networking

[rocky badge]

@peak cloak @clear igloo

Bruh the resnet isp is now doing 11.X ips on WiFi too

[wheat flicker]

anyone knows that these numbers mean on fast.com?

[full monolith]

I guess its the data usage of down and upstream test

Like how much data it will use when its testing down and upload

[wheat flicker]

Oh that makes sense yeah

[thick minnow]

does anyone know anything about miracasting

I just googled it. Interesting, but never heard of it before

i was wondering if someone could help me with it because the internet is somehow not helping

[drifting raven]

Whatchu need

[thorny osprey]

i dont live in such techly advanced country so lol

[thick minnow]

its saying i have it available for my pc right but it wont work

i updated drivers, cleaned already updated drivers i did all fixes i could find but wont work

[drifting raven]

What are you trying to do?

I'm asuming connecting an external display using miracast?

[thick minnow]

wanting to connect my pc to my tv to watch a movie

my tv is roku

[drifting raven]

Right. What's your network situation like? All WiFi, partially wired, multiple routers or AP's?

[thick minnow]

roku is wireless and pc is wired, now thats what i think the problem is because i dont have wifi on my pc

but people said on microsoft forums that it shouldnt really be an issue

[drifting raven]

Shouldn't be a problem, unless you're doing something weird like having a dedicated router just for WiFi (in router mode)

Have you tried something like AirDroid to check if the casting actually works properly?

[thick minnow]

no, just simple wifi pod extender thingys

whats an airdroid?

[drifting raven]

App for your PC or smartphone

[thick minnow]

i can try rn real quick

[drifting raven]

Also, do you have any custom firewall things set in your router/modem?

Oh hang on, what kind of roku device do you have? Seems like not every roku supports screen casting/mirroring

https://support.roku.com/article/208754928

[thick minnow]

mine does, its just i cant get it to work on my pc

im testing the airdroid to see if its even working at all

[drifting raven]

You could also try it with your smartphone's native screen casting ability (asuming you have an android based phone)

[thick minnow]

ok it works

the airdroid so im assuming then something is messed up with the casting ability on my pc

[drifting raven]

Where have you tried this? Smartphone or PC?

[thick minnow]

pc to smartphone

wirelessly and it just auto connected

[drifting raven]

Probably not related to network issues then

Windows being windows?

[thick minnow]

could be

i might try my laptop and call it on my pc but its just annoying since i cant figure it ou t

[drifting raven]

If you try it on your laptop, do it via a wired connection so you create the same circumstances

[thorny osprey]

https://www.startech.com.bd/tenda-ac23-2033mbps-ac2100-router
is this good?

[peak cloak]

@thick minnow do you have only 1 router

[thick minnow]

yes

[peak cloak]

Hmm then idk

As long as it's on the same L2 network it should work

[drifting raven]

Lightbulb moment! Do you have client isolation enabled on your WiFi by any chance? This causes issues in tech like this. It can still announce via multicast, but can't actually receive any data

[sage crow]

Tbh I’d look at Amazon reviews of this router

[thorny osprey]

amazon not available in the country im in

[untold sierra]

Hey all, I'm looking for a rather specific device for a specific use case.
So basically what I need it a NAS in the form of a USB stick. I need 10MB of capacity and it needs to run like a plugNplay usb stick on the host device, but integrate into the network like a regular NAS.
Its use case would be transfering CNC code files to a CNC router because I'm too lazy to keep running around with a USB stick
If this doesn't exist, a regular NAS with PSU could work too, but the stick solution would be cooler since I need essentially no capacity

[peak cloak]

So a NAS with DAS functionality (can also access via USB)?

I don't know if such thing exists due to the difference in how the 2 work

[untold sierra]

Direct Acess Storage, I suppose? Yes

It seems to exist but 3.5" is definitely not the right form factor for this. I have a bunch of 2.5" ssds spare tho, which I could use

Oh and there is no possibility to plug the NAS into the network, it would need to be wireless even when it's not in the form of a stick

[thick minnow]

hm would i change this through my routers settings in the browser?

[drifting raven]

Yeah, depends on the router though

[hollow marlin]

The amount of millions of $ worth of IP space (that is also wasted) that your uni has is 🤮

[clear igloo]

Waste all the IPs!
You get a /64 and you get a /64 and YOU get a /64
Everyone gets a /64!!!!

[hollow marlin]

Nah, everyone gets a /48 because IPv6 is of course "unlimited"

[clear igloo]

I was trying to save some space 😦

Every P2P gets a /64 though right?

[hollow marlin]

Oh yeah definitely!

[drowsy fossil]

i would be shocked if that wasnt nat tho?

[hollow marlin]

It's not. They own multiple /8s and that's what's handed out

[drowsy fossil]

thats quite messed up

[crisp thorn]

my isp gives you few thousand ipv6 ips

[drowsy fossil]

so they own 11 and 26?

[crisp thorn]

I dont like ipv6 addressing

[unkempt thicket]

can relate

[hollow marlin]

If going of just their AS...No, quite a bit more than the two

7.0.0.0/8          *[BGP/170] 5w2d 19:46:02, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified
11.0.0.0/8         *[BGP/170] 5w2d 19:46:01, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified
21.0.0.0/8         *[BGP/170] 5w2d 19:45:58, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified
22.0.0.0/8         *[BGP/170] 5w2d 19:45:58, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified
26.0.0.0/8         *[BGP/170] 5w2d 19:45:52, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified
28.0.0.0/8         *[BGP/170] 5w2d 19:45:52, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified
29.0.0.0/8         *[BGP/170] 5w2d 19:45:52, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified
30.0.0.0/8         *[BGP/170] 5w2d 19:45:52, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified
33.0.0.0/8         *[BGP/170] 5w2d 19:45:51, localpref 100
                      AS path: --- 1299 3356 749 I, validation-state: unverified

[drowsy fossil]

😱

now why on earth do they need 144 million ip addresses

[hollow marlin]

The "internet" in the early days was primarily for universities (then funded by the gov.). IPv4 blocks were handed out upon request via email and recorded in a notepad. This was prior to and RIR to oversee allocation.

So you have Unv's with multiple /8 that they own the rights to and cannot be reclaimed legally.

Once it was realized that handed them out was going to be a problem, then the RIRs were created

[drowsy fossil]

well at least thats not a problem on ipv6 probably

[crisp thorn]

well that depends on the country I guess

nothing stops a country from going rouge and reclaiming the IP routing it to its own server

[drowsy fossil]

yea but they can only do that within their own borders

[crisp thorn]

yeah

[hollow marlin]

Not yet anyway. IPv6 may seem near infinite, but we are closing in on the hardware bubble. Right now there is only so much memory in hardware to hold a few millions prefixes

[crisp thorn]

wdym

[drowsy fossil]

ah so we may end up needing a nat like "compression" algorithm to make routing easier?

[crisp thorn]

you wont hold addresses to all other devices on network on your own device

and for google,cloudflare its not a challenge

[drowsy fossil]

and actually for ipv6, the router technically doesnt neeeeeeed to store the ips right?

[crisp thorn]

ipv6 went too far with the 128bits imo

64bits are enough

[drowsy fossil]

nah
most computers can operate on quads relatively quickly

and it only doubles the storage and memory requirements if you dont use more addresses

[crisp thorn]

you get 18,446,744,073,709,551,615 addresses with 64bits

and 16 bytes for an address is a lot

ipv6 is wasting your ram and will waste ISP's resources too

ipv6 was made in 1995 and still almost no one uses it

[drowsy fossil]

except on their phones

[crisp thorn]

?

I still get ipv4 assigned on my phone from GSM

[drowsy fossil]

most cell networks use 4on6

[hollow marlin]

In the scope of things, routing in hardware (ASIC) is more recent in the timeline (like mid 2000s). This is major if you want to route at line-rate. However, memory in these ASICs are expensive and can only scale so far and getting worse as we're pushing 400G+.

Most service provider routers fall within the mid-range level which can hold a total of a few million IPv4/IPv6 routes. Because IPv6 is around 2-3 times v4 when in memory, each prefix added to the global table is contributing to the maximum number of prefixes in memory.

If IPv6 continues to be allocated without care and not being aggregated. We will be reaching a point where a large portion of SP routers just don't have the memory. There are plenty though that easily have the memory for 16+ millions routes

[crisp thorn]

ipv6 is 4 times bigger than ipv4 in byte size

[drowsy fossil]

so it only requires 4x the memory despite being able to address every atom of everyones body on the planet

[hollow marlin]

Correct, but vendors optimized it to reduce memory size by 2-3 times actual prefix size

[crisp thorn]

they have to use normal ipv4 too because there are many services that dont even accept ipv6 connections

[crisp thorn]

4x the memory is not only

its a lot

[drowsy fossil]

its alot but it will matter as the number of devices per person balloons

[crisp thorn]

you need ~17GB to store all ipv4 addresses pure

[drowsy fossil]

except thats not true

Nat has alooooooot of ips

[crisp thorn]

ipv4 can have only 4,294,967,296 addresses x 4 bytes because its a 32bit

you get 17 179 869 184 bytes

[hollow marlin]

I am referring to strictly to routes. When talking about host within subnet, then that's dealing with ND which is and never will be able to store and entire /64 worth of ND tables

[crisp thorn]

if you were to store all ipv6 addresses the amount of memory needed is incomprehensible

[drowsy fossil]

again tho we have plenty of ip addresses that get reused

[crisp thorn]

you would need 5.4445179e+39 bytes to store all of them

[drowsy fossil]

sure but not every single router needs to keep track of everyone elses subnets

[crisp thorn]

that is the same number of possible combinations of 12 word seed phrase

combined from 2048 words

[drowsy fossil]

again though, not all of it needs to be stored in memory
just like how your computer doesnt need to know my computers exact ip address

