Performance, Exceptions and Events - Monitoring and Analytics

At HackScience we are looking into options for monitoring our deployed devices. Currently we are considering using some or all of the following Datadog, Sentry and Keen.io. We are fairly certain we will be making use of both Sentry and Keen.io.

However for Datadog to work it is claimed that it will not work without an agent installation. I think this essentially means that Datadog would need to install their own container or code on the underlying operating system which is managed by resin.

  • Does anyone have any thoughts or advise on device and performance monitoring in resin?
  • Could there be any way to get Datadog working with resin and if it is possible is Datadog overkill for IOT?
  • What alternatives are there to monitoring the performance of a resin device?

Any help or advise is much appreciated.

Hey @hpgmiskin,

I’ve played around with prometheus before however this will give you container stats and not the host stats it sounds like you are after. To get host stats, we’ll need special mounts as mentioned here: https://github.com/prometheus/node_exporter/pull/376/files.

@pcarranzav will need to confirm, but once we have multi-container support (coming in a few months) you’ll run the containers with a docker-compose file, which will allow you to run (datadog/prometheus) as a container and make the appropriate mounts to the host and docker socket.

Hi @craig,

Thanks for your post. I had not looked into the Prometheus before but it looks to be an excellent solution.

For anyone else reading this both Craig’s blog posts are excellent for setting up performance monitoring.

I cannot wait for multi container support, currently we have a monolith of a dockerfile which makes use of systemd services and takes an absolute age to load.

@hpgmiskin, you and me both, once multi-container support lands we can really take advantage of the docker/resin community.