adammiribyan

5 ms·

1.
Show HN: Wide Events – Rails telemetry for agents and humans in a DB you own(github.com)
4 points by adammiribyan 2mo ago|0 comments
2.
by adammiribyan 7mo ago
Not yet. Current design is run code, return result. Adding virtio-net to forks is on the roadmap. What's your use case that needs it?
3.
by adammiribyan 7mo ago
Glad to see the approach validated at scale! I hadn't seen your blog posts until they were linked here, going to dig into the userfaultfd path. Would love to chat if you're open to it.
4.
by adammiribyan 7mo ago
Great writeup, bookmarked. The fault storm point is interesting -- our forks are short-lived (execute and discard) so the working set is small, but for longer-running sandboxes that would absolutely be a problem.
5.
by adammiribyan 7mo ago
Both. The engine is open source. You can self-host it on any Linux box with KVM. There's also a live API you can hit right now (curl example in the README). Building the managed service for teams that don't want to run their own i
6.
by adammiribyan 7mo ago
1 vCPU per fork currently. Multi-vCPU is doable (per-vCPU state restore in a loop) but would multiply fork time. On Firecracker version: tested with v1.12, but the vmstate parser auto-detects offsets rather than hardcoding them, so it shoul
7.
by adammiribyan 7mo ago
On tail latency: KVM VM creation is 99.5% of the fork cost - create_vm, create_irq_chip, create_vcpu, and restoring CPU state. The CoW mmap is ~4 microseconds regardless of load. P99 at 1000 concurrent is 1.3ms. The mmap CoW page faults dur
8.
by adammiribyan 7mo ago
Exactly —- they skip the OS, we make it free to clone.
9.
by adammiribyan 7mo ago
Agreed, cross-node is the hard next step. For now single-node density gets you surprisingly far. 1000 concurrent sandboxes on one $50 box. When we need multi-node, userfaultfd with remote page fetch is the likely path.
10.
by adammiribyan 7mo ago
Good callout. We seed entropy before snapshot to unblock getrandom(), but forks still share CSPRNG state. The proper fix per Firecracker’s docs is RNDADDENTROPY + RNDRESEEDCRNG after each fork, plus reseeding userspace PRNGs like numpy sepa
11.
by adammiribyan 7mo ago
CRIU is great for save/restore. The nice thing about CoW forking is it's cheap branching, not just checkpointing. You can clone a running state thousands of times at a few hundred KB each.
12.
by adammiribyan 7mo ago
The API in the readme is live right now -- you can curl it. Plan is multi-region, custom templates with your own dependencies, and usage-based pricing. Email in my profile if you want early access.
13.
by adammiribyan 7mo ago
Thanks! Yes, there's going to be a managed version.
14.
by adammiribyan 7mo ago
Fair question. The fork engine itself is general purpose -- you could use it for anything that needs fast isolated execution. We say 'AI agents' because that's where the demand is right now. Every agent framework (LangChain,
15.
Show HN: Zeroboot – sub-millisecond VM sandboxes using CoW memory forking(github.com)
20 points by adammiribyan 7mo ago|1 comments
16.
Show HN: Sub-millisecond VM sandboxes using CoW memory forking(github.com)
311 points by adammiribyan 7mo ago|72 comments
17.
by adammiribyan 8mo ago
Cool website. I built Croft a few weeks ago — very similar. https://api.trycroft.com/landing-draft
18.
by adammiribyan 11mo ago
Does OpenAI use Ory? I thought they’re using Auth0.
19.
Simple sabotage for software (2023)(erikbern.com)
271 points by adammiribyan 2y ago|75 comments
20.
The Git Parable: Making Your Own Simple Version of Git(practical-neuroimaging.github.io)
3 points by adammiribyan 3y ago|0 comments
21.
Is Cloudflare down?(cloudflare.com)
8 points by adammiribyan 3y ago|7 comments
22.
by adammiribyan 3y ago
Their home page shows “Sorry, you have been blocked” and the status page is not event responding.