[crisp thorn]

my computer doesnt

others do

[drowsy fossil]

my computer technically requires 32+16 bits of data to specifically address

[crisp thorn]

if ipv6 will be given away like free candy like it is rn

I wish it good luck

ISP's allocate thousands of ips for you even if you use only one device at home

[hollow marlin]

To give you a perspective on optimization and use of composite-next-hops, this is Juniper MX that is part of my lab at work. 4GB memory, full IPv4/IPv6 tables or 2m/340k in RE and 900k/170k in PFE

Routing Engine status:
  Slot 0:
    Current state                  Master
    Election priority              Master (default)
    Temperature                 43 degrees C / 109 degrees F
    CPU temperature             47 degrees C / 116 degrees F
    DRAM                      3968 MB (4096 MB installed)
    Memory utilization          84 percent

> show route summary
inet.0: 920452 destinations, 2109513 routes (906783 active, 0 holddown, 298639 hidden)
              Direct:     41 routes,     41 active
               Local:     39 routes,     39 active
                OSPF:   2626 routes,   2626 active
                 BGP: 2106785 routes, 904075 active
              Static:      1 routes,      1 active
                RSVP:      8 routes,      0 active
                 LDP:     13 routes,      1 active

inet6.0: 170805 destinations, 341549 routes (170805 active, 0 holddown, 0 hidden)
              Direct:     19 routes,     10 active
               Local:     17 routes,     17 active
               OSPF3:     41 routes,     41 active
                 BGP: 341471 routes, 170736 active
              Static:      1 routes,      1 active

[crisp thorn]

then there comes the problem that ipv6 is too hard to remember or quickly write down unlike ipv4

[drowsy fossil]

again tho i dont need to keep track of it, i just need to pass the data to the end device

[crisp thorn]

and then malicious users

you wont be able to just create a blacklist of ip's

if you block one spammer they just change ip instantly gl keeping up with that

[drowsy fossil]

now that will be a bit of a problem

but thats already true now

[crisp thorn]

with ipv4 its manageable

[hollow marlin]

The problem with memory that I am talking about is on the global routing table, not at the host or customer level.

[crisp thorn]

there are services that you send a request to api and get if the ip is malicious

or you can generally block the ip on your own device

with ipv6 that will be obsolete

unless the amount of IP's assigned per person will be standardized so we know which exact ip's are the same user just by knowing one of them

then we are screwed when we go to IPv6

[drowsy fossil]

sure but quadrupling the cost of all the core routers is better than using nat

[crisp thorn]

because one isp gives you few million other thousands etc etc

with ipv4 getting even 1k addresses is hard to do

it would cost shitton of money for all the vpns

but with ipv6 you just get it for free lol

[drowsy fossil]

well not exactly

you cant exactly slac outside of your subnet

[crisp thorn]

my isp gives you thousands of ip's and they are all dynamic

they all keep changing so you can actually never reuse the same ip

I dont use ipv6 tho and have it completely disabled on every device I have

[drowsy fossil]

yes but they are all from the same subnet

[hollow marlin]

It not just cost. Our larger MXes are already a few million, whats another million on top.

The problem is continuing to add memory to these large chips while keeping them at line-rate. Not including the standard deployment cycle being 10 years, you cannot just keep purchasing millions worth of equipment because engineers are being careless in the advertisements

[crisp thorn]

they are just from the pool that ISP has

[drowsy fossil]

also your isp probably doesnt specify them, it just gives you a range and your devices slac within them

[crisp thorn]

well its good that we dont need ipv6 yet

[drowsy fossil]

now thats fair, however there are many very interesting solutions for that, and we need to get started on it asap

[crisp thorn]

I wish people still behaved like in the early days of computers when ram was worth as much as gold

everything was made to use as little memory as possible and be conservative

[hollow marlin]

This perspective of IPv6 not being as infinite as expected has already been discussed and being pushed in the IETF. The respected leaders such as Russ White and Geoff Huston are already pushing it but many other don't see it as an issue

[peak cloak]

That mentality makes v6 deployments worse

[crisp thorn]

Good

let it die

[peak cloak]

...

[drowsy fossil]

just think of ipv6 like double nat with none of the downsides besides still requiring routers

[hollow marlin]

Here are a few articles that were referenced in APNIC's Ping podcast discussing this if you're interested:
https://blog.apnic.net/2021/03/03/what-will-happen-when-the-routing-table-hits-1024k/
https://blog.apnic.net/2023/01/06/bgp-in-2022-the-routing-table/#:~:text=The growth of the IPv6,of all prefixes are %2F48s.

This is what opened many peoples eyes to IPv6's careless deployment and my main points.

[drowsy fossil]

its true that nat does explicitly prevent you from having these sorts of problems

you still have the problem however that besides these super reserved ipv4s, everything eventually will have to be hidden behind nat

[hollow marlin]

IPv6 was standardized in 1995 at a time where memory was in MB and routing in hardware was thought of as impossible. But they designed it after seeing the growth of v4 and wanted to avoid the same problem so it was overengineered.

We need IPv6 along side IPv4 just to keep progress moving. I have my own issues with IPv6 and it's still fragmented in many ways, but I would not say we don't need it.

[peak cloak]

Android DHCPv6 is so bs

[hollow marlin]

The largest issue with v6 is the OSes. It's super fragmented and in many cases unpredictable. Networking vendor may have much to incorporate, but at least the processes are mostly standardized

[peak cloak]

Yeah

[drowsy fossil]

yeop

[rocky badge]

Uni only has a /16

They don’t own any in 11.X

It’s the awful company they contracted residential networking out to that’s using 11.X

[hollow marlin]

Ah, I thought that was also your Uni.

[rocky badge]

Nopee

[thorny osprey]

im looking for a good gigabit router
what router do yall recommend? and what price should i range on for a average quality? (im on a budget too lol)

[sage crow]

Look here:

https://www.gl-inet.com

They have cheap gigabit routers with loads of features and plugins you can download

[thorny osprey]

are they good?
like reliable and will last long

jeez its still kinda expensive (120$)

[sage crow]

Look on their website their cheaper lol

[thorny osprey]

99%
or 10000TK

[sage crow]

I’ve had the Slate Plus for over a year with no issues

[thorny osprey]

i live in bangladesh

[sage crow]

Ah

For the features and specs you get it’s still a pretty good price

[thorny osprey]

well ye but
is there a retail for like irl buying

[sage crow]

Not that I know of

Could always just use an old PC as a router install pfsense and buy a cheap wifi access point & switch to give wifi and internet to multiple devices

[peak cloak]

@thorny osprey I would look at local stores

and see what you can find

[rocky badge]

@clear igloo

PoE Texas splitter to PoE power this PTZ controller

I am happy now lol

[clear igloo]

😄 PoE all the things!

[drowsy fossil]

heck ye!

[clear igloo]

IPv6 won't die, no matter how back asswards your thinking wishes it to

[rocky badge]

Its so STUPID this doesn't do this built in

[clear igloo]

Yah, there are a lot of things that should have PoE but remember, that's a few cents they can't keep for themselves by adding it

[rocky badge]

The newer ones support it

[clear igloo]

Ah, so it's a legacy model then?

[rocky badge]

technically

[clear igloo]

"legacy"

not the newest shiny off the assembly line, lol

[rocky badge]

https://na.panasonic.com/us/panasonic-professional-ptz-camera-control/selecting-controller-your-ptz-camera#:~:text=Comparing PTZ Camera Controllers

this is the AW-RP50

the newer version is the RP60

[crisp thorn]

i dont care if it dies, I just dont like it and wont use it until it will be mandatory

[clear igloo]

It shall become mandatory then

[crisp thorn]

which it probably wont, I assume something better will be made when we run out of space

[clear igloo]

I shall make it so with the introduction of IPv8

256-bit alphanumeric addresses for everyone!!

IPv6 is already the norm in a lot of countries. In Japan, Rakutan is IPv6 only in it's infra even, it's just slow in the west because there are more IPv4 addresses still around

[crisp thorn]

sucks for them not being able to play some games and use nice services

[clear igloo]

NAT64 still exists

as does CG-NAT, being IPv6 only doesn't stop you from accessing IPv4 only services if done properly

[crisp thorn]

some ps4 and steam games wont work with ipv6 at all, same with smart tvs

some HP printers wont work with ipv6

[clear igloo]

@rocky badge You have IPv6 yet on your home connection?

[rocky badge]

Just because you don't get an IPv4 external address doesn't mean you can't use IPv4 internally

[rocky badge]

ISP is slowly rolling it out which I hate but

[crisp thorn]

the point is they wont work if you dont have an external ipv4

[drowsy fossil]

thats an infrastructure problem on the server side

[crisp thorn]

but its problematic for user

people at HP couldnt care less you cant use your printer because of ipv6

buy a newer one that can

[drowsy fossil]

which is why we need to push to transition to it asap so that noone makes any of the new legacy bs

[drowsy fossil]

theres nothing preventing you from setting up 4on6 on your router

[rocky badge]

NAT64 rewrites & maps

that's the whole point

[crisp thorn]

many ISP given routers dont allow you to choose settings like that

[drowsy fossil]

which is why we need to force them to transition asap
at some point it will be forced for technical reasons, we want them to transition before that

[rocky badge]

@clear igloo I thought WiFi 6 required WPA3

or was it 6E

[drowsy fossil]

6 definetly doesnt

seems that e requires it ye

[rocky badge]

I love how the entire campus is WiFi 6

[drowsy fossil]

very sweet

[rocky badge]

They did the final wave of AP replacements over Spring break

[drowsy fossil]

got our second site up on wifi 6 out of just under 100

:/

[rocky badge]

for this building they do 20MHz only

in the academic buildings they do 40MHz wide 5GHz

PHY speed is like 1037Mbps

[drowsy fossil]

yeee

[rocky badge]

now for the university to get more than 20Gbps of connectivity to the internet lol

[clear igloo]

6E

[peak cloak]

That's an issue with vendor support we were discussing before not with the protocol fundamentally

[potent radish]

I'm trying to setup wol on 2 computers
First is optiplex 3020
Second is my main pc with msi mobo

First had wol enabled and working some time ago but it's not working since i connected it through other switch and then connected it back normally
System shows that wol is enabled on magic packet (after enabling it on every boot)
But ethernet link led is off after i turn off pc
And it does nothing after i try to turn it on using any wol client
Wol is enabled in bios and system

Second one have similar setup and while pc is off, 100M link LED is blinking
But ... port LED on mobo is somehow off
And it's not starting after using wol

[potent radish]

Is this even possible?

