Tuesday, February 17th, 2009 | Author:

Some of you may already know that I built a home server not too long ago. I documented some of the very important parts of how it was built though I was planning on releasing all the documentation all at once. I was using Arch Linux and I hadn’t nearly finished everything, especially the documentation. For example, it was supposed to be a media server. After some disk shuffling, it was supposed to end up having a RAID1 for the boot and RAID 10 for the rest (the media part).

This didn’t work out at all.

I got as far as having an efficient (and wellfirewalled) routing gateway server. I was finally satisfied that the customised local routing*  was working correctly and I was confident that my tests with DHCP meant I could disable the DHCP service on the flimsy ADSL router and have all my flatmates start using the server as the Internet gateway. Instead: I was logged in to the server from the office, I’d just installed Apache2**, and I was about to consult with a colleague regarding getting nice graphs put together so the flatmates could all see who was using up the bandwidth*** — when I noticed a little message indicating that the root filesystem had been remounted read-only due to some or other disk failure.

And then I lost my connection to the server.

And then I gained a foul mood.


When I arrived home, I found that, as I had guessed from the descriptive message given at the office, the (very) old 80GB IDE disk that I was using for the root filesystem had failed. Unfortunately, the server would never boot again and there was little chance of prying everything off onto another disk to continue where I’d left off.

I’m buying a replacement (SATA) HDD this next weekend just after pay day – and I’ve changed my mind about documenting my progress… and backing up my configurations:

Release Early. Release Often.

* ISPs in South Africa charge less (easy price comparison) for “local-only” (within South Africa)  traffic on ADSL but only if you use an ADSL account that CANNOT access web services outside of South Africa. This means that if you want to take advantage of the reduced costs but still be able to access the Internet at large, you need to set up some sneaky routing.

** one-command-install: ~$ yaourt -S apache

*** Internet Access in SA is expensive – you get charged about R70 ($7 / £4.9 / €5.46) per GB when using ADSL, or about R2 per MB if using GPRS / 3G.

You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

6 Responses

  1. 1

    omw dude. u all fighting over bandwidth in your haven? haha
    Too much WOW i say..

    Ps. Remember that time i asked if you had a unix server hosted somewhere so I could use it as an ssh tunnel, affectively avoiding my works proxy so that I could do such things like…errr…ftp?

    This is quite simply.. Karma 🙂

  2. 2

    and could you elaborate on this sneaky routing?

  3. 3

    pps. Dont remind me whose hosting package my current domain is sitting on. That good will eventually find its way to you soon enough. Even if she does wear a wig, fancys green-illumo underwear, has 2 and a half boobs and has more than just on tooth ..

  4. 4


    The sneaky routing involves using 2 PPPoE connections. One using a standard “international” account and the other using the local-only.

    Then, keeping handy a list of all the local-only ranges, we route all traffic destined for those ranges over the local-only PPPoE connection. 🙂

  5. Re RAID, were you using your motherboard’s “hardware” of software RAID? And what went wrong?

    My personal experience is that unless your hardware RAID costs more than a house, you want to go the software route.

  6. 6

    well… the *plan* was to go raid. 😉

    I believe in software RAID as well. Software raid is more portable and you don’t have to depend on a hardware manufacturer keeping stock of the RAID cards 5 years on. 🙂

Leave a Reply » Log in