Sending influx metrics over UDP

[mroysterhead]

The environment I am deploying to has an existing influx stack with a requirement to use UDP. How can I configure harvest to send over UDP?

I would very much like to be able to do this - The only alternative method I can think of is adding a telegraf service alongside and using a UDP plugin - but I would like to avoid this if we can stick with just harvest.

[cgrinds]

hi @mroysterhead not sure how widely used metrics over UDP would be if we added it to Harvest. I thought Influx dropped support for UDP in v2 or did they change their mind? Ah found it.

Since Harvest only supports Influx v2, I think Telegraf is probably your best bet.

[mroysterhead]

That's what I ended up doing. For posterity, here's the telegraf config to basically behave as a translation proxy from TCP to UDP:

# Telegraf Configuration
#
# Telegraf is entirely plugin driven. All metrics are gathered from the
# declared inputs, and sent to the declared outputs.

# Global tags can be specified here in key="value" format.
[global_tags]
  dc = "iad1"

# Configuration for telegraf agent
[agent]
  interval = "0.5s"
  round_interval = true
  metric_batch_size = 1800
  metric_buffer_limit = 1200000
  collection_jitter = "0s"
  flush_interval = "1.0s"
  flush_jitter = "0.1s"
  precision = ""
  debug = false
  quiet = false
  logfile = ""
  hostname = ""
  omit_hostname = true
###############################################################################
#                            OUTPUT PLUGINS                                   #
###############################################################################

# Configuration for influxdb server to send metrics to
[[outputs.influxdb]]
  urls = ["udp://<INFLUX-ENDPOINT>:8094"]
  database = "telegraf"
  retention_policy = "null"
  write_consistency = "any"
  timeout = "5s"
  user_agent = "telegraf" # used for HTTP only
  udp_payload = 1024
  tagexclude = ["junk"]
  namedrop = []

###############################################################################
#                            INPUT PLUGINS                                    #
###############################################################################


# Generic socket listener capable of handling multiple socket types.
[[inputs.socket_listener]]
service_address = "tcp4://:8096"
max_connections = 1024
data_format = "influx"
tagexclude = []
namedrop = []
# These default to the OS default
# read_buffer_size = 65535
# keep_alive_period = "5m"


#Collect statistics about itself
[[inputs.internal]]
  #If true, collect telegraf memory stats.
  collect_memstats = false
  namepass = ["internal_write"]
    [inputs.internal.tags]
    host="<HOSTNAME>"

And you can validate this with:

# confirm port is listening
    $ netstat -lntu | grep 8096
tcp        0      0 0.0.0.0:8096            0.0.0.0:*               LISTEN
# test writing
    $ echo "mroysterhead_test,pipeline=cool_test value=5.0" | nc -w1 <TELEGRAF-RECEIVER> 8096
# confirm the data was written to influxdb
SELECT distinct("value") AS "test" FROM "<database>"."<retention_policy>"."mroysterhead_test" WHERE time > now() - 1h GROUP BY time(1m) FILL(null)

[cgrinds]

Awesome! Appreciate you sharing your config for the next person