Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[wierd]

The real issue with NAT comes in when you have a buttload of ports open / a lot of NAT'd clients.

These are often weak kneed devices running embedded linux on extra weak SoC based processors (that are modelled after chips from the 90s, like MIPS). Having a dozen or more stateful connections in the NAT port mapper PER CLIENT, starts to have a toll on such equipment pretty fast, andbthe only way it can dealwith it, is to hang up sessions.

On the other side, not being able to be reached without initiating from behind the NAT makes hackimg a lot more difficult.

[Starver]

If it were just IPv4 limitations, then IPv6 could have 'solved' that problem. For the foreseeable future..

Security/compartmentalisation is the main thing. You're going to have a modem/router (or a paired combination of the two separate devices, if that's how you roll) which might give everything 'inside' semi-permanent RFC1918(ish) addresses which you can then leave to just be passively serviced by NATting or tweak to your requirements (set up exacting 'port-holes', if necessary, or even setting up a 'red/amber/green'-list to allow your own firewall to do far more complex monitoring, filtering and auto-forwarding).

If you think about it, the process is so very well lubricated for the modern user. Plug in your device with an 8p8c-ended cable (these days, can often be straight or crossover) or over any negotiable hotspot. And most things autonegotiate TCP (or maybe UDP) connectivity without the user having to do anything (you don't even choose whether your browser tries to use port 80, 8000, 8080 or whatever, it just happens). From a security side, it's frightening how much we defer to aggressive autohandshaking and, if anything, NAT-routers are provides with too few speedhumps to their users[1].

I'm intrigued about the "learning about networking", to be honest. It's a whole interesting field (from 'configuring winsock' to packet-shaping techniques), depending upon how in-depth you go into it. And a lot of it is darned archaic unless (in some still 'relevent' cases) you're going to want to work some particularly weird magic at unusual levels of the OSI stack. Though of course it may help to understand some of the more abstracted bits work when trying to debug why you can't seem to open streams to things. Depends where you're hoping to take things.


(You don't actually need to traverse your awkward NAT to experience much of the 'regular' networking experience. If you have two computers on the same router, you can get the basic feel. Or an abbreviation to that could be to just use localhost and 'bounce' between two windows on the same machine, almost regardless of whatever else you're connected to. You can get a surprisingly amount of the Network+ learning experience by such self-study, if that's where you're heading, though obviously there's a lot more interesting things to do if you have a whole network of active machines out there to give you a variety of end-devices/routers to connect to and/or fingerprint.)


[1] Though you can imagine the frustration if, when you first decide to try out VOIP, Torrents or whatever, you have to go into the router-config interface to add whatever new ports you suddenly decide you want to start tunelling through. (Or go to your Dad and get him to grant you this access on your family-home router, possibly trying to subtly avoid why and how much you might be jamming up the local router with your latest obsession.)

[wierd]

re: [1]

Manual port mapping is something I in fact, DID do during the Napster/Kazaa era.  This was the era before uPNP, and the router I had was EXTRA weaksauce, and would hang up the sessions pretty fast if I didn't.

It was also necessary for certain local game serverlets to run reliably.

[heydude6]

*snip*

Interesting, so I probably should do more reading about universal plug and play then. I'm surprised minecraft servers don't have it though.

As for "learning about networking", I just took an intro course in university and thought I'd try to apply my knowledge to code a few personal projects. I know about using localhost and multiple windows to do basic testing, but I wanted my program to be able to work across the internet so I had to look into port-forwading.

Turns out that I wasn't the first person to invent the punch-through, but the limitations are kind of nasty. I have full-access to my router so I can forward all the ports I want, but I wanted to make something less tech-savvy people can use.

[Starver]

I have full-access to my router so I can forward all the ports I want, but I wanted to make something less tech-savvy people can use.

Without knowing specifics (and plenty has changed in the years/decades since this was my primary concern), one way of establishing a handshake is to use a 'guaranteed' port[1] that's already likely to be tunelled.

The network doesn't really care if the stream of data is using a datagram 'pretending' to be web-page traffic, but contains entirely proprietary game-data/etc. Some of the monitoring and traffic-shaping elements might (if a firewall doesn't like the 'gobbledygook 'pretending to be a web-request), or you're crossing a router that 'knows' to NAT web-traffic off into the web-server's rather particular subnet, or maybe it's blocked http and will now only ever accept https (:443?). But most end-users aren't hidden behind systems that don't have one or more 'allowed' sockets to more easily be connected.

Or you could even make it 'properly' formed (e.g. XML data as an actual HTTP(S) return from a suitable 'page request' to a suitable faux-webserver (or actual webserver, with backend extensions).


There's all kinds of ways I might try this (or even have, but long ago, and may no longer work as well as they used to...), but time marches on. What might have gotten around the NAT of times past might not do so these days (and what didn't used to might well do so quite nicely). But torrenting and all kinds of other P2Ping became a big thing, over the intervening years, so whatever technomagery they used must be a valid approach. And probably overkill. Probably all you need is some variation (if even required...) upon the DynamicDNS approach if you just want everyone to chat back and forth with you. And you having carefully ensured that you're open enough (but proofed against spoofs), without needing your client endpoints to have paid any mind to their own router configs (and ISP policies) under most circumstances.


