate +# 2. do the same for the last 10s +# 3. raise an alarm if the later is 10x or 20x the first +# we assume the minimum packet storm should at least have +# 10000 packets/s, average of the last 10 seconds + + template: cgroup_1m_received_packets_rate + on: cgroup.net_packets + class: Workload + type: Cgroups +component: Network + hosts: * + lookup: average -1m unaligned of received + units: packets + every: 10s + info: average number of packets received by the network interface $family over the last minute + + template: cgroup_10s_received_packets_storm + on: cgroup.net_packets + class: Workload + type: Cgroups +component: Network + hosts: * + lookup: average -10s unaligned of received + calc: $this * 100 / (($1m_received_packets_rate < 1000)?(1000):($1m_received_packets_rate)) + every: 10s + units: % + warn: $this > (($status >= $WARNING)?(200):(5000)) + crit: $this > (($status == $CRITICAL)?(5000):(6000)) + options: no-clear-notification + info: ratio of average number of received packets for the network interface $family over the last 10 seconds, \ + compared to the rate over the last minute + to: sysadmin |