News‎ > ‎

BonFIRE Release 4.0.5

posted 6 Nov 2013, 01:16 by Vegard Engen
We are happy to announce that BonFIRE has been upgraded to version 4.0.5, which is a significant upgrade to our services and infrastructure.

Trial accounts

In addition to our already streamlined Open Access application process, BonFIRE now supports a much simpler, even faster process to get access to BonFIRE within a working day. Please tell your friends and colleagues about it:

New site, Wellness Telecom

Wellness Telecom is now on line. This new site offers a taste of production Clouds through the BonFIRE interfaces and expands BonFIRE's portfolio to include VMWare-based systems. Please read the user documentation page to familiarise yourself with the Wellness Telecom capabilities:

More capacity

HLRS has more than doubled its capacity to 344 cores. EPCC has also all but doubled its capacity to 176 cores. Combined with Inria's recent upgrade to 96 cores, BonFIRE is demonstrating its commitment to a continued service, past the end of the current project funding.

For more information please see the infrastructure documentation at


Inria and iMinds (be-ibbt) include the new COCOMA image, which allows you to perform controlled experiments of multi-tenant scenarios (effects of VMs co-located on the same physical machine). With COCOMA, you can generate COntrolled, COntentious and MAlicious pattern loads on the physical nodes where it is deployed, affecting resource such as CPU, RAM, disk and network. This allows you to study in a controlled and repeatable manner how this affects your own application.

Please read the COCOMA documentation for more information:

Scalable APIs

A new component caches resource collections behind the scenes, so as to accelerate such queries. This improves performance of your query. Also, because these collections use data from asynchronous site events rather than queries to the sites, site performance is also improved.

More experiment events

Starting from Release 4.0.5, the Experiment Message Queue includes events about user resources, which are generated by the sites. These provide more detailed, more accurate information about the resources than the previous (and still supported) events coming from the Resource Manager. They come in a new, advanced format, described in our user documentation:

ECO-metrics and Zabbix improvements

Courtesy of BonFIRE-based STREP ECO2Clouds, the sites at HLRS, Inria and EPCC now expose a new set of infrastructure metrics about energy consumption and the CO2 profile of this energy.

These, together with the new events, are available through Zabbix.

Numerous improvements have also taken place to improve the stability of Zabbix in our images, including upgrade to version 1.8.16 of Zabbix, the latest version that supports the same interface as currently exposed by BonFIRE.

Network functionality

# BonFIRE now supports VPN access to its resources. Instructions are available from our user documentation:

# We made it easier to use Unix tools for network impairment on all infrastructures. Just follow the following simple instructions:

# The Virtual Wall also allows to select the queue strategy. More

# EPCC now features an increased number of public IP addresses, along with removal of most of the restrictive firewall rules.


The Portal has been revamped to better present user and group experiments and give users control over refresh of the output. It also exploits the new caching mechanism for faster resource listings and better support of large experiments.

Experiment Manager

The new "aggregator" keyword simplifies initiating monitoring on managed

Also, the Experiment Descriptor now also supports the creation of persistent storage outside an experiment:

Known issues

We will be adding fixes and announcing updates to the following issues:

  - A recent upgrade of the network outside BonFIRE's control has affected the performance of the AutoBAHN service. We are working to address this issue with the UK NREN.

  - HP is not yet available due to delays in upgrading the systems.

  - The accounting service interface was found to be performing sub-optimally and is being worked on. Data are still being collected.