Runc Breaks Pods When CPU Requests Aren't Multiples of 10
Key topics
A bug report for the runc project was filed, but the report was generated by an LLM, causing frustration among maintainers and commenters; the discussion revolves around the implications of AI-generated content in open-source projects.
Snapshot generated from the HN discussion
Discussion Activity
Light discussionFirst comment
3h
Peak period
4
2-4h
Avg / period
2
Based on 12 loaded comments
Key moments
- 01Story posted
Nov 8, 2025 at 11:47 PM EST
about 2 months ago
Step 01 - 02First comment
Nov 9, 2025 at 2:23 AM EST
3h after posting
Step 02 - 03Peak activity
4 comments in 2-4h
Hottest window of the conversation
Step 03 - 04Latest activity
Nov 10, 2025 at 3:25 AM EST
about 2 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
I'm sure it's also a waste of maintainers time to drop a wall of AI bullet points instead of just sharing the critical information.
Is it mainly post-training that causes this behaviour? They seem to do it for everything, like they are really biased towards super verbose output these days. Maybe something to do with reasoning models being trained for longer output?
And I do mean slop.
It was fixed already: https://github.com/opencontainers/runc/pull/4751
So the bot goes ahead and spams every project that has already done the work. And unrelated ones.
Whilst also leaking that this is to do with a GEO-location project building a cluster on AWS, for Accenture. Probably government contract.
"ip-10-7-66-184.prod-eks.newfront.com"
The insurance company is the client getting their info dumped by a careless approach.
But I didn't really want to dox the client.
Thank God for LLM!
> Thank you for the context! That makes sense - runc is correctly rounding to 10ms (410000) as systemd requires.
You're absolutely right! I shall now commit ritualised suicide based on your feedback! <arrow in target emoji>
Having the AI reply to comments is almost like a direct insult.
https://github.com/opencontainers/runc/issues/4990