InfluxDB 2 - Flux 案例

数据：

#_measurement:rig
#tag:ip
#field:gpu_count gpu数量
#field:total_khs gpu总速度
#field:device_count 机器数量，纯粹为了统计机器数用


p = influxdb_client.Point("rig").tag("ip", myaddr).field("gpu_count", gpuCount).field("total_khs", total_khs).field("device_count", 1)

统计设备数量

找了好久也没看到基于flux统计机器数量（基于tag的count统计），转弯在上报数据里，专门加了一个字段device_count=1，表示机器数量

from(bucket: "fxos")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "rig")
  |> filter(fn: (r) => r["_field"] == "device_count")
  |> unique(column:"ip")
  |> drop(columns:["ip"])
  |> aggregateWindow(every: 5m, fn: sum, createEmpty: false)
  |> yield(name: "sum")

统计group

from(bucket: "trial_bucket")
  |> range(start: -15m)
  |> filter(fn: (r) => r._measurement == "http_api_request_duration_seconds")

默认其实就有group，是按‘start’ 和 ‘end’ 来分组的

 进一步

from(bucket: "trial_bucket")
  |> range(start: -15m)
  |> filter(fn: (r) => r._measurement == "http_api_request_duration_seconds")
  |> group(columns: ["_measurement", "_field"])

 

 

第三步

from(bucket: "trial_bucket")
  |> range(start: -15m)
  |> filter(fn: (r) => r._measurement == "http_api_request_duration_seconds")
  |> group(columns: ["_time"])
  |> mean()
  |> group()  

参数 mode

group有个参数mode，2个值

• by: 默认值，根据前面的参数排序
• except: 除了前面的参数，剩下的都参与group

Accordingly, we can modify the above code to get the same result:

from(bucket: "trial_bucket")
  |> range(start: -15m)
  |> filter(fn: (r) => r._measurement == "http_api_request_duration_seconds")
  |> group(columns: ["_time"])
  |> mean()
  |> group(columns: ["_time", "_value"], mode: "except")

参考：

InfluxDB 2 - InfluxQL VS Flux：

Grouping data with Flux in InfluxDB