It all sounds intellectually interesting, though, and if I wasn't currently on the wrong end of a mobile connection right now, then I might even have poked and prodded at my own router (and a handy external server I can convince to engage in meaningful conversation) to check exactly what is problematically locked down and what is problematically far too open. (I've only minimally fiddled with it, previously, lazily accepting it as 'working', for most part, each time the ISP has deigned to deliver me a new modem/router for whatever reasons they gave at the time.)The (re)learning is half the fun, of course.



[1] Or a series of fallbacks... In the early days of the Web (very early!), the access to ports less than... 512? ....or maybe 256/1024/whatever... were restricted use, inward and outward, so it was useful that port :80 for http could also be used for requests/server-listening by :8000 or :8080 (and others), at least until it became more than just a homebrewed thing and allowed just like FTP'S :20/:21, etc, but that would have been later. But you only rarely had to have odd numbers like ":1234" in your URI (when's the last time you specified the port in your address-bar... I bet most people never realise it's an option. ) because it normally would try the several handshakes silently in the background.

[eerr]

hmmm, i've been looking into new computers.
Normally I choose a fairly cheap computer.
but all the low-tier computers on the market seem to have 16 gb ram, and that isn't any better than what I've got now.
what gives?

[dragdeler]

For a laptop it's going to be tough to find more, at a good price, I'd target whatever dedicated gpu I could afford, as a price reference, and pay extra attention to laptops that come with additional empty ram, nvme and sata bays. There won't be lots but they exist.


You'd think with economy of scale you shouldn't have to bother.... And if you can find pricetrackers that crawl the relevant stores, you can find killer sales. But usually laptop maufacturers overcharge when upgrading their models for you.

[wierd]

The issue at hand, is the consumer's love affair with "under 3lbs" laptops.

Lots of things have to be sacrificed for that, and -- since you are sacrificing things, why not sacrifice other nice things too?

Examples include bottoms held on with freaking glue instead of screws, batteries that are never meant to come out and be replaced, (and since you should never take it apart, dear citizen, why would you want to put more RAM in? For this reason, and to give you SUCH SAVINGS, we have omitted the SODIMM slots!) etc.

The next time somebody PRAISES Apple's "Engineering" as "Brave!" and "Innovative!", please remind them how the trajectory they have placed the market has resulted in noteworthy enshitification.

Almost all modern laptops are just an intel or AMD CPU, with the on-die graphics, and some glued on DRAM on the mainboard.  Practically clones of each other. It is getting very very had to find one with a real GPU inside it, let alone more than one M.2 /NGFF SSD slot, or SODIMM sockets. 


I suspect it will only get worse, until general consumers at large raise a stink about it. 

Given the way that Apple and Co have brainwashed people into the White Plastic Dystopian Hells though, I have very strong reservations and doubts of that ever happening.

[dragdeler]

M'yeah, stuff isn't getting much better in general. It has been since 4th gen intel i that we could replace a cpu in a laptop. If you want something durable I'd avoid laptops alltogether. Also allways dell and hp out ofprinciple. I've seen hp ramsticks with a tapered bottom instead of straight line for the pins. If it needs to be portable these days one could do fun builds in a suitcase, and be competitive with gaing laptop prices.

On the bright site, stuff like the steamdeck and gpd mini computers are getting kind of cool. Suire it's an APU but these days, you tend to be able to get more use out of that. They're (x86) computers, they're rugged and they can be had for less than 500 bucks. I really can't justify buying one... I built a stupid thing myself with a rockpi x a few years back. It sucks but it costed about as much as a steamdeck would cost me now.

[Starver]

Yeah, my perpetual advice remains "only get a laptop if the portability of a laptop is what you need". While eerr doesn't explicitly say it's a laptop, though, it sounds like that's what we're needing to assume.

Desktops/under-desks are generally going to be better bang for your buck, though, and if you still feel they're less endowed with RAM than what you like then you'll be fairly unlucky/careless to have it not able to take more that you get yourself out of the "laptop premium" that you didn't pay, as well as other upgrade options.

The processor (and, by extension, the mobo) is the one thing perhaps not to skimp on 'in the hope' that you'll upgrade it, as any such upgrade will entail removing the original (unless you have a hand-me-down use/buyer for it, large lost value), wereas RAM (at least the first time) is an additive improvement to what's already there (vaguaries of slots allowing), and even HDDs can (usually) be added as further space (usually separate drive-letter, unless you go a bit technical or it's a non-drive-letter OS where you can remount and relink things accordingly). Even adding a better video card (than the existing one, or the mobo-inbuilt video) gives you the possibility of combined options (multi-monitor desktops!) rather than necessarily losing the original investment/letting it sit idle - and with GPUs these days being exploited as 'coprocessors' in some interesting ways, even a non-displaying 'original' video card might possibly be useful past the point of any replacement (but knowing what slots, and power, you have available is probably key to this sort of advanced plan).


