On 01/27/2016 09:30 PM, Ross Finlayson wrote:
Just a quick FYI, for those people out there who thought that I was being
unrealistic and/or alarmist:
http://www.theregister.co.uk/2016/01/28/github_outage/
It was reported on NANOG that this was the web interface only. The ssh
interface (which is what developers use to actually update their repos,
push changes, etc.) was still working.
It's also worth noting for those that are not familiar with git that one
of its strengths is that you can trivially clone the repository itself,
and work locally from your cloned copy. So the only thing that actually
wasn't working was the ability to see things on the web, and the ability
for folks to push their latest updates to github itself. Users who were
taking advantage of git's distributed features would still have been
able to share updates with each other.
Regardless of what protocols (‘git’ or otherwise) the IETF decides to use for
document management, we should be using our own servers for this - not those belonging a
third party. We need to be confident that this is going to survive the next recession :-) :-(
I have no objection to us doing this in-house. What I do object to is
continued inaction. For example an organized pilot program taking
advantage of an "outsourcing" opportunity like github could bring
tremendous insight and valuable operational experience; which could be
used to formulate a longer-term plan (whether that ultimately is
in-house or not).
Doug