News for hackers. Stuff that matters.
Re:Hanging S3 downloads by Rjevski on Thursday July 20, 2017 @08:16AM (#14811772)
Yet another problem caused by shitty consumer-grade network equipment.
Re:Hanging S3 downloads by matthewmacleod on Thursday July 20, 2017 @08:46AM (#14811966)
I think you've maybe been unfairly downvoted there!

Almost all of the issues I experience with Internet connectivity day-to-day are down to bad consumer networking equipment. APs start failing, or routers stop routing, seemingly at random.

Re:Hanging S3 downloads by deathanatos on Friday July 21, 2017 @01:00AM (#14818178)
For example: my router/AP/modem combo-box — which is the standard one that Comcast has been giving to all subscribers — when reset or power cycled, will forget the WiFi key. It just gets reset to the factory default. Every, single, time. And it's just they key: the SSID doesn't get reset. It's not the first modem I've had, either; its predecessor was the same make, had the exact same behavior.

I suspect it's because I'm using an actual key, instead of a passphrase, and nobody else does that. (Or, at least, not enough for Comcast to notice that the problem is them.)

The old one got returned b/c it was routinely dropping IPv6 packets. It took a while — multiple vists from multiple technicians — to convince Comcast that the symptom of "100% packet loss over IPv6, 0% packet loss over IPv4" was not a problem with the coax cable.

The predecessor of that modem got replaced when it had extreme difficulty forming IPv4 connections: almost all SYNs never got a SYN ACK, but in the rare case that they did, that connection proceeded with 0% packet loss. I highly suspect there was something going horridly wrong with its NAT processing, but that's a guess.

Re:Hanging S3 downloads by Rjevski on Friday July 21, 2017 @07:37AM (#14819579)
The worst is that the Linux kernel (and any BSD) would be able to make a perfect router, but manufacturers insist on using their own shitty implementation. If they would've just used used the default Linux/BSD they would've been fine.
Re:Hanging S3 downloads by mjw1007 on Thursday July 20, 2017 @08:34AM (#14811886)
Also: yet another problem caused by NAT.
Re:Hanging S3 downloads by the_mitsuhiko on Thursday July 20, 2017 @09:20AM (#14812192)
And yet NAT solves more problems than it causes.
Re:Hanging S3 downloads by Rjevski on Thursday July 20, 2017 @09:37AM (#14812302)
I disagree. What problem does NAT solve, other than allowing people to cut corners and not deploy IPv6?
Re:Hanging S3 downloads by the_mitsuhiko on Thursday July 20, 2017 @01:13PM (#14814409)
> I disagree. What problem does NAT solve, other than allowing people to cut corners and not deploy IPv6?

It by default ensures that machines inside a network are not reachable from the outside world. It gives you one address to the outside world which made IP blocks for abuse work quite well and enabled a lot of other users.

Sure, it's a bit of a pain for UDP communication but given the common NAT behaviours not completely terrible. If it was defined a bit clearer NAT would not have been that controversial.

Re:Hanging S3 downloads by StavrosK on Thursday July 20, 2017 @04:49PM (#14816159)
So does a firewall, in a much less complicated way.
Re:Hanging S3 downloads by the_mitsuhiko on Thursday July 20, 2017 @05:43PM (#14816562)
NATs are not terribly complex and most consumer grade firewalls are significantly worse than a simple NAT is.
Re:Hanging S3 downloads by kingraoul3 on Thursday July 20, 2017 @10:19AM (#14812640)
Well, it allowed us to keep growing the internet before we had IPv6, for one thing.
Re:Hanging S3 downloads by wbl on Thursday July 20, 2017 @10:36AM (#14812820)
You might want to look at when IPv6 was designed. It's quite a bit older than you might think.
Re:Hanging S3 downloads by sulam on Thursday July 20, 2017 @03:53PM (#14815726)
Designed is very different from available in any form. If I couldn't buy a router that supports it with any amount of money, NAT was my only answer. This was the existing state of the world for a long time.
Re:Hanging S3 downloads by Rjevski on Thursday July 20, 2017 @10:36AM (#14812814)
Wasn't IPv6 already out before we ran out of IPv4 addresses? In which case NAT is just a kluge for the lazy who didn't want to deploy IPv6.
Re:Hanging S3 downloads by diamondlovesyou on Thursday July 20, 2017 @11:20AM (#14813236)
One can't control those that must deploy IPv6 first. That's the problem NAT solves. Developing NAT was likely not easy; consider how many implementations f it up anyway.
Re:Hanging S3 downloads by growse on Thursday July 20, 2017 @09:39AM (#14812306)
The main problem it causes is that it breaks the internet: not all peers are routable. It's difficult to think of a greater benefit that offsets that.
Re:Hanging S3 downloads by the_mitsuhiko on Thursday July 20, 2017 @01:14PM (#14814413)
> not all peers are routable.

For me that's a core feature of NAT, not something that breaks the internet.

Re:Hanging S3 downloads by growse on Thursday July 20, 2017 @06:28PM (#14816800)
It breaks the internet because it breaks IP - the internet protocol. Segregating devices onto separate networks that are not routable to each other is something that reduces the overall functionality of the internet, not enhances it.

It doesn't help that it's a shitty hack which causes endless lost time either spent wondering why something is broken, or through designing entire protocols just to get around it.

Re:Hanging S3 downloads by the_mitsuhiko on Saturday July 22, 2017 @02:12AM (#14825689)
If you consider that "breaking" the internet then many thing break it. That seems like a weird argument to make.
Re:Hanging S3 downloads by matthewmacleod on Thursday July 20, 2017 @06:05PM (#14816687)
I understand where you're coming from with that – but I can imagine a different Internet, where devices where by default globally routable but firewalled. That could even have affected the current trend towards 'stuff it all in my proprietary cloud system'.