No announcement yet.

Login is dead?

This topic is closed.
  • Filter
  • Time
  • Show
Clear All
new posts

  • #46
    It appears that we're back in business. Thanks for the patience everyone.
    Active Projects:
    • Priorities
      1. Bug fixes
      2. Fixes for existing hunting grounds/NPCs (ongoing)
      3. Basran Hill story, eventual lead to ToH.
    • Side projects
      1. Malezzerai (dialogue ongoing)
      2. Armour Rebalancing (ongoing)
      3. Assisting org/event GMs with lore consistency and coding support.
      4. Additions to Monlon (TBA), Blackvine (after Monlon), Rock Valley (after BV), Steps (after RV.) Overall goal to add substance to existing regions.


    • #47
      To offer some final comments on the topic:

      The main problem with our user auth yesterday was that the power supply of a machine had died without a replacement being on hand and (worse) it was a very old power supply that wasn't commonly available. We were able to get an equivalent just in the nick of time (e.g., just before stores that could help us were closing yesterday).

      Generally, power supplies are the weakest element of our setup — or at least the weakest element that sees regular failure. All of our discs are RAIDed, which means we can see a single disc failure without missing a beat (though we've at least once had a double-disc failure which had us biting our fingernails more). Three of our machines actually have redundant power supplies (e.g., two in the case), but that's less than half of our machine count, so clearly the rest don't. I actually swapped one of those redundant power supplies yesterday too, which is the sort of near-failure avoided by our setup that you don't tend to hear about. Mind you, I say power supplies are the weakest element, but the mean time between failure of power supplies is also a lot longer than the mean time between failure of discs; I think I've swapped three power supplies in the last few years (two of them yesterday!) and that was unusually high, while I've swapped over a half-dozen discs, which was pretty usual.

      For years we had our main sysadmin/tech offsite, which wasn't conducive to having spares for stuff on site. Almost two years ago I took over the tech onsite and since I have I've been getting duplicates whenever I have to swap something, to make sure we have spares in the future. Thus, there's a *second* power supply on the way for our userauth machine, which would probably reduce the next downtime of this sort to an hour or so. (Maybe; it's one of those tiny, tiny older 1U machines where it's very challenging to get bits in and out.)

      I suppose we could do something more widely, but that'd require us taking down all of Skotos for perhaps a day, to go through every machine, and make sure we had the right models for hard drives and power supplies. And in the process we'd probably lose a machine or two, given how electronic equipment tends to work (e.g., powering down and up are the times you're most likely to lose equipment). As is, some machines have been running steadily for over 2 years, which increases the likelihood of a failure on power down (and in fact it was one of those machines that died yesterday when powered). So, we've opted not to do that.

      So, that's why things took a while yesterday, why things sometimes take a while generally, and how we're slowly improving things as we can. Overall, we're a very small company, so we make the best we can of limited resources.