[fallow wing]

hi there
is anyone here familiar/regular with TrueNAS ?

[drowsy fossil]

noone is familiar enough to know what your problem is without you asking about it

[potent radish]

^ true

[fallow wing]

I don't have a problem, I'm making my own file server and wanted to ask if there was "bullet points" of stuff to know about that you don't find on the popular reviews or info like that

it's my first time so I'm not aware of what to look out for at all

[velvet jolt]

Does anybody know why this Ubuntu server static IP file isn't working?

[peak cloak]

Isn't working in what way?

Isn't applying? No internet access?

[velvet jolt]

It's giving error claiming inconsistent indentation right now

[pseudo blade]

Well I can see "addresses" is one space too far in for valid yaml straight up

[pseudo blade]

Ok yeah had a look at netplan's structure and neither are correct

Nameservers, gateway4 and addresses are keys on the interface

Blindly typing into Discord on my phone, something more like this:

network:
  version: 2
  ethernets:
    enp3s0:
      dhcp: no
      addresses: [10.0.0.231/22]
      gateway4: [10.0.0.1]
      nameservers:
        addresses: [10.0.0.22, 10.0.0.21]

Can't speak for the accuracy of the addresses as that's obviously on you

idk if "renderer" is a mandatory key or not

[velvet jolt]

Is it 2 more indentations when it indents more?

[pseudo blade]

2 is standard, any works as long as it's consistent

[pseudo blade]

If so copy the renderer specified in your distro's default config

[velvet jolt]

It was dhcp so it didn't have any other lines other than the dhcp yes

I mean below

[pseudo blade]

It'd be above actually, goes in the network object

I mean you can structure it to be below if you wanted but it probably wasn't

[peak cloak]

Very possible LTT got breached

Doesn't surprise me tbh

It may not be just their yt channel

[remote dirge]

it got breached?

[drifting saffron]

@peak cloak if you were to look at #public-chat, #tech-chat-1 or the subreddit in the last 2 hours you'd know that they got pwned by crypto scammers

[peak cloak]

Yeah but it could be a bigger thing than just the channel

[drifting saffron]

main channel is already banned by YT, Techlinked and Techquickie are currently streaming the crypto scam

so they'll go down soon too

[peak cloak]

Maybe got access to internal systems like pw vault if they use one

Which would be a really messy situation

[drifting saffron]

Lastpass had a massive breach a few months back, could be spoofed sponsor link, could be a backdoor in some YT extension for content creators

it's not the first time scammers take over a YT channel and start broadcasting crypto chats with Elon while removing old videos. What is fairly new is reuploading old vids with links to the scam

[waxen scroll]

https://www.youtube.com/user/LinusTechTips

@clear igloo

[waxen scroll]

https://www.reddit.com/r/LinusTechTips/comments/11zftc5/welp/?utm_source=share&utm_medium=ios_app&utm_name=ioscss&utm_content=1&utm_term=1

[snow elk]

That would be a PITA to recover from

1: SSL certs

2: very possible that acc/pw combinations were already used

But I don't think that's the case, they should have a strict access policy

[magic arrow]

Does anyone have a good place online when it comes to ordering a 9U server rack in the US?

[glass anvil]

Anywhere really. TrippLite is a solid brand now owned by Eaton.

[merry elbow]

Linksys Question - Can I add a wifi 6 child node (MR7350) to a MR8300 (wifi 5) parent/main router? With the Linksys mesh app (which is sorely lacking) It appears i can create a separate network with the MR7350, or use it as a bridge with a second SSID. Also looking at the MX4200 (White velop wifi 6 node). Thanks for any insights

[rocky badge]

@waxen scroll @hollow marlin Well....University seems to be doing 802.1X on wired now...

Which means we need to contact networking to get a MAC bypass for several devices.

[narrow gate]

Or you can just install the certificates on those devices

[rocky badge]

We can't

[narrow gate]

why not?

[rocky badge]

because they're not devices like that. One's a PTZ controller, several embedded stuff, etc.

[narrow gate]

802.1x isn't exactly now. Why should they not support it? It supports IP

[rocky badge]

because there's no way to set it up?

not exactly gonna be installing 802.1X certificates on this

[narrow gate]

I mean, still don't see why not. If it communicates via IP, surly it does more?

But that device might not have support for that

[rocky badge]

It does not support 802.1X

[peak cloak]

just because it's in IP spec doesn't mean that the device itself supports that feature

or implements it correctly

[thick minnow]

what's the maximum bandwidth of 5ghz? I had to plug the ethernet because downloading a game was slow

[peak cloak]

Depends

But Ethernet is always more consistent and reliable

In a perfect environment you can get over gigabit, but no environment is perfect. You have noise, interference, etc.

[thick minnow]

it was downloading at 35mbits more or less, with peaks of 120mbits

with ethernet I was getting triple the speed

[peak cloak]

Makes sense

[fallen totem]

I’m running cat6 in my house for PoE APs. Is now a bad time to buy Wi-Fi 5 APs when Wi-Fi 6 and 6e are out?

[peak cloak]

Depends, are you on a strict budget?

Personally I would just jump to 6

[obtuse dragon]

Depends on your clients. No reason to spend a ton of money on wifi6(e) APs if your clients can't utilize them. This table shows the differences in speed/frequency/standard. https://en.wikipedia.org/wiki/IEEE_802.11#Generations

[fallen totem]

Looks like I’m not on a 6e budget but I could probably swing for the 6

[fallen totem]

good point. Also thinking about the future and new phones and tablets in the next couple of years might have 6e

[peak cloak]

I mean it also depends on what your situation is. Wifi 6 is really an improvement in congested areas

And what your internet connection is

[fallen totem]

yeah. I do have gigabit fiber

[obtuse dragon]

Until everyone defaults to 320mhz and we end up with 5ghz all over again 😦

[fallen totem]

As far as clients go, really just 3 devices at a time high bandwidth. And lots of IoT

[peak cloak]

Well the actual specification is better

On how it handles lots of devices

[cedar cloak]

Curious, what does everyone prefer openwrt linux or opnsese bsd? I'd do vyos but that seems like a lot of effort tbh dunno if it would be worth it

[peak cloak]

for what exactly?

[cedar cloak]

just a home network, have a small x86 machine laying around with 4x i225 nics lol thought process was

xfinity xb7 in bridge mode -> small computer router -> unmanaged switch -> other devices/asus router in ap mode - pretty basic needs honestly, adguard home dns and codel/qos for latency lol

[manic nova]

Anyone ever seen a massive fluctuation in internet speed like this? I ran 3 speed tests back to back and got 65mbps the first time, 12 the second, and 40 the third. Real life reflects that pattern

[sullen olive]

Recommendations for a budget network switch

[pseudo blade]

Can you be more specific as to what it should be capable of? Port counts, speeds, management...

[sullen olive]

4+ ports 1gig and not trash

[waxen saddle]

Just about any switch will work just fine for you then.

[pseudo blade]

I suggest an RB260GS or - if you can resist the urge to plug it into itself - whatever you can find in your local tech or office supply store lol

[sullen olive]

Andddd it doesn't ship to canada

[pseudo blade]

Basic desktop gigabit switches are all much alike unless you're buying a managed one like the probably unnecessary RB260GS

They're available in Canada, you buy from resellers

[sullen olive]

80 bucks

[pseudo blade]

Buy an unmanaged one then

Or find a reseller that isn't ripping you off

[sullen olive]

Looks like a unmanged one is 15 bucks

Hmmm 15 bucks or 80

[pseudo blade]

Well there you go

I wouldn't buy a 6-port switchOS switch for $80 either

[sullen olive]

Now I'm wondering metal case for 3 bucks more or plastic

[sullen olive]

Well I lied I thought the plan was to run part off the router part from the switch

Were running it all off the switch

I need like 10-15 ports

Also whats managed vs unmanaged

[tawny temple]

managed gives you network configuration options to tweak. on unmanaged, you just plug your cables and let it do its thang

[peak cloak]

managed allows you to configure features such as vlans, STP, and more

unmanaged doesn't allow for such features

most people do not need a managed switch

[sullen olive]

But I want it now

[peak cloak]

you prob do not

most features are useless without other network devices supporting them, and that you configure your network in a way that can utilize those features. Other than that it's just more money for nothing

[sullen olive]

Its only 10 bucks

[peak cloak]

which one

[sullen olive]

https://www.amazon.ca/TP-Link-Gigabit-Ethernet-Network-TL-SG116E/dp/B07GRG63P6/ref=asc_df_B07GRG63P6/? https://a.co/d/j0WMmb9

[peak cloak]

I guess it's fine. Don't know much about tplink managed switches

[sullen olive]

also i just wanna fuck around with tech

[sage crow]

I've got one they're pretty solid I'd say just as good as any other brand of consumer managed switches

[bold drum]

bah I need to find a 24x2.5" chassis

[thick minnow]

i can vouch for there normal switches

[sullen olive]

do they switch?

