Hacker Newsnew | past | comments | ask | show | jobs | submit | rumblefrog's commentslogin

Easier said than done, vendor lock-in is costly to move from.


I feel like that job would fall on them :P


I'm not sure about that


I tried to deploy a small cluster in the US VA region, but the cluster status kept flipping between Failed and Creating with no clear way of troubleshooting it: 7ad975fb-3c8e-47a9-b03d-9e6bec81f0db


Hello there, sorry for that I will look into it right now.


I wonder how long before Hetzner adds something like managed Kubernetes to their native product line. They already have S3 compatible object storages, load balancers and more.



No idea about the timing but I imagine it's coming.

Would make a lot of sense, especially if you can combine it with the hardware servers. You could get a lot of grunt in your cluster for a lot less than for example AWS.


Given how rarely they offer specific software solutions and at what pace, I would say 5+ years from now or never.


This looks great! Haven't tried it yet, but should I presume this also does k8s and OS updates for me? Or how managed is it?


Thanks for the feedback! The platform is mostly self service, but it is very easy to upgrade the Kubernetes version, just change the version in the cluster configuration. For OS updates, you can replace the nodes and it will automatically pick the latest OS image from Hetzner. I also run it isolated for some small companies, as a fully managed service, so that option is available as well.


Could you point to the source of the on-device model? Moreso for curiosity.


No, but I can tell you that the moment you open the browser console, it stops scanning and marks the scan as failed.

The vendor is https://www.k-id.com in Discord's case


That just seems like a standard anti-tampering measure, I don't think it necessarily means the model is local or anything


https://www.k-id.com/post/adapting-discord-for-the-uk-online... says that:-

> Identity documents are deleted after a user’s age group is confirmed, and the video selfies used for facial age estimation never leaves their device.


I find the map the most interesting map: https://map.bgp.tools/

And did not know Mercedes holds such a large block (53.0.0.0/8), and it appears to be mostly dark or unused.


See "List of assigned /8 blocks to commercial organizations" at: https://en.wikipedia.org/wiki/List_of_assigned_/8_IPv4_addre...

Mercedes-Benz isn't the only corporation with its own /8 block - although it is the only non-US one; and it's a pretty small exclusive list, only six in total. I guess someone at Mercedes-Benz just happened to hear about this Internet thing relatively early (53.0.0.0/8 was registered Oct 1993 according to Wikipedia - the relevant European authority, RIPE NCC, was only founded in Apr 1992), and/or happened to have a buddy at IANA / RIPE NCC, and the rest is history.


Does anyone know what the red striped sections are? Hovering them only yields < <range> is excluded from scanning, but that doesn’t say much


They're part of AS749, which is US Department of Defense IP space that appears to be unused and which, based on public statements made by the Pentagon, might just be safely parked or might be part of a network scanning observatory [1]. Either way, scanning that space is probably a waste of bandwidth.

[1] https://blog.benjojo.co.uk/post/ip-address-squatting


Maybe a bit off topic, but years ago I had a home linux server, used for usenet/torrenting. I was just poking about on it one day and ran some variation of netstat and could see a connection with a 6.x.x.x range IP address, which stood out. I didn't know a whole lot about networking at the time (still don't), couldn't say if I had misinterpreted what I saw in the netstat output. But it stood out so I looked it up and hoo-boy, it was a shock to see who owned that IP range, Army Information Systems Centre. I chalked it up to them maybe running a torrent tracker or something, or was the DoD in my PC?


Service providers (especially mobile ones) often squat on large, unused IP blocks (of which the DoD has plenty of), especially at peak capacity. I suspect this is what occurred here.


> Either way, scanning that space is probably a waste of bandwidth

That's what the DoD wants you to think ;)


Just a fascinating map. Like looking into Neuromancer’s cyberspace.

Is there a legend of what the different colors mean?


The XKCD Map of the Internet is good too (2006): https://www.xkcd.com/195/


The map also does a good job showing why we need to do away with multicast/class E and reuse the address space.