For specifics, I'm woefully out of date, and I don't know what options (and prices) you'll have to deal with, but... assuming you can avoid laptops (and perhaps even if you can't/won't), research opinions and stats on the CPUs for better one(or  that's not dangerously close to being a dinosaur) and then see if it's possible (including physically...) to bundle in anything like RAM upgrades either at time of purchase or as a near-future option. Specific other needs, e.g. video cards sufficient for a desired game release, would also be aims to take account of, but I know what my priorities are.

(For one thing, I'm no longer using my old 29-year-old computer,  but the one 25 or 26 years 'young', with the 'newer' 24yo OS installed on it perhaps 20 years back is still running nicely, for the purposes I use it, and I only hope the machine bought last year (to replace one a mere 15 years of age that (breaking my 'rule', being a laptop),started to fail in series of particularly hard to repair individual ways) will provide as much service... Heck, it could easily outlast me, if it does.)

edited to correct failed closing tag...

[eerr]

hmmm if computers aren't really getting significantly better, I might want to address the issue on my current computer.

I've had the current browser tab reset entirely just from watching youtube/twitch.

I've had plenty of games freeze up, including games that are extremely low spec, like one originally designed to run on phones.

I've had twitch repeatedly refresh on some streams and not others. (which I assumed wasn't an issue with the computer itself)

I've had constant issues playing runescape, with instability included. (sadly not the beacon of 'low spec gaming' that it used to be.)


With issues like that, I'm fairly sure i need to replace the computer.

[Starver]

Could just be the "too much electronic krud" problem. In the old days, often deleting thousands (or tens/hundreds of thousands) zero-byte-length files under C\WINDOWS\TEMP would stop some programs slowing themselves down[1]. Or even going down the line of disabling virtual memory, reboot 'clean', full defrag, re-enable the virtual memory.

But more modern OSes are both much better at not getting messy and storing cached krud in all kinds of weirder and less obvious spaces. (I've seen Win11, sitting doing essentially nothing, report more than 100GB of free space[2] one minute and then the next complaining that it's running out of disk-space, and the check shows we are indeed down to 1.5GB or so.

There's plenty of "WinCleaner"-type products, which might have actual tried-and-tested methods for overcoming both Windows' own garbage-cleaning oversights and other commonly offending applications. But, honestly, I don't trust any of them. And some are closer than a few steps away from being trojans/malware/spyware, if you're unlucky, and 'merely' horrible pesterware if you are.


That and (from my experience with using decades-old systems, as above) the possibility that corruption has crept into some obscure linked-library/resource which mostly doesn't matter but occasionally raises an exception. That's not unknown. Sometimes easy to solve (and, first, to identify), other times not so.


[1] "I'm thinking of storing temp data in PROGNAM000000.TMP... No, that exists. I'm thinking of storing temp data in PROGNAM000001.TMP... No, that exists. I'm thinking of storing temp data in PROGNAM000002.TMP... No, that exists. [Skip a few iterations] I'm thinking of storing temp data in PROGNAM083629.TMP... Yes, at last... Ok. Restart the real-time screen-memory writing!"

[2] Which is perhaps less than I'd suggest you regularly leave, these days, but I probably push it more than I should if I'm collating data.  Noting that the collation is often over/not started at the point it fluctuates by several 10s of GB, in the above tale... And while it also seems to happen prior to announcing a wish to autoinstall Windows Updates on the next reboot[3], it also happens when it's not even connected to any network for a week or so.

[3] Which makes one wonder how or if it's managing to accumulate approaching 100GB of "pre-update" preparatory files. I presume there's unpacking and making strategic copies of the current setup for possible fall-back.

[Flying Teasets]

Dumb question: Can Microsoft Copilot be permanently disabled in Windows 11 Home Edition?

[Starver]

Good question. I haven't got 'my' W11 in front of me, right now, but when the "Copilot (Preview)" popped up on it without being asked I found various instructions on how to banish it from any obvious presence. Like Personalisation>Taskbar items turning it off, as per Widgets, and I think some regeditting to stop it even trying to start up in the background on Startup (with more success than with Microsoft Teams[1].

Do you just not want it cluttering your screen (and possibly responding to accidental activating inputs), or do you want to make absolutely sure that it's not running silently in the background with no obvious reason? From what I half remember, I think it's either unable to be completely uninstalled or tends to get 'updated' back in again if you do (which may undo any prior 'just don't bother me' setting changes), the very next time you get a network connection. But it was yet to get beyond the introductory Preview version when I checked what was and wasn't possible to do with/about it. This may have changed.

[1] Unfortunately required every now and then, but seems to be in some super-secret "autorun"-like element in the Windows startup process that I can't ellide.

[Akura]

I should probably ask this on the Steam forums, but I think it'd be pretty quickly buried before it got properly answered.

I have an issue where achievements don't pop up in game anymore. Rather, whenever I turn on my VIVE headset, every achievement from every game I've unlocked since the last time I used VR will start popping even before the VR environment finishes loading. That is, non-VR games don't have the achievements pop up until I use the headset, VR game achievements do appear in game. Well, maybe, it's been a while since I've played a VR game other than Blade & Sorcery which doesn't have achievements. I think restarting my PC halts this issue, at least until the next time I use my headset.