[thick minnow]

yes

and swap

[peak cloak]

yeah, problem is a vlan capable switch is basically almost useless without a router that can support them

[iron kelp]

That does need a router that can handle vlans and the isp shit probs wont

[lucid siren]

Or even turn your switch into a router

Even the cheap older enterprise switches do l3

And you don't need to use a crappy gui

[peak cloak]

Not a good one

And that' requires a L3 switch

[peak cloak]

Problem is they are loud and power hungry

[lucid siren]

They do routing in the tbps

In hardware

[peak cloak]

Maybe routing, but not NAT

[lucid siren]

Fuck nat nevertheless

[peak cloak]

Which is essentially what a home router does

[lucid siren]

For nat just use a Linux box

[drowsy fossil]

and many do not...

[lucid siren]

Anything 10g does ish

I mean, Cisco, Arista, juniper

Those i meant

[drowsy fossil]

again wont be cheap tho

unless it requires licenses and doesnt have them ofc

[clear igloo]

Cisco you can use without, it's all honor based

[lucid siren]

Arista doesn't have on hardware checking

[drowsy fossil]

ah nice love it

[clear igloo]

but still, that's noise and power for the old stuff to deal with too

[lucid siren]

You can even deactivate vendor lock

[drowsy fossil]

expecially noise

[steady girder]

Im building a new PC soon and the new motherboard has 2.5g ethernet. I have a Catalyst 3750X running my house and it has a spare SFP port. Is 10g backwards compatible to 2.5g? like will it negotiate to 2.5g? I probably dont need 2.5g but its there, so I kinda want to use it lol. My server is also connected via 10g. Or is there a 2.5g SFP module I should use instead?

[lucid siren]

50W for a 1.5 tbps switch

[drowsy fossil]

i removed the fan from my switch, but now its probably gonna die

[clear igloo]

It is not

[lucid siren]

Noise however, true

[drowsy fossil]

also what model?

that seems like a very modern number

[clear igloo]

50w at 100% idle with no optics and the OS stopped?

[lucid siren]

Yes, you can, if you get the right insert

[drowsy fossil]

probably better off getting a switch

[lucid siren]

You'll have to disable autoneg

[steady girder]

@lucid siren any suggestions for the insert?

[clear igloo]

The 3750X was out LONG before 2.5g was even a twinkle in the eye

[drowsy fossil]

patrick just did a roundup that included a few 2x10g+ #x2.5g switches

[clear igloo]

Yah, instead of attempting to try a nBase-T SFP, of which there are maybe two or three on the market, just get an nBase-T switch with 10g uplinks

[drowsy fossil]

and those sfps are super power hungry

[steady girder]

so Im probably better off keeping at 1g as my switch it probably too old and doesnt support 2.5g?

[clear igloo]

It 100% doesn't

[steady girder]

bummer lol

[drowsy fossil]

you can always get a second switch for faster stuffs

[steady girder]

oh well, not like I would need it anyways. only have a 200/10 internet connection

[clear igloo]

Yah, if you have 10g ports on it via the module you could uplink from a 2.5G switch with a 10g uplink to that with 10g fiber or DAC

[steady girder]

gotcha. makes sense

[pseudo blade]

Could just get an old 10 gig NIC for fun

[drowsy fossil]

yep

[lucid siren]

I'll check which one we used in a sec

[drowsy fossil]

also 40g qsfp can be turned into 10g sfp+ with a $12 adapter and is often alot more than $12 cheaper

[lucid siren]

SFP or SFP+?

[lucid siren]

Just use dac

Then u have 4 10g in one

Ultimate space saver

[clear igloo]

or a 40g mpo optic with mpo breakout cable

[drowsy fossil]

dac is cheaper

[clear igloo]

DAC is cheaper and less power though

[steady girder]

you guys lost me lol. I really need to brush up on my networking. Spent too long in client services.

[clear igloo]

10G is SFP+ always
1G is SFP

[lucid siren]

I wasn't Shure if he said SFP or SFP+

[pseudo blade]

SFP's only capable of a tiny bit more than gigabit

So it's useless when talking anything beyond gigabit interfaces

[clear igloo]

SFP28 is 25G
SFP56 is 50G
QSFP+ is 40G
QSFP28 is 100G
QSFP56 is 200G
QSFP-DD is 400G
QSFP-DD800 is 800G
Still waiting to see what they'll use for 1.6Tb

[steady girder]

I added a C3KX-NM-10G to get the extra ports

[clear igloo]

Yah, thats got 2x SFP+ for 10g ports

[pseudo blade]

I'm sure it'll have a fun new name

[clear igloo]

and I think 2x 1g ports too, it's been so long since I worked with the 3750s 😦

[marsh oracle]

1.6TB link, Linus wet dream

[lucid siren]

Pretty shure just qsfp-dd1600

[clear igloo]

1.6Tb
it's bits not bytes, capital is important 😉

So 200GB/s

[lucid siren]

I mean, most DWDM waves barely got turned into 800g ones lately here in europe

[steady girder]

@clear igloo I believe you are right. 2x capable of 10g

[marsh oracle]

ok.

[pseudo blade]

Yeah good luck pulling data off disks at a speed and scale reachable by ltt

[lucid siren]

Most still only 200

[pseudo blade]

PCIe 5.0 is too slow

[lucid siren]

So i don't see 1.6 being mass soon, 800g barely made it to DC's now

[marsh oracle]

it aint matter

[pseudo blade]

You'd need multiple sources and destinations

[drowsy fossil]

actually youd be surprised
a single epyc processor has the required bandwidth

[clear igloo]

Yah, 800g is still barely in it infancy for switching, most of that is backhaul stuff right now

[pseudo blade]

Ah, but does it have double?

[drowsy fossil]

just about yep

but it doesnt need double for just file transfers

[lucid siren]

I'm happy I got my stuff to 100g peering ports now

[pseudo blade]

It does unless your files live in memory and your memory is also fast enough

[marsh oracle]

I stick with my 1GBit home network, no need for me to upgrade at the moment

[drowsy fossil]

and that was on 7702s, i suspect that the 9000 series is plenty capable on its own

[drowsy fossil]

nope
nvidias demo was nvme to gpu

unless you count nvme as memory

rdma is quite OP

[lucid siren]

This

[pseudo blade]

And how do these NVME drives electrically connect, hm? You still need double bandwidth, rdma is a CPU-saving feature

[drowsy fossil]

why?

[pseudo blade]

Has to get off the drives and to the NICs

2x

[drowsy fossil]

oh no i mean throughput bandwidth

[lucid siren]

Rx tx

[pseudo blade]

Yes, but to have that capacity bidirectionally you need a bus or switch with double the bandwidth

[lucid siren]

Usually switches are Rx tx

[drowsy fossil]

good idea but worded wrong
to get that capacity full duplex yes
bidirectionally no you can just be half duplex

[fallow wing]

hi there fellas
I'm looking for external opinions on my project, suggestions, advice, relevant questions...
thanks in advance
https://linustechtips.com/topic/1496088-my-first-personal-tinker-project-as-an-it-student/

[pseudo blade]

I mean it wasn't worded wrong thanks, I just used different terminology

[drowsy fossil]

bidirectional doesnt guarantee full duplex

[pseudo blade]

No. But I'm talking about the busses

[drowsy fossil]

which can be half duplex

but they arent

[pseudo blade]

Not in this case

If they were on a switch, it would be true to the CPU

But they aren't on a switch

They are electrically connected directly to the CPU

As such, all connections are full-duplex

[drowsy fossil]

actually in nvidias example they were connected to the cpu through a switch

but the switch was full duplex so it didnt matter

[pseudo blade]

This is in the weeds and I mostly replied because you said I phrased stuff wrong when I didn't

So whatever lol

[drowsy fossil]

regardless, pcie is full duplex, and 1x epyc 7003 has 4096gbps of bandwidth for a theoretical maximum of 4096gbps of data

[pseudo blade]

I mean that doesn't really hold up when you need to do anything more with your data than stream it.

[drowsy fossil]

and?

[pseudo blade]

Nothing. It's pretty self-explanatory.

[drowsy fossil]

point is this is incorrect

[pseudo blade]

Go actually think about that statement a bit

[lucid siren]

Tbh, unless you're building a supercomputer you don't need more than 100g on a single socket

Or unless you got unlimited money

[pseudo blade]

Density is valuable

It's nice being able to scale stuff up further rather than out

[drowsy fossil]

and 100g isnt enough to saturate a single nvme drive any more with gen 5

[lucid siren]

That's usually supercomputer shenanigans tho

[drowsy fossil]

i want to be able to copy files to my nas at full speed of my drives 🤷

[lucid siren]

I mean

[pseudo blade]

Should be 64gbps each direction, should be able to saturate gen 5 with 4 lanes

[lucid siren]

Do you REALLY need that speed

[drowsy fossil]

a single gen 5x4 drive is 128gbps full duplex

[drowsy fossil]

need ofc not
but can use? yes

[pseudo blade]

Uh... 4GB/s, x4 is 16GB/s... oh ffs yep 128gbps

[drowsy fossil]

i remember bc its the same as 3.0x16

which there are plenty of 1x100g 3.0x16 nics

but to get 2x100g on a single nic like i have you need 2x3.0x16 or 4.0x16

speaking of 100g tho, does anyone know of anything like the intel cwdm4 modules in the 40g generation price wise?