At this point, time would be better spent moving to IPv6, don't you think?


The problem is that you can't really move completely to IPv6. You actually will have to run both IPv4 and IPv6 networks indefinitely. Which isn't a lot of fun.


Now, yes. But eventually, we'll reach a tipping point where that isn't necessary. When? Your guess is as good as mine.


"You actually will have to run both IPv4 and IPv6 networks *indefinitely*."


Sure you can. You can do it today even, let alone in some unspecified number of years.


We have been trying to deploy IPv6 for 20 years now. This would be comparatively easier and buy us another 20 years to finish v6 deployment.


A ton of old equipment would need to be upgraded to use 240/4 for IPv4 unicast. We'd run into weird issues where it works for some people and not others. I'm not convinced. If this was done 25 years ago, maybe.


And 50% of the internet traffic is IPv6. The proposal here is to introduce a separate (arguably harder) change which would start at 0% support again. Beyond that, it'd just be a temporary fix.

Just 240/4 allowed specifically for private network use (like the 10/8 range)... that I could get behind though. This would still exclude 255.255.255.255/32 of course.


and what, earn 5 more years till we have the same problem?


That's a problem for the people in 5 years


You can’t use multicast because multicast addresses work today on private networks


Surely something from https://en.wikipedia.org/wiki/Reserved_IP_addresses could be repurposed?


Why? It would be to the benefit of relatively few and be an enormous technical problem for decades for everyone. If we did this, the IPv4 addresses would run out again after a few months at most, leaving us all in the same position we are in today, but also with a huge technical problem of fixing all the old devices which had these now-repurposed networks hardcoded.

If you are fine with doing the work of obsoleting old equipment, then just start using IPv6.


240/4 can be hijacked/used as private IP space currently by a lot of devices. I think Windows might be the hold out. But for internal routing and IoT it’s very useful. Can never be used as publicly routable space.

I’d like to see it added to the official list of private space alongside RFC1918.


Not sure why you'd need more than 10/8, but I suppose it looks even cooler too those who know and find such things cool.

Does Windoze prevent you from using certain IPv4 addresses?


> Not sure why you'd need more than 10/8

Large organizations have moved to IPv6 because they, and everyone else, are using 10/8, and so when mergers and acquisitions happen trying to connect the networks together becomes a nightmare.

See this talk from Wells Fargo as an example:

* https://www.youtube.com/watch?v=EzTWjNUb4H4


The pain of merging acquired company networks cannot be overstated!

IPv6 can help, but in my experience there's just soo much old IPv4 tech that no one is prepared to migrate to IPv6.


Aws sized companies, we also use them in aws. And no doesn't work on windows.



No. It will take 10 years for everyone to update their router configuration/software to treat the new “formely-reserved” addresses as global unicast. There’s no point in doing that whatsoever. That effort would be spent much better by adopting IPv6.


Holy cow. Didn’t realize there is an entire /4 reserved for future use.


Lol, then what do we do with all the multicast traffic?


I'd guess they bought a whole block early on because they could and it wasn't too expensive yet. I wonder if they could be compelled to auction it off. But they might be using it a lot internally.

(a lot of guesswork in this comment, I don't know anything lol)


no, no one bought anything, they just emailed some dude and said "hey we're connecting to The Internet, we need some IPs" and the reply was "here is the 2^24 IP addresses dedicated to you".


In the early days no money changed hands at all. You just asked for a block and got one.


But Andrej is no longer with Tesla?


Is there an alternative that integrates with a Jira instance?

Many of my tasks already exists in forms of a Jira ticket, would be interesting to prompt it to take over a specific ticket & update its ticket progress as well.


For such kind of tasks I would go with Taskmaster AI. It had mcp integration and probably could connect with jira.

Backlog is more for smaller projects where you wouldn’t normally have a project management tool


Isn't the Github repo a mirror?


Yeah the commit messages suggest all the heavy-lifting was done on kernel.org's server. It's just matched his GH user to the email in the commit.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: