This month has been punctuated by uptimes for me.
The first to come was a webhead in a farm we run, running Solaris 10 on 64-bit x86 and about 9 instances of Apache 2.2.x, which had managed 928 days without a reboot. That’s two-and-a-half years. The second was an external VPS that was rebooted after 356 days as it only seemed right.
The funny thing is, I have mixed feelings about uptime.
It takes a certain breed of person to get all excited about uptimes. I think in simpler times, at the risk of sounding twee, is was a badge of honour. Those TANDEM or VAX boxes stuffed away in banks and exchanges didn’t have to worry about a daily onslaught of attackers and ner-do-wells exploiting software stacks deep enough to reach the knee. At my time at a highstreet British bank I came across TANDEM boxes with uptimes in years and S/390s robust enough that I only saw one outage during my years working there. And I think that was something to do with ILOVEYOU.EXE anyway.
Today, if you are professionally involved in systems administration and architecture, long uptimes are seen as an omission that you’re not doing your job correctly. The ’Modern Administrators Handbook and Gateway To A Gittish Car And Suit Guide 101’ has it that you should be patching often throughout the network – the whole “hard shell soft core” thing is bad. This is certainly good practice for any trivially redundant site. We’ve all done the “fuck we run NT4” thing, learnt our lessons and thank $DEITY that we’re at least running a UNIX flavour.
That said, I said I have mixed feelings. That our facilty kept a box up for 2.5 years without interruption is something worth at least a nod. That we considered Solaris 10 to be robust enough to not warrant a reboot for two-and-a-half-years, let alone the fact that the thing didn’t fall over by itself, is a testament to all of the people who’ve had a hand in that operating system. That the Supermicro box, the Skoda of data center kit, didn’t falter/catch on fire/spit out the PSU, was on and powered and working without fault for that amount of time is, still to this day having done this lark for a decade, quite amazing.
I would be interested in hearing any uptimes stories you have. Just like fishing, feel free to add a coupla-hundred days as it suits the story.