$7 per 100g singlemode transceiver is such a good deal, but they arent compatible with 40g

[lucid siren]

Just use fscom ones

Cheap asf

They are pretty reliable

[pseudo blade]

For some reason I did 4^3 mentally, instead of 4*4*8

Which is not helpful

[drowsy fossil]

they are individually more expensive than an entire cable using the intel modules

[lucid siren]

You mean cables with attached modules?

[drowsy fossil]

nope

eg 2x intel modules = $14, 50 meters of os2 lcto lc duplex cable is $20ish

$34 for 50m
the cheapest 40g transceiver i can find on fs is $40
min price being $80

[lucid siren]

I can't find 10$ 100g transcievers anywhere in europe

Only used

[drowsy fossil]

yes used

[lucid siren]

Trust me, don't use used in a prod env

I tried

[drowsy fossil]

my home is not really a production environment

[lucid siren]

Oh well

Then go for it xD

[drowsy fossil]

the problem is they arent compatible with the connectx-3 nics that are cheap

you need cx-4 or cx-5

[lucid siren]

Well, why you want 100g @ home tho

[drowsy fossil]

i would prefer 40g, but i bought 100g because it was cheaper

but 40g should be cheaper

right?

I also much prefer duplex lc transceivers

[lucid siren]

Just get fscom ones, i think ur time worth more than scouring ebay 🤣

[drowsy fossil]

meh got plenty of time to do mindless tasks while troubleshooting aps

[hollow marlin]

We're still waiting for the equipment for 100g upgrade on all our backbone links. I don't think I will ever see even 400g even though most our core/edge supports it, let alone 800g

[clear igloo]

Yah, I think there are 3 or 4 ASICs in total for 51.2T switches right now and I don't know of any on the market for routing yet. I'm sure Broadcom and others are working on them for routing but just haven't released yet afaik

[drowsy fossil]

my work has a 1g backbone lol

[lucid siren]

Just got 100g a year ago

[clear igloo]

The biggest thing is the gap between hyperscale and enterprise is probably just going to widen in the coming years as companies stop needing as much in terms of bandwidth increases year over year. I see it hitting a plateau for most around the 800/1.6T mark for a lot of places outside of the hyperscale realm for at least a while until moar data is needed lower down the stack

Plus until CPO catches on you're going to start seeing 20-30+ watts of power per optic meaning a 64 port switch would chug down over a kilowatt of power in optics alone let alone the probably 2-4+kW of power for fans and the ASICs

[drowsy fossil]

the google dlp switch is insanely sweet

[hollow marlin]

Yeah nothing I know of on the routing side either. I know Juniper's PTX is the first on the road map followed by MX for 800g. But most their MX/PTX line support 400g with the Trio.

I see it hitting a plateau for most around the 800/1.6T mark for a lot of places

Agreed. Even with the hyperscalers, load-balancing/ECMP being priority for resiliency/redundancy/uptime, 400g -> 800g is really 4x increase or more in capacity depending on n links.

[lucid siren]

They do that already

[clear igloo]

Yah, I was checking and broadcom's fastest routing silicon is 12.8T still. Everything above that is all switching focused. Cisco has their SiliconOne 19.2T ASIC but still nothing at 25.6T even but I think Juniper might have something for that if I remember right

[lucid siren]

If you use XR or er transcievers

For long haul

[clear igloo]

400G optics only pull about 10-15w in most applications
Yes ER/XR optics pull more but you're not filling a data center switch with those 😛

[hollow marlin]

In the SP space, multiple 10/40g links with ECMP > single 100g so we never had the need for 100g. However, with the stark increase in customer traffic over the past 1.5 years, we need the increase to 100g.

[lucid siren]

We do DDoS protection Soo we need shittons of bandwidth

[rocky badge]

@clear igloo I had a little adventure last night with the network here lol

[clear igloo]

Actually ZR optics are targeted at 15w, ZR+ is 25w x.x

[clear igloo]

did you get all the MAB stuff sorted?

[rocky badge]

No

I got a couple ports shutdown by accident Lmfao

[clear igloo]

lol, rip

[rocky badge]

University port security shuts down multiple MACs on the same port

And plugged in a Windows machine with a bridge and it cut that drop lol

[clear igloo]

welp, time to call IT

[rocky badge]

@clear igloo I do hate how the AV integrator designed the tech booth

If it were me I would’ve done 10G or 20G fiber to the rack in the tech booth instead of running individual lines

[hollow marlin]

Yeah that'll do it too. In the SP space, outside of long backhaul, throughput is not as high as most people think unlike DC environments.

[rocky badge]

We only have 8 “user accessible” data drops for the AV LAN and they’re all taken

[drowsy fossil]

i asked my boss why we werent using the sfp+ ports on our switch, but he had been monitoring it and discovered that we had not even once in several months hit full gigabit on any port of our core switch

[hollow marlin]

Sounds about right. We have multiple downstream region providers and as an example, one of them has 500/100mbps as their base plan and they're barely pushing 16-19gbps at peak. Netflow shows around ~2100 customers.

[drowsy fossil]

i mean internally within our network even

and we have most of our documents stored on the network

[hollow marlin]

That said, if he is going off SNMP monitoring with standard 5m polling, you won't catch the burst unless you look for output drops

[drowsy fossil]

ill check how fast it polls

p sure its 60s

and users waiting 60s for their documents is fine

at least on our budget lol

[hollow marlin]

60s is pretty low but fine in a "small" environment. Our NMS would keel over at that rate with the number of nodes.

[hollow marlin]

I mainly gave the example from an average residential user as many overestimate how much throughput they use. It's even lower in most average business scenarios too.

[drowsy fossil]

oh wait no it is 5 mins

[peak cloak]

So LTT hack was classic email attachment phishing

[lucid siren]

Yeah, we have terabits to the level 1's and 10s of t's in some pni's

[drowsy fossil]

sure on average i use nothing, but i want to be able to download stuff super fast
and i want to be able to do complete backups of my computer in minutes locally

[lucid siren]

Yeah

[peak cloak]

I wonder if they even have some sort of filtering system

[hollow marlin]

Yeah we have no where near the need for more than a few 100g to our L1s nor to our IXs.

[ripe hatch]

Looking at major wifi upgrade

Does the Ubiquiti Dream router have any huge glaring issues

Besides imo it’s lack of wifi 6e for $200

[rocky badge]

Doubt it

[livid hawk]

Is this good for a home internet set up? The repeaters only go up to 300 mbs but they will be for the other side of my home that doesn't get great wifi coverage.

Or did I maybe overpay?

[drowsy fossil]

you definetly overpaid, and for asus routers its highly recommended to use other asus routers in mesh mode as the extenders rather than repeaters

[shell sail]

Hoping to get some help. I have OPNsense set up with Adguard plugin set for my DNS server. I am not using UnboundDNS on this so nothing is going through that. I set up a wireguard client to connect to a VPN server (surfshark). The tunnel is up, but when i enable the rules I set based off of this guide, all external traffic fails but my internal traffic connects fine, ie to my router or my servers. I'm guessing I am having an issue with a firewall policy somewhere... I am doing policy based routing and set for only device to go out through the vpn connection right now. Link for the guide I used. I followed these settings to a T, except did not add the Kill switch. I have tripled checked the settings and double checked that. Any thoughts is appreciated, let me know if needing screenshots of anything.

https://docs.opnsense.org/manual/how-tos/wireguard-selective-routing.html

[shell sail]

It does, If I disable this rule my connection works through my normal outbound ISP connection. When I turn it back on, my traffic does not go out.

As for outbound NAT rules, here it is.

Reading some other posts, do I need to have a firewall rule that allows access to the Adguard DNS server, even if it is hosted on my OPNsense box?

[pastel hound]

Hey all, does anyone here have experience with Mikrotiks and their performance with GRE over IPSEC?

I have a RB5009 and I'd like to have a fully meshed site to site IPSEC VPN topology with 5 remote sites and to be able to use OSPF in the future, I need an internface which is why I'd like to use GRE on the Mikrotiks. But I remember reading somewhere that using GRE has quite the performance impact CPU wise (of course MTU and fragmentation also plays a role).

[sharp sparrow]

Why do you need to do it with a virtual interface?

[pastel hound]

because Mikrotiks don't support VTI, so without GRE or EoIP or something like that I do not have ANY interface on the Mikrotiks

[sharp sparrow]

Ok, let me rephrase sorry. Why do you need an interface?

[pseudo blade]

Everything has a performance impact, what kind of throughput are you expecting over the tunnel?

RB5009's are pretty good for performance, you might get away with whatever you're looking for

If you're not married to IPsec wireguard's great on my Mikrotiks and does expose an interface

[vital gorge]

how to make internet better?

[pastel hound]

that I mentioned initially. so I can implement OSPF, later on

[sharp sparrow]

Sorry mate, I skimmed 🙂

[pastel hound]

I would probably actually switch to Wireguard but unfortunately I will still be having some older Fortigate peers for the time being and they do not support Wireguard.

[pastel hound]

no worries

[sharp sparrow]

Have you tested perf over gre/ipip on those devices?

[pastel hound]

throughput max. 80/20 Mbit anyway but my concern was also due to the big amount of low bandwith sessions I'm expecting. also syslog logging of all traffic, that's gonna put some load on the Tiks 😄

[sharp sparrow]

Can't the 5009 do like 1.5Gbps of ipsec?

