Features Release

Pandora FMS 3.0 Rc2 is here !

November 10, 2009

Pandora FMS 3.0 Rc2 is here !

Hi everybody !

Pandora FMS v3.0 RC2 has been released. This is a major upgrade from 2.0
version and includes a lot of new features and a surprising performance
boost. This version is a Release-Candidate, this is similar to a beta
version, but with months of testing done. Since RC1, one month ago we have fixed about 50 bugs.

This includes a new installation CD (Software appliance based on
OpenSUSE11), an vmware/VirtualBox image, RPM packages, DEB packages and of course, source tarballs. Some files has not changed since RC1, like the  Windows Installer (for our Windows agent), and the 360 page user manual in PDF.

More information about this release at our website at http://pandorafms.org

To get downloads go directly to:

https://sourceforge.net/projects/pandora/files/Pandora%20FMS%203.0/RC2


3 comments
  1. Hello! Have you try Pandora FMS on BIG installations? About 1000-5000 computers? I need the monitoring solution for about 5000 computers and about 100 chilers, conditions, etc... On every computer I need to get about 15 sensors (cpu, swap, network, temperature, special network events) every 15-20 seconds. I can use several secondary monitoring servers. Is Pandora usable with this configuration, can you appologize?

  2. Slerena

    We're working in a project for a big telco company in Spain. Project is for 6000 computers, with a total of almost 500,000 total alerts.15-30 seconds for all modules is _too much_ for any system in a single server (just make the math opeation for how many modules you will have in a single day:5000 agents x 15 modules x (86400/30) = 216,000,000 items PER DAY. That's impossible to manage in a single database, at least in MySQL, even using a big cluster. Probably in Oracle you can get much more, but using extraordinary hardware and a well-done oracle setup, not very cheap ;-)You need to use long intervals or split monitoring in different domains, with different hardware. Another option is to do not use history on modules (do not store performance data, just get the last value), but even in this case, is a HUGE ammount of data, you will need several systems in the frontend and for use a MySQL Cluster very well configured to manage this load: 5000 XML files each 30 seconds or 75000 network checks each 30 seconds, that makes 2500 "pings per second" if the checks will be a simple PING.There is NO WAY to make 2500 pings per second, even using a MEGA-CLUSTER system. Just think what happen if a ping needs 2-3 secs to finish the check, times will be delayed....You need to split this data or processing in paralel with separate systems and aggregate data to a final system just showing bad data or totals.

  3. I know about all troubles :) I want to use hierarhical monitoring system with aggregations of some (not all) parameters. So, databases can be hierarhical too on several servers. I don't need to store all data forever. E.g. I need to store cpu/swap/net usage with 15-20 secs interval for all nodes for about a week. Then I can aggregate them to 5-10 minutes interval. Temperatures, humidity, etc can be tested every 30-40 secs and can be aggregated to about 5-10 minutes untervals. Etc. All parameters must be tested for alerts values. Some values may be not stored, only tested... Is it possible?

Leave a comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.