gphat / censorinus Goto Github PK

Hey @gphat, i've seen that scala 2.13 support is merged in the master branch. Is it possible to publish it ?

The metrics 2.0 spec is interesting. It seems that unit is the only addition that would be required, aside from an encoder.

DogStatsD's sample rate uses a pipe in front. StatsD doesn't.

The DogStatsD encoder test should be modified to match the StatsDEncoder test.

I would like to add tags to my counters when sending data to telegraf.
Similar to this:

> time                code host     message metric_type value
> ----                ---- ----     ------- ----------- -----
> 1556787040000000000 200  server-3 success counter     1
> 1556787120000000000 200  server-3 success counter     5
> 

Here, code and message being the custom tag, I did through python client.
Same can be done using this library?

We tried to use your library, but there is no output, even when using the client this way:

    new StatsDClient(hostname = "localhost",
                     port = 8125,
                     asynchronous = false,
                     allowExceptions = true)

We can see data which we pushed via nc like'

echo "foo.bar 42 `date +%s`" | nc -c 127.0.0.1 2003

Is there a way to enable debugging?

We are running this docker image: https://github.com/kamon-io/docker-grafana-graphite
Tried also another image, which is not working as well: https://github.com/hopsoft/docker-graphite-statsd

The DogStatsDClient is not not using prefix parameter for events names whereas the official statsd client from datadog does.

I think this library should have a behaviour as close as possible to the datadog official library.

It's common for StatsD clients to allow setting a prefix for any metrics, saving some typing. Add it.

This will make it easier to pass in fake/mock clients and test dogstatsdclient. Specifically, I'm currently having problems hitting socket limits since I can't stop my mocked DogStatsDClient from creating a UDPSender in another class that uses DogStatsDClient (using ScalaMock).

From @jpellerin:

From scala @com_github_gphat_censorinus//:censorinus:
external/com_github_gphat_censorinus/src/main/scala/github/gphat/censorinus/dogstatsd/Encoder.scala:71: warning: Using toString on something that is already of type String.
      sb.append(h.toString)
                  ^
external/com_github_gphat_censorinus/src/main/scala/github/gphat/censorinus/dogstatsd/Encoder.scala:75: warning: Using toString on something that is already of type String.
      sb.append(ak.toString)
                   ^
external/com_github_gphat_censorinus/src/main/scala/github/gphat/censorinus/dogstatsd/Encoder.scala:128: warning: Using toString on something that is already of type String.
      sb.append(h.toString)
                  ^
three warnings found

These aren't client agnostic, but I'm pretty sure I left room for client implementations to have different features.

Telegraf logs show lines like these:

2021-04-07T08:21:11Z E! [inputs.statsd] Parsing value to float64, unable to parse metric: xxx.yyy.zzz:2,2|g

Note that I have LANG=fi_FI.UTF-8 which should have comma (,) as the decimal separator. (This is of course only a problem when testing on my own workstation.)

The metric is sent like this client.gauge("yyy.zzz", doubleValue, tags)

It seems that in github.gphat.censorinus.dogstatsd.Encoder the value format is initialized as:

val format = new DecimalFormat("0.################")

I think it should be initialized with ROOT locale:

val format = new DecimalFormat("0.################", new DecimalFormatSymbols(Locale.ROOT))