[pastel hound]

no, I'm doing a "pilot"/PoC in a virtual environment with the free CHRs but they are limited to 1 Mbit. I'll get my 5009s in the next few days and will test with them

[sharp sparrow]

I use a 5009 as my edge at home, if there's any sort of basic testing you would like me to do :p

[pastel hound]

yeah, something like that, but GRE seemingly is loading the routers hard

i'm planning to replace my aged Fortigate 60D which is currently my edge firewall + vlan router for all my labs and DMZ networks with one 5009

I have like 25 VLANs on that fortigate + something like 7 or 8 ipsec tunnels with quite strong encryption (site to site and client to site) and I hope that I can route wire speed between the VLANs with the 5009 + logging all the edge traffic to a syslog server 😄

• later OSPF

that 5009 is gonna be under load lol, especially if I use the 10G interface for my DMZ uplink

that firewall filter rule config is gonna be a pain in the butt compared to the fortigate xD

I'm also concerned because of that crapton of firewall filter rules that i'll have to put on the 5009... but I guess there's no better way to find out than to do it

[lucid siren]

?

[pseudo blade]

That seems pretty achievable even with an inefficient config, I wouldn't be worried at all, low bandwidth sessions aren't very expensive

RB5009 has hardware-accelerated IPsec and GRE's cheap enough to get 100mbps easy - even on Mikrotik's older single-core MIPS stuff the RB5009 beats the pants off.

[thick minnow]

Boy howdy pfsense sure is fun

[olive trellis]

not sure if correct channel but I use a tp link T3U plus wifi adaptor and my wifi speed is stable 200mbps but sometimes on my pc I get 50mbps or even 5mbps I have newest driver installed any idea why?

[pseudo blade]

Tech support question should go in #1027757333117415424 , but could be interference on the frequencies used in your house, could be driver issues. WiFi is a shared medium and can be very difficult to troubleshoot, which is why running a cable, MoCA and powerline ethernet are common recommendations for people who want consistent performance.

I will say that my experience so far with USB WiFi adapters has been consistently abysmal.

[vital gorge]

💀

bruh

how

[lucid siren]

Datacenter 🤣

[pseudo blade]

How to get fast internet:

1. Spin up a large VM in AWS/similar
2. Perform speedtest

[lucid siren]

or get colo :)

speedtest kinda slow

[clear igloo]

Or have a fiber plan with 2 and 5Gb options 😛

[lavish hollow]

or 3/3 😻

[lucid siren]

This is on 100g tho 😂

[thick minnow]

you may be running into system hardware limitations. most systems can't keep up with that much network bandwidth. Cache, Memory, I/O bus, network interface, tcp overhead, all have a price.

[manic cape]

When it comes to home routers, is there any specific one that is superior to another? I've been looking at these three, which all have extremely different price points

TP-Link AXE5400 ($200)
ASUS RT-AXE7800 ($300)
ASUS GT-AXE16000 ($630)

[vital gorge]

what sort of speeds could I get with 2.5g ethernet?

[lavish hollow]

I was going to start typing about that

that's with 2.5Gbe NIC on a 3/3 fiber plan

this is direct from my modem/ont combo https://i.imgur.com/SYeDPNY.png

[lucid siren]

Without nat

[lavish hollow]

there we go

[lucid siren]

Didn't find a Speedtest server with more than 10g yet however

[lavish hollow]

once I get my plexyserv active again I want to look into a pair of 10gbit nics and a switch with a handful of 10gbit ports

[lucid siren]

Should technically be 25gb the Speedtest (uplink to the switch from the hypervisor)

[opal pagoda]

few years back it was hard to find speedtest server with more than a gig bandwidth

[lucid siren]

I think I can't get around setting a 100g Speedtest server up...

[opal pagoda]

this is what speedtest cli is for

[clear igloo]

Up to 2.5Gb/s if you pay for it

[clear igloo]

Depends on which one you use 😄

[opal pagoda]

this is the speed i dont pay for (its free real estate)

[peak cloak]

just host your own

[vital gorge]

icic

[lavish hollow]

and as whistl said earlier there are some things that can limit you other than your NIC

[thick minnow]

1gbe fiber is only $70/mo here. 2.5gbe would require a router upgrade, and would be closer to $125/mo

I've never saturated my 1gbe internet link. See no reason to pay more. I'm the only heavy user on this lan

[lavish hollow]

Bell offered us 3/3 for $60CAD the other week for 2 years so I took it

[thick minnow]

cool!

[lavish hollow]

1.5/940 is the next tier down, so 2.5Gbe is still nice to have coming in on more consumer devices

[thick minnow]

they offer 5gbe too, but really, I'd only be getting that for bragging rights.

[lavish hollow]

I was coming from very consistent and stable 1gbit/50mbit cable, the download speeds were over 900mbit most of the time but cutting the latency by more than 50% and the massive upload increase will be very nice

[opal pagoda]

i cant even get more than 16mbps down 1mbps up

[lavish hollow]

I need to find some good literature and dig into some things and see if I can get my current 2.5Gbe nic to perform a bit better when I run into a server that can feed it

[opal pagoda]

only thing you could do is enable jumbo frames

[mellow hemlock]

How much network info might I learn from nothing if I read every message from this channel?

Do you know of any that you recommend?

[clear igloo]

ccent is dead, CCNA is the lowest and it covers EVERYTHING in the networking realm (collab, security, automation, etc)

but I would focus on routing and switching portions of the CCNA at least to get a good footing

[mellow hemlock]

I’ll have to give these a look. I appreciate the feedback. If you have other ideas too feel free to ping me as well. Ty guys

[shell sail]

that didn't work.

I'll have to go through and rebuild that. Is there something I need to do with Adgaurd and the vpn tunnel? Unbound is off and I have my dns set to adgaurd on opnsense.

[devout scaffold]

should probably fix this 🙃

[lucid siren]

Ran into some random issue

[shell sail]

I have a live view screen shot and do see some blocks coming from the device i have set to go out through that tunnel. I can post it, but need to know the best way to redact IP address but to give good information still. How can you tell what rule is causing the block?

Nevermind. this is a different device getting blocked. Need to find out what device that is, but it's not going through my vpn tunnel

[thick minnow]

my home network now runs on pfsense

[hard arch]

did i do my ethernet cable wireing right

[clear igloo]

You mixed up the last two color pairs 😦

[hard arch]

ahhh fuck thats why its not working then

do sometimes mix up the orange and striped orange tho so a few of my cables got them swaped dont cause issues anyway so no big deal

[shell sail]

So I rebuilt my wireguard tunnel, removed all the firewall rules for it and rebuilt those as well. Went back to unboundDNS and turned off adguard for now and still nothing. 😦 And used the docs.opsense.org/manual/how-tos/wireguard-selective-routing.html guide to the T. Might have to pop on their forums and post this there... So confusing.

[lucid siren]

Fucked up some routing 😂

Easy to fuck bgp

[worn viper]

Deco x20 honestly surprised me,I am temporarily using it as a router and it's been pretty decent

[ancient arch]

I find it quite difficult to fuck BGP up. It gives the most granular control to user so nothing is advertised without your consent

[devout scaffold]

Who here is familiar with DumaOS? Is it stable yet?

[opal pagoda]

what is dumaOS?

[devout scaffold]

A gaming oriented router OS, typically ran on Netgear’s gaming routers such as the XR1000 which was totally bunk for the first year it came out

I dropped it after getting their R2 as a replacement for the XR1000 and ended up going Ubiquiti instead, because the R2 literally bricked itself one day without any user input

It’s got cool features though, it’s just very buggy and kinda slow

[ancient arch]

Gaming routers are all bloated software wise.

[devout scaffold]

Yeah definitely

[lucid siren]

if you announce routes trough your network trough up to 5 routers behind each other it gets complicated

[devout scaffold]

Ubiquiti has been a breeze so far, but I got the UDMP right before the SE came out so that kinda upset me

[lucid siren]

and if you then add anycast to the mix

you can have fun if you fuck up

[ancient arch]

Why would anyone ever have 5 routers behind each other? And with BGP?

[devout scaffold]

https://tenor.com/view/colbert-popcorn-popcorn-stephen-colbert-eat-popcorn-popcorn2020-gif-16075618

[pseudo blade]

BEHOLD: THE NETDUMA R1

[devout scaffold]

HAHAHAHA

It’s a gaming switch bro

[pseudo blade]

That's not a switch

[devout scaffold]

I know

[lucid siren]

well, 2 out of them are filtering, 1 is core, 1 up to 2 are distribution

[devout scaffold]

It just looks like one

[lucid siren]

all my switches do routing / bgp

[pseudo blade]

That's a Mikrotik Routerboard with a sticker

[devout scaffold]

Their R2 is dogshit

[pseudo blade]

But uh yeah it's gaming-optimised routing with a healthy dose of snake oil

I mean you can class traffic and prioritise it sure

[devout scaffold]

But the VPN features that were the main selling point for me was removed from the XR1000 after launch when I bought it because of some dumb partnership with some shitty anti virus security company which bundled its paid-only VPN service with it

And don’t get me started with the Wi-Fi drop outs

That were only resolvable through a reboot

[pseudo blade]

I actually knew a few people who sold rebadged Mikrotiks with "gaming-optimised VPNs" way back in 2014 or so

Though unlike Netduma they pivoted

[ancient arch]

Something like Exit Lag?

The app

[pseudo blade]

Honestly I don't keep track of them

[devout scaffold]

Netduma just blocks connections to/from countries, exitlag actually does route you differently iirc

[pseudo blade]

I actually ran a VPN for a few days to bypass a routing issue that made Brawlhalla unplayable

[clear igloo]

add vxlan evpn or srv6 when?

[pseudo blade]

It was helpful then, and only then

[devout scaffold]

I think James runs a netduma router

[lucid siren]

vxlan for customer shit

[devout scaffold]

They used one in one of the Wi-Fi upgrade videos a year ago

[lucid siren]

but theres switches in there that dont support it so theyre carried along xD

[clear igloo]

OSPF or IS to IS underlay?

