Trace Logger

Distributed logging of embedded traces through docker, fluentd and elastic search

What is this?

Trace logger is a simple utility to flash nordic devices with a target hex file, timestamp with the host time the application output and redirect it to fluentd and elastic search.

This utility features a simple device discovery which is used to mass flash devices and attach a logging container to them.

Requirements

• Docker engine
• Docker compose
• x86 or ARM v7 platform (Raspberry Pi)
• Linux host

Usage

Let's say you have multiple devices plugged into a computer, which you wish to flash with the latest version of your app. In addition to that, you would also like to capture RTT traces straighout of your devices and into a remote database.

To achieve that, execute the launch inside the client folder with the following options:

./launcher.sh --hex $(pwd)/target.hex --app app-name --flash --log

This will start a local logging container which you can view with

docker logs LOG-xxx

Note! It also starts an additional container which establishes the connection to the device and allows capturing RTT traces.

You can also learn more about the launcher with

./launcher.sh --help

Sending traces to Fluentd and visualising them in Kibana

Copy the contents of services into the remote machine of your choice and fire them up with docker-compose. Once they are up and running, run the launcher with:

./launcher.sh --app app-name --log --fluentd --fluentd=myhost:24224

Note! If you don't provide a fluentd host, it will default to localhost:24224.

Now, the configuration provided here will push the data coming in from these containers into elastic search. You are free to route this information elsewhere and please feel free to contribute it.

The routing and matching of data at fluentd is configurable in conf/fluent.conf. In there you will see that fluentd follows a matching rule based on the tags defined at each container. By default trace logger tags all containers with docker.device_id. This makes it easier to identify the devices later on.

In fluentd's configuration you have to chose how to tag the output stream. In this example we use dockerd as an index key. This will be asked from you when you fire up kibana at:

http://localhost:5601/

After you have set the index, data will be indexed and presented to you. Be patient, it can take awhile.

Enjoy!

Contributing

Feel free to send me your pull requests or get in touch with me at @pedrofigssilva