3 ms·

github.com is having issues across the site, the API including git operations (and CLI) still work. Status page is manually updated, and we're working to get it

github.com is having issues across the site, the API including git operations (and CLI) still work. Status page is manually updated, and we're working to get it updated. EDIT: it's updated now. EDIT EDIT: github.com is back up and running, apologies for the disruption :( Source: GitHub employee

Miner49er 5y ago
Why manually update vs automatically?
- jondwillis 5y ago
  They wait for the HN thread to appear so they don't give false positives.
- JaimeThompson 5y ago
  I think it is so they can "hide" small outages that don't rise to the level of making the news sites so they can look better. A lot of sites do this sort of thing these days.
  - asciimike 5y ago
    Ish? I wrote some stuff up here: https://news.ycombinator.com/item?id=30182591 https://news.ycombinator.com/item?id=30182591 When I worked at GCP it was all manually updated as well so we could add a sentence about what was actually affected. In any sufficiently large system it's hard to indicate exactly what's broken/how to work around it, so it was just easier to `/status <system> <color> <reason for status>`.
- kxrm 5y ago
  Not many places I have worked for allowed for this to be automatic. A lot of it was so they could provide a coherent explanation as to what the current state of internal attention was directed at vs what everyone can plainly see.
  - Miner49er 5y ago
    I guess this makes sense, but I don't understand why you couldn't have it change status automatically, and still allow a person to go in after and manually add an explanation.
    - asciimike 5y ago
      Mostly because it's more work to manually remove a bunch of spurious robotically added statuses (which look bad, because you're either down more than customers are noticing or your detection is flawed) than it is to manually add and remove real ones.

3 ms·

github.com is having issues across the site, the API including git operations (and CLI) still work. Status page is manually updated, and we're working to get it

by asciimike 5y ago

Miner49er 5y ago
Why manually update vs automatically?
- jondwillis 5y ago
  They wait for the HN thread to appear so they don't give false positives.
- JaimeThompson 5y ago
  I think it is so they can "hide" small outages that don't rise to the level of making the news sites so they can look better. A lot of sites do this sort of thing these days.
  - asciimike 5y ago
    Ish? I wrote some stuff up here: https://news.ycombinator.com/item?id=30182591 https://news.ycombinator.com/item?id=30182591 When I worked at GCP it was all manually updated as well so we could add a sentence about what was actually affected. In any sufficiently large system it's hard to indicate exactly what's broken/how to work around it, so it was just easier to `/status <system> <color> <reason for status>`.
- kxrm 5y ago
  Not many places I have worked for allowed for this to be automatic. A lot of it was so they could provide a coherent explanation as to what the current state of internal attention was directed at vs what everyone can plainly see.
  - Miner49er 5y ago
    I guess this makes sense, but I don't understand why you couldn't have it change status automatically, and still allow a person to go in after and manually add an explanation.
    - asciimike 5y ago
      Mostly because it's more work to manually remove a bunch of spurious robotically added statuses (which look bad, because you're either down more than customers are noticing or your detection is flawed) than it is to manually add and remove real ones.