Red Hat Slave node cannot connect to master

[grahamjjohnson]

Hi

Linux eurlinux02 2.6.9-22.0.2.ELsmp #1 SMP Thu Jan 5 17:13:01 EST 2006 i686 i686 i386 GNU/Linux

we are trying to set up cluster on 2 linux boxes using GFS 6.1.

When We issue the following we get "failed to logon to master":-

root@eurlinux02 ~]# service lock_gulmd restart
Stopping lock_gulmd: [ OK ]
Starting lock_gulmd: failed to login to master [FAILED]

When we run clustat on Master we get following:-

[root@eurlinux01 ~]# clustat
Member Status: Quorate

Member Name Status
------ ---- ------
eurlinux01 Online, Local, rgmanager
eurlinux02 Offline


Any Ideas would be greatly appreciated!!

Thanks in Advance...

 

[jstreich]

Never used GFS, we used MPI libraries, ssh, rsh, ganglia...
As far as sharing the drives, NFS works, but can slow the network a little... We just moved files using sftp/scp and rsh/rcp when we need to get remote files... Most cluster software knows to dump output data to the control node, and only uses the compute-node's drive for scratch.

This was a 21 node cluster of P4 machines with HT technology. I noticed your using the SMP kernel, our testing showed that taking advantage of the hyper threading actually slowed the simulations down. If your using a dual core machine, you may not have this problem.

http://mirror.centos.org/centos/4/docs/html/rh-gfs-en-6.1/

[plug=shameless]

http://game-master.us/phpx-3.4.0/

[/plug]