Mon blog-notes à moi que j'ai

Blog personnel d'un sysadmin, tendance hacker

Paris Zabbix user group, first meetup

Around 50 people attended the Paris Zabbix user group first meetup on the 23th of june 2016 at BlaBlaCar’s office in Paris. Alexei Vladishev, creator of Zabbix was present.

Paris Zabbix user group, first meetup - Crédits

Paris Zabbix user group, first meetup - Crédits @parismonitoring

After the usual introductions by myself on behalf of BlaBlaCar, sponsor of the event, and Steve for the Paris Zabbix user group, Alexei Vladishev himself gave us an introduction to Zabbix.

Introduction to Zabbix by Alexei Vladishev - Crédits

Introduction to Zabbix by @avladishev - Crédits @parismonitoring

Alexei first recalled that Zabbix favors a full open source approach, the software is released under the GPLv2 license, rather than opening a core feature set and providing additional ones, often among the most important, in a pay and closed version (known as an “open-core”). We obviously fail to draw a parallel with Nagios.

Zabbix is open source, not open core - Crédits

Zabbix is open source, not “open core” - Crédits @parismonitoring

Alexei then talked about Zabbix architecture with its components: the server itself (zabbix-server) but also proxies (zabbix-proxy) and finally agents (zabbix-agent). He insisted that both proxies and agents can operate in passive or active mode depending on the needs and constraints imposed by the network.
These features make Zabbix a comprehensive and flexible monitoring tool whose deployment is also simplified by the particularly small size of its main components:

  • zabbix_agent: 339KB
  • zabbix_proxy: 1.3M
  • zabbix_server: 1.4M

After sweeping recent additions Zabbix 3.0, among which encryption communications using TLS but also the prediction based on the evolution of a metric which is very useful when monitoring disk space for example1, Alexei presented some of the features planned for Zabbix 3.2, scheduled for August 2016.
We should get some event correlation as well as tags. Combining these two features should help to reduce the number of alerts in the event of a major incident in a more flexible way than the current system of triggers dependencies, to show only alerts reflecting the root cause of the incident.

How to avoid flapping - Crédits

How to avoid flapping - Crédits @ghileskheddache

Right after Alexei presentation and a QA session, I made a short come back about Zabbix usage at BlaBlacar.

Zabbix at BlaBlaCar by Jean Baptiste Favre - Crédits

Zabbix at BlaBlaCar by @jbfavre - Credits @parismonitoring

Talk was shorter than during Zabbix Conference 2015, 5 minutes only.
I did not dive deep into Zabbix technical details because part of the attendees weren’t familiar with Zabbix. Instead, I made some recommandations about monitoring: architecture, segmentation, standardization of a monitoring infrastructure, and explained how we applied them at BlaBlaCar with Zabbix.

The evening ended up with traditional discussions with non less traditional pizza. Of course, most of the discussion were about Zabbix.

This first meetup really was a success, only troubled by Wifi automatic shutdown2.

Share more, learn more about Zabbix - Crédits

Share more, learn more about Zabbix - Crédits @SteveDESTIVELLE

Can’t wait for the next Zabbix Conference in Riga to discover more precisely the new Zabbix 3.2 features !

See you there !

  1. problem is no more to alert when disk space runs below a given threshold, but more to know when it’ll be full so that alert gives enough time to remediate the situation. 

  2. once more sorry for that folks, I swear it won’t happen anymore :)