Skip to content
 

System Uptime Personal Record: 4.75 Years And Growing (Debian Linux)

Background

*update* currently 5.2 years (04:40:08 up 1900 days, 15:05,  1 user,  load average: 0.00, 0.00, 0.00)

I’ve worked for a small company for many years that does a lot of things, one of them being hosting.  I have the opportunity to run and maintain dozens of machines for various private sector contracts.  My personal choice would be FreeBSD for a bare metal OS but, since it is fairly hard to find BSD admins compared to Linux (read: pay grade), I have to run Linux in case I ever leave.  Windows has gotten better over the years yet Linux and especially BSD has always been renowned for its ‘rock solid’ behavior.

We have one server in a group that thus far has not seen a reboot since it went live.  This debian based machine is firewalled off except for web services, Apache got needed security updates, and we started with an older stable kernel that has yet to see a critical security update that applies to the network environment.   The contract has expired and the machine is not currently used ( gotta love bureaucracies ).  I’ve taken screenshots of the uptime for myself over the years… at 2, 3, 4 and now 4.75 years.

 22:17:35 up 1737 days,  9:43,  1 user,  load average: 0.04, 0.08, 0.07

This is a testament to Linux and open-source.  I would not be surprised if Windows could match this now but it still impresses me.  Below I have posted two nearly famous images depicting the call stack of a web request for Linux based Apache and for Windows IIS.  By itself it does not mean much but history has generally shown that the more complex software is, the more prone it is to problems.  Although being over half a decade old, these images just about speak for themselves.  Efficient software can be made proprietorially I am sure, but from my experience does not happen unless it is a serious requirement.  Windows is more of a pumped out ‘make it work’ system whereas large projects in the open-source arena benefit from a lesser bureaucracy, letting more eyes see the code, and more work and acceptance towards elegant solutions.

These images are unfortunately not full resolution.

Thanks Linux and the open-source community.  I will continue to update this until the server is rebooted or decommissioned.

3 Comments

  1. Bill Lewis says:

    We built a server and a router using two atom K510 mini-itx motherboards in a single 1U case (we have to pay for rack space per U). the design is fanless and the server provides websites and email to over 1400 customers.

    I built the server and commissioned it into the data centre. The data center is closing down later this year and the server has not been rebooted since we fitted it :-

    ]# uptime
    00:14:34 up 860 days, 14:33, 1 user, load average: 0.62, 0.48, 0.50

    uname -a
    Linux arunxxxxx.net 2.6.32.16-150.fc12.i686.PAE #1 SMP Sat Jul 24 05:25:42 UTC 2010 i686 i686 i386 GNU/Linux

    bill

    • robert says:

      That is impressive for random commodity hardware. I love how server systems today are trending towards a more redundant commodity hardware setup instead of ‘high end’ systems that cost a fortune.

  2. Sadly, this one will be shut down in tne days from now because we’re moving our infrastructure after (bet it…) three years:
    root@inanna ~:5017 $ w
    00:36:35 up 1095 days, 5:18, 1 user, load average: 0.43, 0.52, 0.66
    USER TTY LOGIN@ IDLE JCPU PCPU WHAT
    root pts/3 00:33 0.00s 0.00s 0.00s w

Leave a Reply