Nagios World Conference 2013 was held between Sep 30th and Oct 3rd at St. Paul, MN. I represented Mozilla IT/SRE along with Sheeri Cabral, who spoke about MySQL plugins. I wanted to share some observations and my best takeaways from the conference. I attended about 10 talks in all and spent more time discussing setups and best practices.
The biggest draw at the conf this year was Nagios 4.0 that was announced at last year’s keynote. 4.0 brings in some long awaited and much needed rocket power to Nagios. The changelog has detailed information about the big features but the ones that interested me the most were:
- Core Workers – I have been researching on how to scale up service check execution on some of our bigger instances. mod-gearman has till now been the tool of choice but with Core Workers, Nagios natively steps up to the task. The legacy forking-for-each-check model was unsurprisingly hitting limits in some places and 4.x replaces it with worker processes that get check execution delegated to them. There is a massive performance gain and I’m looking to leverage that vs. integrating with mod-gearman.
- Query Handlers – This feels like baked-in MK Livestatus. It’s made available via a socket. Unlike livestatus, it doesn’t yet have a lot of fancy and it’s mostly basic at the moment. I’d expect it would get a lot of attention in future versions.
Among other things I’m looking forward to integrating Multisite in our infrastructure. We have close to a dozen Nagios instances here at Mozilla and our primary interface to each is via IRC bots. As one would imagine, it doesn’t scale well and isn’t ideal for dealing with mass changes. This is where Multisite comes in very handy. Along with Livestatus, Multisite provides for a supercharged way to deal with multiple instances and multiple service/hosts within each. Do try out the demo because it’s hard to put awesome in words
Some nice talks that stood out:
- Effective monitoring by Rodrigue Chakode where he spoke about filtering false alerts and actionable alerts and using business processes to monitor the most effective elements of a system.
- Nagios at Etsy by Avleen Vig, who had an eventful road trip to the conference and discussed some cool things Etsy has done, particularly measuring alert fatigue by correlating alerts and sleep inputs from fitbit worn by oncalls. He also spoke at length about “Monitoring hygiene” and how Etsy went from 300 alerts/day to 45 alerts over the course of two years.
In all, it was a great conference, like last year. Looking forward to a year of 4.x and trying to get the in-house puppet module out on github