[devout scaffold]

I think it was the XR500 which doesn’t have features ripped out, but only supports 500Mbit/s

[lucid siren]

ospf

[ancient arch]

Wut?

[lucid siren]

bot go brrrrr

[ancient arch]

Anyone used IS to IS in production?

[clear igloo]

ISPs mostly

[lucid siren]

i hope not, but probably

true, forgot about those suckers

[ancient arch]

I haven’t seen yet. I work at ISP

Perhaps older ones use it

[lucid siren]

could explain some things

[clear igloo]

My customer wants IS to IS because Arista can't do SRv6 in OSPF so we need to convert to it from OSPF

[opal pagoda]

mikrotik goes brr

[lucid siren]

i do it diffrent for v6

its not a legacy setup

[ancient arch]

Ooh.

[lucid siren]

v6 is just straight switching to core router

[pseudo blade]

It really does, I've sold Mikrotiks as something else/ part thereof in two companies now

I think I'm done with RouterOS for that kind of shit though, the hardware's neat but the software just doesn't real allow for that bullshit

[lucid siren]

I hate Mikroshit for routing

They ok for l2

But l3, hell nahw

[pseudo blade]

Mikrotik's fine for routing though dynamic can be fun

[lucid siren]

They do a lot on the cpu

[pseudo blade]

Not in this decade they don't

[clear igloo]

v7 moved off the CPU right?

[lucid siren]

So they die if you send a few Mpps

[pseudo blade]

You can do switch hardware accelerated L3 now yes

And what you can't offload gets calculated and put back on the switch wherever possible

Would it really kill them to put a semi-decent CPU in their switches though?

[rocky badge]

@clear igloo

??????

[clear igloo]

rip

[opal pagoda]

we do cisco only

[rocky badge]

“Disable APIPA” bruhhh

[clear igloo]

but why??

[pseudo blade]

Yeah Cisco doesn't offer anything that would do anything I'm trying to do lol

[clear igloo]

those octal cables on top 👀 ?

[pseudo blade]

I mean sure their super big switches with Xeons in them would, if you'd like to strap one of those to your car

Do it, it'd be funny

[opal pagoda]

no, power cords with this kind of end

[clear igloo]

ah, picture was a bit blurry and looked blue-ish from a distance

cat9k switches or cat8k routers?

[opal pagoda]

its blurry for a reason

[clear igloo]

haha, fair 😛

[pseudo blade]

Yeah they look blue in the preview but not if you open it fuly

[opal pagoda]

c9200

[clear igloo]

Nice, they're good switches

Just be careful with TACACS on the latest releases 😛
Until 17.6.6, 17.9.4, or 17.11.1 or later

[rocky badge]

I wish university would just rollout PEAP TLS/cert based auth @clear igloo

[clear igloo]

That would be easy, lol

[rocky badge]

Logging into a laptop is so fucking annoying

[clear igloo]

yah

[rocky badge]

It’s doing pre logon user wireless authentication

Not a machine cert

[clear igloo]

damn, that stinks, super slow and annoying

[rocky badge]

It is…

Computers already auto enroll to client authentication certs too

But they have a shitty PKI

this computer was issued a certificate DIRECTLY FROM THE CA

[clear igloo]

lel

[rocky badge]

“Firepower rpc access” lol

“Firepower dynamically mapped ports”

[sweet coyote]

Hello, could someone help me with that situation?

[pseudo blade]

So you have ethernet running from the switch to each outbuilding?
Tbh I can provide suggestions but I'm not sure how many of them you'd adopt - do you trust the people in the outbuildings, is this an airbnb-style situation?
If they're not trusted you want to replace the switch with a router and isolate each one from the others. it also sidesteps a bunch of problems - collisions as you say but probably more relevantly if you plug one of them in wrong and you end up with two competing DHCP servers

I notice you've got it isolated from the home network and can't think of another reason why you'd do that

-Or have four buildings unless you sleep in them on rotation :P

[sweet coyote]

Yes, the people in outbuildings are airbnb style situation, and I thought about putting another router in lieu of the ethernet switch

[pseudo blade]

I'd suggest having them all nicely isolated on individual subnets and using the routers in the outbuildings as plain access points

[sweet coyote]

Thing is, I'm not knowledgeable enough to know how to do that

It took me a lot of time to develop that huge network (it's a big place, and I ran a lot between the 4 buildings and was sweating a lot haha)

the only thing I didn't do myself is runnin the ethernet of the 3 cottages (an electrician did it)

and the ethernet switch was existant

[opal pagoda]

get a managed switch, mikrotiks are deacent value

[pseudo blade]

Well if they generate revenue for you, you could always get a consultant to do the setup for you, testing and all.
Else you can learn to use something like a Mikrotik hEX in place of your office switch.

[sweet coyote]

It's not my home/cottages, and I'm the one paid to do the setup, I usually do computers more than everything, but here they called me for their macbook, HDD and then asked me to make this network...

I tried and failed, well not entirely, the internet is the house is very very good

[sweet coyote]

aren't these expensive? what about a managed wifi router with 1 WAN and 3 LAN? I have one in hands

[pseudo blade]

Mikrotik's don't really do WAN/LAN, you define it all yourself as per your needs

[sweet coyote]

is the MikroTik Hex Lite, priced 50eur okay? thanks a ton for your inputs @pseudo blade and @opal pagoda

[pseudo blade]

I'd suggest not going for a managed switch because you'll be doing NAT for each unless your 4G router is awesome and need to do that somewhere

hEX lite is 100mbps only and half the CPU speed

[sweet coyote]

I don't go over 40mbps with 4G+ internet

[pseudo blade]

It'll do 40mbps but will be useless if they ever decide to offer more than 100

[sweet coyote]

and people in cottage won't be able to much a lot of stuff

[pseudo blade]

The regular hEX isn't much more expensive

[sweet coyote]

okay okay, and if I have a router that does gigabit, it would be good, right?

[sweet coyote]

double the price in my crap country

[pseudo blade]

I mean you're going to give the customer a quote, right?

[hollow marlin]

IS to IS > OSPF

[opal pagoda]

what country? since you use eur i assume somewhere in eu

[pseudo blade]

idk, I get bad feelings on this one - don't be afraid to tell your customer to get someone else to do the work if you're not really familiar with networking like this. Especially if you're coming down to the last euro on pricing. Beats a bad experience.

Otherwise you're signing yourself up for a fair bit of learning

[sweet coyote]

Well, to them they spent enough already and they told me that "last summer, all three cottages were working before you installed internet in the home"... the 4G internet for the cottages was EXISTANT and when I tried them, I realised Internet wasn't working in the 3 at the same time

so I told them and they were like "well it worked before"

[sweet coyote]

thing is these are clients potentially bringing me other clients, like a f*ck ton and all are paying money

[sweet coyote]

I'm sorry but I prefer to avoid answering this

But you're right @pseudo blade, shouldn't have taken a task I cannot compelte

well!! again, the house part I did everything and it works

then for "fun" I checked the cottages part and it wasn't working

I told them and they think it's because of the house networking

even though both aren't linked at all

[opal pagoda]

anyway, senetic is a store where lots of networking equipment can be found and prices are resonable and also they ship preatty much everywhere

https://www.senetic.net/

[sweet coyote]

In fact, I tried to plug the "PowerLine Wifi" onto the "Ethernet Switch" to get the Starlink internet is both the house and the 3 cottages, and it didn't work

but when I plugged back the 4G router to the Ethernet Switch, it wasn't working either

Cottage 3 constantly get access, but If I put power into Cottage 1 & 2, they don't get internet from their respective router

[pseudo blade]

80% chance either you've got a router's switch ports plugged in a loop or you've got the "LAN" port facing the house with DHCP enabled

[sweet coyote]

but if I plug a PC into the wall ethernet socket, it works

[sweet coyote]

which LAN port? on which device I mean?

[pseudo blade]

On the routers in the outbuildings

[sweet coyote]

oh

[pseudo blade]

That would nicely bridge the house's networking with any guests who connect to the access points, offering no security or isolation whatsoever considering the lack of any network design to prevent it.

Absolutely awful idea unless you had a router set up with appropriate firewalling rules in place where they can't touch it.

[sweet coyote]

We abandonned the idea, the cottages will work using the 4G+ internet router, so that it's separated from the house network

So I need to check if the cottages's routers are DHCP enabled and disable that?

[pseudo blade]

Depends on the rest of their config

And what you plugged into where

If you actually just had them set up as per default with the WAN port facing the house they need DHCP because they're actually each doing NAT

[sweet coyote]

they are defaulted with WAN port facing the house, indeed

Cottage 1 & 2 have the same router model, Cottage 3 has a different one (Netgear), and this one works nicely

I suspect the brand of Cottage 1 & 2's routers to be crappy, they're the same brand of the Ethernet Switch and they were in place when I arrived

[pseudo blade]

So why is it your fault they're broken if they were already in place?

[sweet coyote]

because the client tells me it worked before I worked on them

[pseudo blade]

I'd be correcting that misconception before any networking problems tbh

[sweet coyote]

the only thing I did is renaming the SSID so that the guests could find them easier to connect

but if we only take this

[pseudo blade]

You can't and should not be responsible for any networking equipment you've ever touched, especially if it failed later and you didn't supply it

[pseudo blade]

But I get it, money

[sweet coyote]

it's not a complicated network, is it?

[pseudo blade]

No. But it's also not properly isolated.

And that lack of isolation (and of course your lack of control over the equipment in use) leaves you with a few particularly fun failure modes

[sweet coyote]

so in order to isolate every cottage on the network side, I'd need to replace the Ethernet Switch, right?

