3 ms·

I thought it was pretty serious. Github is no longer a tiny company and when so many people rely on you for everything a 5 min outage becomes significant. E.g.

by soheil 13y ago

I thought it was pretty serious. Github is no longer a tiny company and when so many people rely on you for everything a 5 min outage becomes significant. E.g. I was not able to deploy to my server a few mins ago due to this outage. This makes me question the decision to use github going forward.

ColinWright 13y ago
What value does it add to HN to have outages like this reported? As far as I can see, and as I said elsewhere, if you're affected you'll already know, and if you're not affected it won't matter. I concede that it was important to you, as no doubt it was important to many. The question is, why should it be posted to HN? There might be value in a longer post with real analysis of outage statistics and patterns, but this isn't it.
- soheil 13y ago
  I don't think the point is to merely be aware of the outage (as most who rely on github so heavily would most likely already have checks in place to notify them), but rather start a conversation, no matter how un-actionable, on whether the next guy should be using github.
dalore 13y ago
Surely you have a local git repo? You shouldn't need git to deploy, build a system package and deploy that.
- soheil 13y ago
  There is a local git repo for local env. But I don't see why it's wrong to use git to deploy to staging/prod using different branches. Surely there is a more perfect solution but who said anything had to be perfect, we're hackers.
  - dalore 13y ago
    Not wrong, and what I view as an intermediate step of the evolution of a project. But the crux of your argument, also invalidates it :) as a hacker you should be able to hack around github being dead.
andyhmltn 13y ago
This is where people go wrong. Why throw the advantages of Git out the window and depend upon one master repo? You are aware that you can use your local repo are you not?
- Dylan16807 13y ago
  Git makes it pretty awkward to sync a branch via multiple paths. A globally-accessible repo that never goes down is the only easy way to transfer revisions.
  - andyhmltn 13y ago
    Yes, that's true but that's acting like there's no fallback when things do go down. A 'repo that never goes down' doesn't really exist (as is proven here)

by soheil 13y ago

ColinWright 13y ago

What value does it add to HN to have outages like this reported? As far as I can see, and as I said elsewhere, if you're affected you'll already know, and if you're not affected it won't matter. I concede that it was important to you, as no doubt it was important to many. The question is, why should it be posted to HN? There might be value in a longer post with real analysis of outage statistics and patterns, but this isn't it.

soheil 13y ago
I don't think the point is to merely be aware of the outage (as most who rely on github so heavily would most likely already have checks in place to notify them), but rather start a conversation, no matter how un-actionable, on whether the next guy should be using github.

dalore 13y ago

Surely you have a local git repo? You shouldn't need git to deploy, build a system package and deploy that.

soheil 13y ago
There is a local git repo for local env. But I don't see why it's wrong to use git to deploy to staging/prod using different branches. Surely there is a more perfect solution but who said anything had to be perfect, we're hackers.
- dalore 13y ago
  Not wrong, and what I view as an intermediate step of the evolution of a project. But the crux of your argument, also invalidates it :) as a hacker you should be able to hack around github being dead.

andyhmltn 13y ago

This is where people go wrong. Why throw the advantages of Git out the window and depend upon one master repo? You are aware that you can use your local repo are you not?

Dylan16807 13y ago
Git makes it pretty awkward to sync a branch via multiple paths. A globally-accessible repo that never goes down is the only easy way to transfer revisions.
- andyhmltn 13y ago
  Yes, that's true but that's acting like there's no fallback when things do go down. A 'repo that never goes down' doesn't really exist (as is proven here)