Checks with joins

InfluxDB 2
1

I am trying to create a check for load average per cpu where I take the loadaverage5 and divide it by the number of CPU’s.
This is so that I can get a per-cpu load average metric to alert on.

I have got the following filux query working in the data explorer

loadavg = from(bucket: "metrics_pg")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "system")
  |> filter(fn: (r) => r["_field"] == "load15")
cpucount = from(bucket: "metrics_pg")
  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)
  |> filter(fn: (r) => r["_measurement"] == "system")
  |> filter(fn: (r) => r["_field"] == "n_cpus")
join(
  tables: {load:loadavg, n_cpus:cpucount},
  on: ["_time", "_stop", "_start", "host"]
)
  |> map(fn: (r) => ({
      _time: r._time,
      _value:  r._value_load / float(v: + r._value_n_cpus),
      host: r.host
    })
  )
  |> aggregateWindow(every: 5m, fn: last)
  |> yield(name: "last")

But when I try to create a threshold check using the HTTP API I get the following error:

$ curl --header "Authorization: Token ${TOKEN}" 127.0.0.1:9999/api/v2/checks -X POST -d @pg_cpu.json  | jq
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100  1768  100   113  100  1655   8692   124k --:--:-- --:--:-- --:--:--  143k
{
  "code": "invalid",
  "message": "Could not create task from check: expected a single field but got: [load15 n_cpus]"
}

From the output it seems to me that the check is using the output from the first two queries where I fetch the loadavg and cpucount.

Here is the query section of the json I use to create the check

    "query": {
        "text": "loadavg = from(bucket: \"metrics_pg\")\n  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)\n  |> filter(fn: (r) => r[\"_measurement\"] == \"system\")\n  |> filter(fn: (r) => r[\"_field\"] == \"load15\")\n\ncpucount = from(bucket: \"metrics_pg\")\n  |> range(start: v.timeRangeStart, stop: v.timeRangeStop)\n  |> filter(fn: (r) => r[\"_measurement\"] == \"system\")\n  |> filter(fn: (r) => r[\"_field\"] == \"n_cpus\")\n\njoin(\n  tables: {load:loadavg, n_cpus:cpucount},\n  on: [\"_time\", \"_stop\", \"_start\", \"host\"]\n)\n  |> map(fn: (r) => ({\n      _time: r._time,\n      _value:  r._value_load / float(v: + r._value_n_cpus),\n      host: r.host\n    })\n  )\n  |> aggregateWindow(every: 5m, fn: last)\n  |> yield(name: \"last\")\n",
        "editMode": "advanced",
        "name": ""
    }

Any help will be really appreciated.