[pseudo blade]

Sure, but I'd hate to see how you get paid for it if they'd reject a $30 more expensive router

[sweet coyote]

well, I got a free wifi router with 3 LAN 100 plugs, a client gave it to me

if I can reuse it in that scenario, i'm okay

my working time is nicely paid, and the client is nice and cool

he just wants it working for next summer

[pseudo blade]

Does it allow you to independently manage the interfaces?

If not it's no better than the switch was

[sweet coyote]

i'm plugging it to check the management intercace

[pseudo blade]

if it's a consumer-grade router you can default to "no"

[sweet coyote]

it's tp link

[pseudo blade]

No chance lol

[opal pagoda]

many routers employ this kind of structure
router>internal ethernet switch
and not
router > bunch of independant interfaces

[pseudo blade]

-So it's basically just adding an extra NAT in and you get a crappy 3-port switch.

It will let you determine if the existing switch has dead ports but little else.

And it's not an especially likely failure mode though it does happen.

Here's a block diagram for the hEX lite - it's architecturally the same as described above but the switch and router operating system expose them as if they were regular interfaces attached to the router CPU, which is not something a home router manufacturer is going to bother to do.

[clear igloo]

[sweet coyote]

Well, trying to use the tplink router broke my Internet

And I don't really understand the difference to be honest, but if I need to order that... I will

[pseudo blade]

It's not just ordering that

It's knowing how to do what you want

And what a secure and robust enough configuration is so they can't break it again

[sweet coyote]

Honestly the tplink home router interface seems pretty "complete" to me

May I try it ? to replace the ethernet switch?

could I be able to isolate the 3 cottages with it?

[sweet coyote]

thanks for the help anyway, sorry if I'm a bit dumb

[pseudo blade]

It's not hard for something to look complete if you don't know what complete is, and your product targets a different market with different requirements.
This is a hEX I'm using to stage and configure another attached router and handle a few failure cases, with two networks.

it's configuration is an absolute mess but that's beyond the point, I attach and remove stuff from this all the time

Just for fun, this one has three ways out to the internet and two are disabled or removed

[sweet coyote]

But I don't need 6265431 settings for my use case, do I? I mean, what specific SETTING to I need for my usecase?

[pseudo blade]

Multiple LAN networks

or some way to define them

[sweet coyote]

zero computer will be connected to LAN

only through Wi-Fi, all I want is 4G+ internet <-> 3 LAN ports <-> 3 Wi-Fi routers

[pseudo blade]

You could also use port isolation on a managed switch

Yes, and you do not have sufficient understanding to know why this is not best practice and can cause problems

[sweet coyote]

so if the router I'm currently owning has port isolation, I am fine?

[pseudo blade]

It won't

[sweet coyote]

what kinds of problem?

I'm failing to understand what's complicated in my use case

[pseudo blade]

Honestly I've already explained it multiple times and don't feel generous enough to repeat it again

[sweet coyote]

alright, sorry

I thought having 3 access points with separated wifi was a simple task, my bad

[pseudo blade]

In a house: yes

For untrusted guest access: no

[sweet coyote]

Guests won't be creepy trying to do something weird, they are on vacation for a wedding, they want to use internet in the countryside, that's all

they spent 1 to 2 days there, max

Imagine it's a house, for a moment, how come it wasn't working in the first time, is what I fail to understand

[pseudo blade]

Additionally, you're using routers as APs, which gives you four ways to plug it it in wrong per building

[opal pagoda]

i once stayed in a airbnb that had a isp router in the room i was staying in and it used default password and username
could have done quite a bit of tomfoolery but i am a good citizen so i didnt

[sweet coyote]

I mean, I have an internet access, that I sent in a ethernet switch, to 3 different routers, and it's not working, even BEFORE WE TALK about security

[pseudo blade]

I completely get that you fail to understand. But I keep telling you and you reply that you fail to understand. Not sure what you want to hear

"Yeah all good mate just do what's presently broken"

There you go. Good night.

[pseudo blade]

Yeah it's common when every idiot thinks they can set up motel internet, last place I was in they just gave the office WiFi router password to all guests.

Booking server, Windows 7 reception PC and all guests on a VDSL modem-router

[opal pagoda]

i have been able to access router management portal multiple times when on vacation, i mostly try that if internet is broken and was able to fix it few times

[sweet coyote]

Okay, I reread your first proposed solution, which was, "I'd suggest having them all nicely isolated on individual subnets and using the routers in the outbuildings as plain access points"

can I do this with the current devices, or not? thanks

[pseudo blade]

No.

[pseudo blade]

Nowadays I just ignore it and use my phone directly/hotspot if I'm travelling, it's rarely worth bothering with and 4G in Australia's more populated locations tends to be quite good.

-And if I'm too remote for even that I probably have better things to do

[sweet coyote]

What speeds do you get on 4G in Australia?

[pseudo blade]

Obviously it's quite variable, but in my house here I can get 200mbps on my phone's Cat-20 modem, in my last place I could get 400 down easy off a tower in an industrial area that empties after work that I was near, never saw it below 200 actually

During the day more like 50mbps here, speeds pick up later in the night

[sweet coyote]

wait what, 200mbps on LTE?

[pseudo blade]

Yeah definitely.

[sweet coyote]

you're very lucky, I never saw anything above 75mbps

with my devices of course

[pseudo blade]

It depends a lot on the tower you're connected to and your modem.

For example, here's a result of a cat-6 modem under test https://www.speedtest.net/result/14529727634

[sweet coyote]

yep, that's about what I get

[pseudo blade]

This one's off a cat-20 modem in my phone, same position one minute later: https://www.speedtest.net/my-result/a/9160412819

It obviously has a rather different antenna layout and supports greater carrier aggregation

[opal pagoda]

problem that i had is congestion in 4g band
4g speeds were as bad as 0.5 down 5 up
so i switched their router to 3g band (HSDPA) and it went up to 10 down 3 up

[pseudo blade]

Yeah it's 1:45AM here lol

Those are not daytime speeds though the Cat-6 modem will usually do more than that, closer to 100mbps at this time of night

[fleet gate]

Normal for a new router?

[pseudo blade]

It looks like what I'd expect here ~500km from the datacenter plus infrastructure, give or take.

My friend in Sydney gets ~4ms to Sydney datacentres but everything's close to him

It's fine enough for most networked gameplay, maybe not ideal for live game streaming (GeForce Now etc.)

[fleet gate]

50km from the data center or 33 miles

Supposed to get 125mbps which I normally do but since the new router im getting 15-20

[pseudo blade]

The latency doesn't look like too much a problem but the reduced throughput is probablt worth a call, that's rubbish

[lime copper]

Hello, I am setting up a secondary router connected to the main one to use it as an extender and as a separate testing network, I changed the gateway to 192.168.2.1 since the main router has a gateway of 192.168.1.1, when I use the same gateway, I don't get managed to access secondary router GUI panel, but when I change the gateway of the secondary router to 192.168.1.1, I get managed to access both of the routers and have access to the internet, but when I use my laptop which is connected to the main router, it connects automatically to the secondary router gateway and I don't get managed to connect to the internet until I turn off the secondary router or disconnect it, why does that happen and how to solve it without having to manually setup gateway and IP for every single device?

[glass glacier]

What ports do you have the two different routers connected on? I could see two different issues depending on which ones, either a Double NAT or a DHCP Server conflict.

[lime copper]

Give me a minute to port scan each of them

[glass glacier]

I mean physical ports, like WAN or LAN

[lime copper]

Oh, it's connected to LAN port

[glass glacier]

LAN port to LAN port, or LAN port to WAN port?

[lime copper]

LAN port to LAN port

Sorry, just reverified

My bad, it's LAN port to LAN port

[glass glacier]

Ok, each of the routers have their own DHCP server that tell devices connected to them what IP address and gateway to use. With how the two routers are connected, devices connected to either router could grab that info from either one "at random".

[lime copper]

Oh no

[drowsy fossil]

With router 2's wan port going to a lan port of router 1 you should be able to get the effect you want?

[lime copper]

I can't insert the cable to the WAN port, it's different

Router 1 is Fiber and router 2 is ADSL

[drowsy fossil]

Ah so they are modems

You probably won't be able to do what you are describing with these devices

[glass glacier]

Yeah, only pain and suffering will be found trying to setup advanced networking on consumer grade combo units.

[lime copper]

Oof, then I guess all I have to is statically setup the gateway for each device

[drowsy fossil]

Why tho?

[lime copper]

Because it can randomly choose the DHCP server which means I have to statically setup the IP and gateway for each device like I did on my laptop and it will work just fine

[drowsy fossil]

Just turn off the DHCP server on the second one

[lime copper]

Wait, is it that simple, cool, I will try it right now

[drowsy fossil]

Basically turn off everything you can on it and it'll be a simple switch

[lime copper]

Okay

And success

It works like a charm, now I can just normally on other devices access to router 1 while isolating router 2 and have access to both of them in my desktop testing lab

Thank you guys

[opal pagoda]

not sure about isolating, you should try to ping a device connected to router 1 from a device connected to router 2

[sage crow]

Odd I’ve done this exact thing with 2 gl-inet travel routers and had no issues whatsoever

Oh nvm one is fiber

[lime copper]

All I had to do is to disable DHCP in router 2

No, it has nothing to do with fiber

[sage crow]

I didn’t in my experience but it may just be down to the firmware of the router

[snow anchor]

does any one have any ideas of how I can get faster wifi or even ethernet in my room? I have to use a USB wifi adapter and its pretty slow expatile when everyone is using the wifi.

[sage crow]

Get a really longer Ethernet cable

[snow anchor]

I don't have ethernet in the walls and can't install it because we have asbestos. and I don't have working cable so i can't use it