A few hours on the Hacker News front page

Categories

Background

Earlier this month, one of the blog posts on this blog came to the first page of Hacker News. It was early hours of 17 Oct when that happened. I was in Vietnam and actually a day before that had a long day at VNIX-NOG 2026. During the event, Cidy Yang from Akamai gave a talk on Akamai and IPv6 (slides here). During the talk, she mentioned RFC 8950 - routing IPv4 over IPv6 sessions.

In her talk, Slide 14 was interesting.

I ended up tweeting that, along with a context link to my own blog post which I published in June 2026.


Someone picked it up and posted the link on Hacker News. I strongly doubt it was a coincidence that someone picked it up that same day. It was likely a result of that talk. I was back in my hotel room and crashed early. There was an outing planned for early the next morning. At 3 am I woke up with a couple of alerts triggered via Prometheus due to high visitor traffic on all my CDN nodes. I put this alert rule in place months ago when some large scraping was happening & was curious to get alerted. I looked around; there were over 10,000 visits on that post, and as I looked around, I found this:

My post was on #28 on the front page of Hacker News. 😀



Traffic and infra

The post stayed on the front page for a few hours before dropping to page 2, where it remained for a while. This brought very high traffic by the standards of this blog. Most posts here are about BGP routing, DNS or DevOps stuff. These are niche topics, so traffic is usually low. Served around 40,000 visits while the post was on the first page. This includes traffic from Hacker News and from websites that mirror its content. Some of them also send summaries by email.

This blog itself is a Hugo based stateless site. I ran it on WordPress between 2008 and 2022. In 2022, I moved to Hugo & in phases I moved the blog to a few of my servers for the “CDN effect”. I covered parts of these in post here, here and here. As of now the blog is hosted on 5 servers around the world. One of them is a dedicated server in Mumbai; others are cheap VMs in the US and EU. Traffic lands on these servers based on the source of the DNS resolver or client subnet EDNS data. My auth DNS server (Self-hosted PowerDNS) takes care of resolving traffic to the nearest node. Next, it lands on an NGINX reverse proxy where all TCP 80/443 traffic terminates and in the backend has:

  • Web files for blog generated by Hugo
  • Static files (usually pics, videos, screenshots etc) on a S3 NGINX Gateway which pulls this origin via Backblaze B2 in Amsterdam.

All web page content sits locally on these servers and static content is cached for a day to speed things up. For now, this setup has worked well, and I enjoyed the traffic while it lasted. 😀