Add tcp/udp check connection input plugin

[titilambert]

Rebased PR of #476

[titilambert]

@sparrc Thanks ! I fixed the sample config ;)

[sparrc]

could you write a README in the format of https://github.com/influxdata/telegraf/blob/master/plugins/inputs/EXAMPLE_README.md?

[sparrc]

since we can specify multiple [[input.connection]] entries, can you change config to look more like this:

[[inputs.connection]]
  # must be either "tcp" or "udp"
  type = "tcp"
  # Server address (default IP localhost)
  address = "github.com:80"
  # Set timeout (default 1.0)
  timeout = 1.0
  # Set read timeout (default 1.0)
  read_timeout = 1.0
  # String sent to the server
  send = "ssh"
  # Expected string in answer
  expect = "ssh"

[titilambert]

@sparrc Done !

[gianarb]

Awesome!! A good way to decrease the latency.. 👍 Thanks

[sparrc]

This looks good @titilambert, the only problem I have is that I feel like the name connection is a bit vague.

But, I'm also having trouble coming up with a good name, maybe just response_time would be a better name? or net_response?

[titilambert]

@sparrc I had the same feeling when I choose the name ...
Nagios equivalent plugins are check_tcp and check_udp...
I think I like net_response, but I would suggest some others: net_connection, tcpudp_connection, port_check, tcpudp_check.
What do you think ?

[sparrc]

I like net_response or net_connection, with a slight preference for net_response since it describes the data that you get out of the plugin (response time)

I'm leaning away from tcpudp because at some point I think we could add http response times to this plugin as well.

[titilambert]

@sparrc let's go for net_response !
I'm currently doing a input plugin for kubernetes, once finished, I will make this change.

[sparrc]

OK, but someone has already tried to do that: #665

I rejected the PR because it seems that the kubernetes client library is very tied up with their entire repo, and adding it requires adding 10-12 huge dependencies to Telegraf that almost doubled the binary size.

Although there's also the option of requiring kubernetes-cli or some external executable as a dependency of the plugin

[titilambert]

@sparrc I do not use kubernetes client lib :)
I use "/metrics" endpoints which show data with prometheus format :) So I just use prometheus client lib.
I'm currently adding a new parser in plugins/parsers folder.
And for other metrics, I would recommend https://github.com/kubernetes/heapster which can send data to influxDB :)

[titilambert]

@sparrc changes done !

[sparrc]

great, thanks for another contribution @titilambert

[sparrc]

@titilambert I sent you an email for some swag ~2 weeks ago, did you receive it?

[titilambert]

@sparrc no I didn't anything. Could you PM me on twitter ?

[sparrc]

@titilambert sure, will do