Many array, 2 disks on RAID 1, and rest on RAID 6. The full system is physically separated from the user data. One can expand user data partition without touching the operating system. Also it is easy to keep a spare system disk and save it off-site.

For servers with limited number of slots, one can mix RAID 1 (/boot) with RAID 6 (other partitions). This gives maximum disk redudency protection with available disk slots.

What RAID for swap?

This error occurs when I run eog and gedit from remote ssh to a Ubuntu server. The problem doesn't exist if I switch to root.

A "just works" solution is to run "rm -rf  .dbus/session-bus/*"

http://ubuntuforums.org/archive/index.php/t-1169825.html

Background

MPICH2 on the cluster runs a global mpd as root. The initial tests include bash scripts, cpi, some user made programs all passed. But one user program almost always disrupt mpd on slave nodes and prematurely exits. This user program was reported running fine on other cluster with MPICH 1.

So MPICH 1.2.7p1 will be installed on to the cluster to see if mpd was causing trouble for whatever reasons (load, compatibility ...)

Procedure:

One day I found a script hangs in the middle when it was executed as a remote command through ssh. I log on to that node and run the same script without any problem.  There was nothing special except a background command in the script. Let's just look at this script:

test1.sh

#!/bin/bash


sleep 20 &
echo done

You get message "done" and the shell immediately after issue the command test1.sh directly on the node, but it hangs for 20 seconds if submiited as "ssh localhost test1.sh".

However, this script

test2.sh