💾 Archived View for perso.pw › blog › articles › hacker-news-impact.gmi captured on 2024-05-12 at 15:30:27. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-05-24)
-=-=-=-=-=-=-
Since beginning of 2021, my blog has been popular a few times on the website Hacker News and it draws a lot of traffic. This is a report of the traffic generated by Hacker News because I found this topic quite interesting.
From data gathered from the http server access logs, my blog has an average of 1200 visitors and 1100 hits every day.
The blog was featured on hacker news: 16th February, 10th May, 7th July and 24th July. On the following diagram, you can see each spike being an appearance on hacker news.
What's really interesting, is the different between 24th July and the other spikes, only 24th July appearance made up to the front page of hacker news. That day, the server received 36 000 visitors and 132 000 hits and it continued the next date at a slower rate but still a lot more noticeable than other spikes.
Visitors/Hits of the blog (generated using goaccess)
The following diagram comes from the tool pfstat, gathering data from the OpenBSD firewall to produce images. We can see the firewall is usually at a rate of ~35 new TCP states per seconds, on 24th July, it drastically increased very fast to 230 states per second for at least 12h and the load continued for days compared to the usual traffic.
I don't have much more data than this, but it's already interesting to see the insane traffic drag and audience that Hacker News can generate. Having a static website and enough bandwidth didn't made it hard to absorb the load, but if you have a dynamic website running code, you could be worried to be featured on Hacker News which would certainly trigger a denial of service.
Wikipedia article on the "Slashdot effect" explaining this phenomena
