Nagios Conference 2012 - Alexis Le Quoc - Pretty Nagios Charts (alq)

Pretty Nagios Charts

Alexis Lê-Quôc (@alq)http://dtdg.co/nagios2012

Pretty Nagios Charts

Alexis Lê-Quôc (@alq)http://dtdg.co/nagios2012

@alqDev & OpsNagios user since 2008Datadog co-founder

Survey

Compared to Nagios World

2011...

Is your Nagios doing...

(a) Better(b) About the same(c) Worse

Why does it matter?

How do you find out?

MemoryAnalytics

First, the pipeline

Your Nagios logsare a treasure

Comprehensive

Accurate

Written for you

Hard to comprehend

IngredientsRaw logs

Simple parserSQL database

R ggplot2 or d3.js

Query

Extract

Analyze

Present

nagios.log

PostgreSQL

R

ggplot2

Nice graphs

Python

nagios.log

nagios.csv nagios table

Postgres

R Data Frame

ggplot2

PDF

RPostgresql

Let’s play!

5 lines of R

ggplot(dd_by_day_raw, aes(occurrence_doy, daily, color=factor(notifying)))

+ geom_line()

+ xlab("Day of year")+ ylab("Service Alerts")+ ggtitle("Notifying v. silent alerts per day")

Data

Geometry

Chrome

ggplot(dd_by_day_raw, aes(occurrence_doy, daily))

+ geom_line()

+ geom_smooth()

+ xlab("Day of year")+ ylab("Service Alerts")+ ggtitle("Daily trends")

Linear regression

ggplot(dd_by_day_raw, aes(occurrence_doy, daily, color=factor(notifying)))

+ geom_line()

+ geom_smooth()

+ xlab("Day of year")+ ylab("Service Alerts")+ ggtitle("Notifying v. silent alerts per day")

Tuesdays???

ggplot(dd_by_hod, aes(occurrence_dow, daily, group=occurrence_dow))

+ geom_boxplot()

+ scale_x_discrete(breaks=seq(0, 6), labels=c("Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat"))+ xlab("Day of the week")+ ylab("Number of alerts")+ ggtitle("Daily distribution")

Fun with Nagios stats...As a

Service

Performance Metrics

Nagios TrafficOther Sources

Real-time graphs + analytics

Play Stack:•parser•postgres•R•ggplot2

Service Stack:•parser•postgres•Hadoop•d3.js

Quick demo

And more...

Real-Time Graphing

CorrelationMetric-based Alerts

“Share what you see”

http://datadoghq.com

Curious?