Joe’s Blog!

Something I happened to see over here this weekend was Disco. It is a Map/Reduce framework written in Erlang. A user/implementer doesn’t need to know a lick of Erlang to get rolling but according to their site most folks use Python to write the actual jobs. If you as me a Map/Reduce framework built using Erlang makes a great amount of sense due to its message passing and light weight processes.

I have gotten a few comments and etc regarding whether or not I was able to get Hadoop to talk to Ganglia. Sadly I wasn’t able to get this to work properly either but I did contact the Hadoop mailing list (this thread) and got the following information. There is actually a bug. The link includes a patch but note that the trunk has changed and the patch currently only works on Hadoop version 0.16.0. I have not had a chance to test everything out yet but it is at least a step in the right direction for those of you who are curious. Hope this helps.

Bernard Li pushed a new version of gexec out in response to my inquiries on the mailing list, it includes the Ganglia switch.

I did some further code changes and was able to generate a tarball
which builds fine without any modification, can you please try it on
your system and see if it works? All you need to is run `rpmbuild
-tb` against the tarball:

http://therealms.org/oss/ganglia/gexec-0.3.8.1375.tar.gz

Check out this sweet machine that the University of Antwerp built.

Install Ganglia

wget http://voxel.dl.sourceforge.net/sourceforge/ganglia/ganglia-3.0.7-1.src.rpm
rpm -Uhv http://apt.sw.be/redhat/el5/en/i386/rpmforge/RPMS/rpmforge-release-0.3.6-1.el5.rf.i386.rpm
yum install libpng-devel libart_lgpl-devel rrdtool-devel freetype-devel rrdtool-devel
rpmbuild –rebuild ganglia-3.0.7-1.src.rpm
rpm -ivh /usr/src/redhat/RPMS/x86_64/ganglia-gmetad-3.0.7-1.x86_64.rpm /usr/src/redhat/RPMS/x86_64/ganglia-gmond-3.0.7-1.x86_64.rpm /usr/src/redhat/RPMS/x86_64/ganglia-devel-3.0.7-1.x86_64.rpm

Install libe

wget http://www.theether.org/libe/libe-0.3.0-1.src.rpm
rpmbuild –rebuild libe-0.3.0-1.src.rpm
rpm -ivh /usr/src/redhat/RPMS/x86_64/libe-0.3.0-1.x86_64.rpm

Install authd

yum install openssl-devel
wget http://www.theether.org/authd/authd-0.2.2-1.src.rpm
rpmbuild –rebuild authd-0.2.2-1.src.rpm

You will run into an error like the following, don’t worry about it we clean it up next.

Installing authd-0.2.2-1.src.rpm
warning: user bnc does not exist – using root
warning: group dusers does not exist – using root
error: Legacy syntax is unsupported: copyright
error: line 5: Unknown tag: Copyright: GPL

Finish up authd

mv /usr/src/redhat/SPECS/authd.spec /usr/src/redhat/SPECS/authd.spec.1
sed ’s/Copyright/License/g’ /usr/src/redhat/SPECS/authd.spec.1 > /usr/src/redhat/SPECS/authd.spec
rpmbuild -ba /usr/src/redhat/SPECS/authd.spec
openssl genrsa -out auth_priv.pem
chmod 600 auth_priv.pem
openssl rsa -in auth_priv.pem -pubout -out auth_pub.pem

Copy auth_priv.pem and auth_pub.pem to ‘/etc’ on each node of the cluster

rpm -ivh /usr/src/redhat/RPMS/x86_64/authd-0.2.2-1.x86_64.rpm

Installing gexec (using my SRPM, includes the ‘–with-ganglia’ option)

echo “gexec 2875/tcp # Caltech GEXEC” >> /etc/services
yum install glibc gcc gcc-c++ authd expat-devel
rpm -ivh /usr/src/redhat/RPMS/x86_64/gexec-0.3.8-4.x86_64.rpm