I have a lab with 17 seats. The server is one of the seats and is six years old. The clients are 8 years old. One of them has a failing hard drive with increasing failures at fsck and it had to have the OS re-installed. For Debian GNU/Linux that should not happen. Rather than change the hard drive, I installed LTSP on the server and the failing machine now boots PXE from the server and boots in 5s less time. It took about the same time to install LTSP as to change the hard drive on the client:
Booted the client and it is in business. Since the booting is 5s faster than from the hard-drive (spin-up and GRUB menu delay) I will switch all the clients to do this as we do not really need the storage on the clients anyway. I could unplug the drives to save power.
How cool is that? I love GNU/Linux. It is the perfect solution to so many problems of IT in education. I have used LTSP since 2003. It just keeps getting smoother although there is some bloat (115MB RAM…). LTSP4.2 could run in 44MB. The difference is mostly that LTSP5 uses the typical structures of a thick client rather than a minimalist chroot. The advantages are that I can use apt-get inside the chroot to use Debian’s management tools.

17445
12772
206
3
2
23891
11850
11702
4641
4257
1642
198
14
2
0
0
0
0 Responses to “Fixing a Dying Hard Drive in the Lab”