Fladi.at

Some pictures of me taking a flight in a gyrocopter near Weiz/Austria.

Wie der Standard und das Science Blog berichten, wurde für 2010 die “Lange nach der Forschung” in Österreich abgesagt. Frau Mag. Dr. Beatrix Karl führt dafür eine nicht näher spezifizierte Evaluierung an.

In gemeinschaftlicher Arbeit möchte ich nun zusammen mit anderen Personen aus meinem Bildungsumfeld eine Veröffentlichung der Ergebnisse dieser Evaluierung erreichen. Eine Email ist zur Zeit in Arbeit und wird morgen Mittag mit mehreren Unterzeichnern an Fr. Karl verschickt.

Because of this: $90 for a simple Plug-in? You must be kidding!

At least this is the final prove we all needed to realize that the bean-counters have taken over and the techies are leaving. Reminds me of a sinking boat. At least we at the MUG dropped plans for an Oracle cluster in favor of a PostgreSQL cluster.

The performance comparison between Oracle and PostgreSQL was absolutely dominated by the second one. A paper on this topic (only in German by now) will be published later this year.

Today I’ve released my templates which are used by myself at the Medical University of Graz to gather usage data of our Novell eDirectory deployment over SNMP.

SNMP access to internal statistics of eDirectory makes it possible to generate pretty graphs out of it and therefor I built a simple integration for Cacti. In fact it’s just a bunch of XML files for cacti to make it aware of the OIDs necessary to address the appropriate values through SNMP plus it includes ready-to-use graph templates for all available eDirectory usage-data.

Instructions are provided inside the archive but installation and usage should be relatively self-explaining.

Novell eDirectory itself provides a service agent which integrates with net-snmp, thus my instructions are only valid for servers running GNU/Linux or any other UNIX derivates.

Download edirectory-0.1.tar.gz

Here’s a sample screenshot showing the operations statistics for a relatively new server:

eDirectory Operations

I’ve finished my setup of a two-node cluster running SLES 11 HAE for a campus system. It provides a loadbalancer with shared Apache Jackrabbit cache for a JBoss cluster running a proprietary application and Jackrabbit.

Packmaker proved to be the way to go in modern cluster deployments as it provides a solid base for heartbeat with it’s OCF framework and it integrates OCFS2/O2CB to a full degree.