Latency/packet loss graph demo for InfluxDB 2 (Flux Query Language) + Grafana

grafana-latency-loss-graph.md

Grafana dashboard setup:

• Query options
  • Max data points: 600 if your data collection interval is 1min, adjust on your needs
• Visualization: Graph
• Display:
  • Bars: off
  • Lines: off
  • Points: off
• Series overrides
  • max
    • Color: Yellow
    • Fill gradient: 2
    • Fill below to: min
  • mean
    • Color: Green
    • Fill gradient: 2
    • Fill below to: min
  • min
    • Color: Blue
    • Lines: true
  • packet_loss%:
    • Color: Red
    • Y-axis: 2
    • Lines: true
    • Line width: 2
    • Staircase line: true
• Axes
  • Left Y
    • Unit: ms
    • Y-Min: 0
  • Right Y
    • Unit: Percent
    • Y-Min: 0
    • Y-Max: 100

References:

• https://peter.run/blog/2019-07-28-visualising-latency-variance-in-grafana-in-2019/
• grafana/influxdb-flux-datasource#42 (comment)

grafana-latency-loss-graph.png

query.flux

bucket_name = "telegraf"
src_hostname = "hostname"
dst_url = "8.8.8.8"
outputColumns = ["_field", "_time", "_value"]

rawLatencyQuery = from(bucket: bucket_name)
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "ping")
  |> filter(fn: (r) => r["_field"] == "average_response_ms")
  |> filter(fn: (r) => r["host"] == src_hostname)
  |> filter(fn: (r) => r["url"] == dst_url)

meanLatencyRet = rawLatencyQuery
  |> map(fn: (r) => ({_value: r._value, _time: r._time, _field: "mean"}))
  |> keep(columns: outputColumns)
  |> aggregateWindow(every: v.windowPeriod, fn: mean, createEmpty: false)
  |> yield(name: "mean")

maxLatencyRet = rawLatencyQuery
  |> map(fn: (r) => ({_value: r._value, _time: r._time, _field: "max"}))
  |> keep(columns: outputColumns)
  |> aggregateWindow(every: v.windowPeriod, fn: max, createEmpty: false)
  |> yield(name: "max")

minLatencyRet = rawLatencyQuery
  |> map(fn: (r) => ({_value: r._value, _time: r._time, _field: "min"}))
  |> keep(columns: outputColumns)
  |> aggregateWindow(every: v.windowPeriod, fn: min, createEmpty: false)
  |> yield(name: "min")

packetLossRateRawQuery = from(bucket: bucket_name)
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "ping")
  |> filter(fn: (r) => r["_field"] == "percent_packet_loss")
  |> filter(fn: (r) => r["host"] == src_hostname)
  |> filter(fn: (r) => r["url"] == dst_url)

packetLossRateRet = packetLossRateRawQuery
  |> map(fn: (r) => ({_value: r._value, _time: r._time, _field: "packet_loss%"}))
  |> keep(columns: outputColumns)
  |> aggregateWindow(every: v.windowPeriod, fn: mean, createEmpty: false)
  |> yield(name: "packet_loss_rate")

telegraf.conf

[global_tags]
  dc = "he-fmt2"

[agent]
  interval = "60s"
  round_interval = true
  metric_batch_size = 1000
  metric_buffer_limit = 100000
  collection_jitter = "10s"
  flush_interval = "10s"
  flush_jitter = "10s"
  precision = "1s"
  omit_hostname = false

[[outputs.influxdb_v2]]
  urls = ["https://influxdb.corp.contoso.com"]
  token = "5pyJ5YaF6ay877yM5YGc5q2i5Lqk5piT77yB"
  organization = "demo_org"
  bucket = "telegraf"
  user_agent = "telegraf"
  content_encoding = "gzip"
  influx_uint_support = true
  # tls_ca = "/etc/telegraf/rootca.crt"
  # tls_cert = "/etc/telegraf/cert.pem"
  # tls_key = "/etc/telegraf/key.pem"
  insecure_skip_verify = false

[[inputs.ping]]
  urls = [
    "8.8.8.8",
  ]
  method = "exec" # do not use "native" as of Telegraf 1.16.1
  count = 10
  ping_interval = 1.0
  timeout = 1.0
  deadline = 20