This year I visited Zabbix conference in Riga. It was great event in nice city. You can find all the presentations (including videos) on zabbix website.
Here are my notes:
All +
All –
- + –
Welcome- attendees from 27 country
- zabbix doing well in Japan (all silver partners from there)
- + –
how zabbix make living?- 50% income from profesional services
- 10% trainings
- 15% commercial implementation
- + –
this year- + –
2.0 released- JMX monitoring
- polished fron-end
- q: upgrade?
- no new features
- + –
development desk- specifications link
- + –
- + –
roadmap 2.2- fixed 12 months release cycle
- trunk stable
- beta 2.2 March 1st, 2013
- 2.2 release May 1st, 2013
- + –
new features- remote command execution
- + –
databse monitoring- experimental now
- alerts on unknown states
- rework of API
- + –
reporting- top N used
- vmware virtualization
- + –
drop current note based distributed monitoring- new generation of distributed monitoring
- + –
zabbix next- vote for features
- how is enterprise prioritiezed?
- + –
questions- + –
upgrade to 2.0- load history after server is running
- + –
- + –
performing RHEL- + –
little’s law- lengh of queue = wait time + number of custommers walking in
- + –
tuning info comming from:- /proc
- /sys
- + –
system tap (RH Dtrace)- compile module on one machine and load it on other machine (only same kernel version)
- + –
use case- tcp restransmissions (NFS restransmissions)
- read/write
- can we use zabbix template for debuging performance?
- noah scheduler for virtuals/ssd/raid
- + –
- + –
Hybrid Cloud Monitoring- + –
presentor- developed Chrome Extension for checking Zabbix Triger (Chromix)
- android app for zabbix
- + –
zabbix extension- host registration done by Zabbix
- Machine Setup Details kept trackable
- Operate from zabbix dashboard
- + –
libcloud.apache.org- meta cloud api
- billing data graph
- connect to ssh console directly from dashboard
- + –
- + –
Zabbix development process- to improve commits – hammer method – pre-commit hook
- weekly digest to rate comments in git – enzyme
- + –
quality of main dev branch- separate dev branches – only merge when it is done
- before merge – code review
- automated tests
- + –
should developer write docs?- yes, initial documentation
- + –
CESNET -Macura- + –
monioring- either we don’t have users
- or we can predict crashes
- + –
nav visualizer for l2 networks- based on mac (using SNMP)
- + –
- + –
Investigating Production Problems- syslog view
- zapcat java proxy
- + –
usual intestigation- feed with lot of data
- play with graphs and correlate graphs
- + –
from 0 to bussines metrics- developers tend to monitor everything and frequently
- pick just key metrics
- what is eact mechanism you collect data?
- online graph of $
- google graphs on top of zabbix – more capacity reporting
- + –
HA Zabbix- + –
problem- if monitoring doesnt work, like flighing blind
- + –
why not cluster- split brain
- no passive checks
- distributed
- + –
- + –
Miracle Linux- selling zabbix appliances
- + –
building maps- topology helps zabbix to prevent false positives
- + –
python script puts together topology using SNMP- non cisco added manualy
- graphviz to visualize data (alternative netviz)
- + –
single zabbix server for more custommers- + –
zpa.cloudz.pro- create iso
- + –
raspberry pi appliance for 50EUR- sanding data from trucks
- + –
- + –
Java EEE- + –
performance of mysql- mysql log size
- partitioning
- doing same thing as we do – modify configuration using API
- + –
autoregistration- java start request and automaticaly register in proxy
- + –
deregister- cron job to query unregistred instances and remove them
- + –
- + –
zabcon- + –
ruby zabbix api- use active record
- command line interface based on ruby api
- + –
lua script- way to extend functionality
- predictive threashold
- separate thread for each script
- + –
- + –
Zabbix API- improve documentation
- improved interface
- + –
strict validation- no incoplete parameters
- error reporting
- + –
scale- up to 1milion items
- + –
new api- + –
maps/graphs- generate image
- + –
- future concepts
- + –
mobile- moZBX
- based on twitter UI
- + –
challanges- too many devices
- limited resources
- + –
frameworks- + –
jqTouch- ajax – cool unless you have to debug it
- + –
twitter bootstrap- simple html and they make it pretty
- just mobile site
- + –
- + –
php api- generate api from frontend (using builder)
- then php interface
- + –
redhat- + –
products- jboss
- rhel
- cloud products
- nas storage
- + –
- + –
Service Health Monitoring- RingCentral – cloud based PBX
- + –
hight level monitoring- matrix of heath
- + –
analyze each service – on what does it depends on?- then triggers for each functionality
- + –
functional checks- host leve
- + –
pool level- agregated on the nasis of pool data
- + –
dashboard- map with background image
- 3 level deep structure