MSteggink.com

Welcome to my personal website. At this website, you can view some activities I am involved at, like my hobbies and work.

• Matthew: About me, my work, short resume
• My photoalbums, made in several countries.
• My portfolio, where I store samples of my work such as created websites, webapplications, written papers and articles, code examples.
• Lambdacomplex or lc is my private small scale data center.
• I also created some Tutorials and code snippets for beginning coders and administrators.
• Downloads: some handy dandy applications that might be usefull for daily usage or administrators.
• Links: Some interesting links I picked up while surfing the net.

Some other sites I run are: http://dev.msteggink.com (development purposes)

I have finished my cluster setup with failover on the masters and the meta data servers. Failover is a pretty hard nut to crack as it introduces several implications: what to do when failover takes place, how to let it detect a failing node correctly etc. I have solved most of this by not using Stonith on the masters, as a faulty action would result in both masters being down :(.

Finally a new update. I am still researching Lustre with failover metadata (MDT's). The storage (OSS) will be two 12 TB redundant RAID6 storage machines. The setup seems simple, but because both Lustre and the Open Fabrics stack have to be installed on it, that makes it quite complex.

For my work I am looking into Lustre, and a tutorial will be published here as well.

Also I have upgraded Vortessence from a dual-core E6750 to a quad core 9550 which will boost performance on virtualization.

I got a cluster up and running with one dualcore compute node. I will be using this to develop and test MPI test applications. Probably not very useful programs, but they are for educational use . I might add another compute node in the future.

The cluster is equiped with SGE as the queueing system. Of course the cluster is not really done yet, and additional software will needed to be installed.

No real updates, I am currently working on installing a virtual cluster for research purposes on LC/Vortessence. I am switching between Rocks or OSCAR. The installation is documented in Tutorials.

I have recently upgraded my server with a 8x 1TB disk RAID6 array and two dual GBit Intel MT 1000. I am currently researching high performance ethernet.

I will also run a cluster based on OSCAR (Open Source Cluster Application Resources) with 4 virtual nodes just to test functionality.

This week I will be preparing a massive 230 node (1840 cores) cluster in Germany. The house in Hengelo has almost been cleaned now.

I've finished installing a 672 core cluster at CERN. It is my biggest cluster up to now. The computer room was gigantic. I will add the photoalbum as soon as possible.