What stats to monitor on a production go server?

[Paul Dix]

Hey all,

I wrote a blog post about some stuff we're doing to monitor our production Go processes (https://errplane.com/blog/monitoring-a-production-golang-server-with-memstats). We're mostly just using runtime.MemStats, but I'm curious... what other metrics are people tracking in their production systems?

Thanks,

Paul

[Sébastien Ménard]

The 3 general metrics are CPU, memory and web traffic use. You can even monitor I/O uses (to switch to SSD if needed later, which is nice for big databases).

[Mike Arpaia]

To throw my hand in here, I work at Etsy and we monitor pretty much everything that can be monitored. Here's a video of our SVP of Ops giving a walkthrough of some of our in-office dashboards and explaining our metrics architecture (spoiler: ganglia and nagios): http://vimeo.com/25807922

We also have a pretty neat blog post about monitoring anything and everything within your application using StatsD and Graphite: http://codeascraft.com/2011/02/15/measure-anything-measure-everything/

The concepts of monitoring and metrics are mostly language independent so I hope this is of use.

--
You received this message because you are subscribed to the Google Groups "golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.