Post #2103864

2026-05-06 09:16 UTC

Whenever I want to explain what I mean by &quot;availability is political&quot; and &quot;uptime is a Noble Lie,&quot; I&#39;m going to start by discussing these two pages: https://mrshu.github.io/github-statuses/ vs https://www.githubstatus.com/ Even the per-component uptimes are different! And the aggregate calculation is… interesting

Replies (2)

@jpetazzo@hachyderm.io 2026-05-06 09:23

@hazelweakly this is giving me flashbacks to conversations 15 years ago, when I was running a PAAS and my boss wanted us to provide uptime figures. "Well, do you want the availability of our API endpoint on some basic call? End to end testing that deploying a basic app works? Aggregated uptime of our customers' apps? (Why if they're down by their own fault?) Of their deployment attempts? Should we compute an average weighted by, say, the revenue for each customer?" The worst part is that 15 years later, I don't think I'm in a better position to answer these questions 😅

Open ##2120088
@adrianc@hachyderm.io 2026-05-06 11:42

@hazelweakly im wondering how exaggerated each one is now… I know people have been complaining about GitHub being down a lot recently

Open ##2120107