inputs.opcua missing support for arrays

[mirkocomparetti-synesis]

Relevant telegraf.conf:

[[inputs.opcua]]
   name = "codesys"
   endpoint = "opc.tcp://192.168.2.247:4840"
   connect_timeout = "10s"
   request_timeout = "5s"
   interval = "1s"
   security_policy = "None"
   security_mode = "None"
   auth_method = "Anonymous"
   nodes = [
     {name="joints", namespace="4", identifier_type="s", identifier="|var|CODESYS.Joints", data_type="float", description="Joint values"},
   ]

System info:

Ubuntu 20.04, Telegraf 1.16.2 (git: release-1.16 962082f), InfluxDB v1.8.3 (git: 1.8 563e6c3d1a7a2790763c6289501095dbec19244e)

Steps to reproduce:

1. Define an array value in OPCUA like in the picture
   
2. import it into telegraf with above configuration

Expected behavior:

Arrays should be imported as separated fields, just like it happens in mqtt consumer when an array is fed to the plugin in json format.

Actual behavior:

Array values are not imported, as shown in picture

Additional info:

As an example, this json

{
	"data_epoch": 1606398721.25,
	"J": {
		"1": 0.8607293824546176,
		"2": 4.381520765315148,
		"3": 0.8909938918704291,
		"4": 3.6710059274584418,
		"5": 4.801455750458236,
		"6": 5.99702671486787,
		"7": 6.237032901194868
	}
}

is correctly translated into multiple fields by the mqtt consumer with this configuration

[[inputs.mqtt_consumer]]
  servers = ["tcp://1.2.3.4:4321"]
  connection_timeout = "15s"
  topics = [
    "some/#",
  ]
  tag_keys = [
    "sometagA",
    "sometagB"
  ]
  tagexclude = [
    "host",
    "topic"
  ]
  persistent_session = true
  client_id = "someid"

  data_format = "json"
  json_name_key = "measurement"
  json_time_key = "data_epoch"
  json_time_format = "unix"

see picture below

Thanks!

[mirkocomparetti-synesis]

This can be potentially achieved using a combination of processors and aggregators, but the configuration for an array becomes very long and not practical. Adding the possibility to handle this automatically would be better.

Thanks

[sjwang90]

Thanks for the issue and info @mirkocomparetti-synesis. I'm going to link this to #8083 and close this issue so we don't have duplicates open. Feel free to add any information to that issue.

[nicolasj92]

I also would very much appreciate this feature! In the meantime I wrote a quick and dirty workaround that stores the array as a string in influx

func arrayToString(a interface{}, delim string) interface{} {
    if reflect.TypeOf(a).Kind() == reflect.Slice {
	    return strings.Trim(strings.Replace(fmt.Sprint(a), " ", delim, -1), "[]")
    } else {
	    return a
    }
}

[werbungvonmir]

Hi! I am trying to import arrays and am wondering if there is a solution available that I have not stumpled across so far. I have not been successfull in importing arrays as of now.

If you have any advice that can help me I would be delighted!

[alexmc1510]

Hello @nicolasj92 , could you explain how you have implemented the function? Inside telegraf?

@mirkocomparetti-synesis , do you have an idea about the processors and aggregators that I should use?