Hacker News Reader

Yes, HN effect lol-sob.

Mitigation: reducing servers life time temporarily so more people can try.

Usually I roll my eyes when someone posts their own website to HN and it crashes under load. But given the nature and complexity of yours I think there's room for understanding and patience :)

Thanks, I did some stress-testing and infra is scalable enough but I forgot about the AWS quotas, my bad. Quota increase requested and servers are killed off so hopefully "soon" the issue will go away.

Scaling this service without breaking the bank could become its own "sad server" scenario.

I'd start by moving the test VMs to bare-metal servers running libvirt. You can get a 128GB RAM server for ~110 EUR and that should be able to run around 120 concurrent VMs assuming 1GB of RAM to each (CPU isn't a major issue in this case).

Reply on news.ycombinator.com