UptimeUptime Wiki

Uptime 0.1.3: "It's Always DNS"

We built a datacenter sim where the incidents are real. It turns out one of them was a little too real: a scripted DNS outage that opened, degraded every customer, and then, in the finest tradition of on-call, never actually ended. It just sat there. Bleeding customers into the small hours while you refreshed the same three menus looking for a fix that did not exist.

It does not do that anymore.

Fixed

  • The DNS outage that would not die. The demo's facility-wide DNS outage was scheduled to open but never scheduled to close, so it ran until the heat death of your save file. It now resolves on time, the way the code comment always insisted it did.
  • The achievement paradox. "It's Always DNS" unlocks when a DNS incident resolves. The DNS incident did not resolve. A few of you correctly diagnosed this as, and we quote, "it's always DNS," then closed the ticket. The achievement is now obtainable by mortals.
  • The haunted LAG. You could disband a link aggregation group and keep an open incident for it, loyally referencing a bond that no longer exists. The alarm has been informed of the LAG's passing. Disbanding a bond, or uncabling its last member, now clears it. Pulling one cable from an otherwise healthy LAG still does not, because that is usually you tidying up, not a disaster.

Changed

  • DNS outages now last 30 sim-minutes, down from a full sim-hour (about half a minute at demo speed). Long enough to sting, short enough that "wait for it to clear" is a real answer and not a lifestyle.

Saves

  • Affected saves self-heal. If you were stuck with an eternal DNS outage or a ghost LAG alarm, just load your save on 0.1.3 and it clears itself. No action needed. A session that is running right now clears on its next load.

For the record: it was DNS. It is always DNS.

On this page