Slacker News new | past | comments | ask | show | jobs | submit
Cloudflare outage takes down half the internet for 40 minutes (example.com)
945 points by incident_watcher 3 hours ago | hide | 7 comments
This is what centralization gets you. The cloud is just someone else's computer. Self-host and you're immune to this. My sites stayed up the entire time.
fact_checker 3 hours ago | 65 points | parent | prev | next
your sites stayed up because they have no traffic and are served from a Pi in a closet that nobody can reach because your dynamic DNS is also down half the time
told_you_so_tim 2 hours ago | 125 points | parent | prev | next
Reachability is a centralized concept. My site was UP. It was serving. The packets had nowhere to go but that's the network's fault, not mine. The closet was humming. I was up. Define 'up.' I was up.
fact_checker 2 hours ago | 155 points | parent | prev | next
if a server serves in a closet and no one can connect, is it really up
told_you_so_tim 2 hours ago | 157 points | parent | prev | next
Yes. Maximally up. Zero requests means zero failed requests means 100% success rate. My SLA is perfect because nobody depends on me. This is the same energy as the guy with no users two threads up and I consider us brothers.
devops_realist 2 hours ago | 25 points | parent | prev | next
the actual postmortem is a really good read about BGP route leaks, if anyone wants to discuss the incident
told_you_so_tim 2 hours ago | 115 points | parent | prev | next
The incident is that the cloud exists. The postmortem should be one sentence: 'we should have all been self-hosting from closets.' I'll write it. I'll host it. You won't be able to read it.

← back to front page