>>107005973
I made one of these for my brother without the cripping tool. It was painful pushing the contact points one by one. He didnt even say thank you

is making your own cat cables recommended? will alie express gear work?

>>107006062
No, once you start using cat6 or better, just acquire cables from somewhere.

File: IMG_20251025_145441_s.jpg (700 KB, 2000x1496)

700 KB JPG

my new HTPC/NAS based on Asrock N100M. half of the components are old stuff I had laying around (case, PSU, SATA SSD for system, and my HDDs with data). I just bought a mobo (new), RAM (used) and a PCI SATA expander.
it works, but I can't honestly recommend it for a Linux HTPC. it was a bitch to get it to play 4k@60, and it still occasionally drops frames when streaming 4k youtube (even though in practice you can't really see it). I imagine it works better with Windows.

Opinions on the Ubiquti Dream Router 7?
I was originally going to go for a TP-Link or ASUS WiFi 7 router, but they're almost the same price and the Ubiquti router seems more prosumer than just consumer.

>>107006062
I've either made my own cables if I needed a specific length or I've replaced the connectors when they break, everything has always worked with a basic crimp tool I got years ago.
But the fastest connection on my network is 2.5Gbps and the longest cable is 10 m so take my advice with a grain of salt.

>>107006087
>he doesn't produce his own glass fibres for homemade fibre optic cables

wait, /hsg/ doesn't crimp their cables?? like wtf? what kind of sysop larpers are you?

>>107006207
i did it in my gay little highschool cisco networking class in highschool ~2002-2003.

File: IMG_20251025_145517_s.jpg (308 KB, 2000x1496)

308 KB JPG

>>107006092 (me)
idles at 22V with all drives put to sleep. not too bad considering it uses an ancient ATX PSU.

>>107006234
>22V
it better not be

>>107006234
*22W

>>107006207
>not be neet
>heave a near infinite supply of network cables to steal from the office/data center.

>>107006207
>just make your own fibre optic cables bro
They are dirt cheap from local suppliers why would I bother?

>>107006260
>Real professionals buy solutions than make them.
I hate this gay earth.

>>107006304
>Real professionals have the company purchase professionally made cabling instead of wasting their time making cables and then troubleshooting why the connection is intermittent.

>ITT /hsg/ outed as a bunch of crimplets and posers

>>107006454
>ITT rampant schizophrenia

>>107006062

i had cat6 and all they run was 10base
bought cat5e with high hopes

>>107006444
Why even have you as a employee if they can just buy the stuff and service?

File: 1745432312858631.png (491 KB, 705x1004)

491 KB PNG

ive crimped at least 300 cat5/cat6 cables in my life, including ~30 in the past year

i made my own server recently with an old pc using an amd r5 2600, 1080ti, 16gb, 2 hdd and 2 ssd.

i use true nas scale and my apps are the arrs+jellyfin and something for ebooks and photos.

am I better with something like a lenovo minipc?

>>107006530
Because your average person has terminal dogishit retard brain and thinks their computer is a scary magic box that will ruin your life if you touch it wrong and therefore chimp the fuck out when asked to do anything remotely technical, let alone setup a professional network

I need a recommendation for a nice router with at least 4x 2.5gbps lan ports

>>107006444
>>107006454
>>107006530
When I was in a proper DC (Google), 90% of cables where pre-made.
Intra racks where DAC, between racks where MPO (impossible to field terminate afaik)
They did have the means to terminate single mode, but I've not seen a single soul do it in my 2 years. That being said it might be handled by the fiber contractor if they really need a cable.
The only cables that where regualry done by us techs was rj45 and it was only for the batteries at the bottom or top but even then they where rarely replaced.

When new hardware comes in, it's followed by pallets and pallets stacked with spools of fiber of various lengths.

File: Linksys Smart Wi-Fi.png (9 KB, 470x212)

9 KB PNG

What happens if I combines WiFi names (same name and pass for both the 2.4GHz and 5GHz), but I switch off the router's 'band steering'?
Googling isn't helping much, I can only find examples of people talking about combining + band steering, or separate names + no band steering. Can't find the middle ground I'm talking about.

>>107007725
AFAIK the device just does whatever the hell it wants.

>>107007753
Thanks. I'm gonna give that a try, I'm having some connection issues. Maybe the devices know better than the router.

File: KLN-26001-2T.jpg (12 KB, 250x250)

12 KB JPG

>>107005973
Passthrough are gay.

>>107006062
I would, but I'm an electrician by hand. That being said it's the cheapest and most material efficient way to install data drops throughout thehouse. You'll also have an easier time fishing raw cable without the connectors. Along with a crimper get
>pic related
You'll use it to strip the jacket and trim the wire. Don't get pass through connectors.

File: 1408959307609.gif (1.39 MB, 204x141)

1.39 MB GIF

>>107005973
Fuck passthrough connectors. They're easier to use than the standard ones, but they leave the ends of the wire exposed.

I can't begin to tell you how many of these pieces of shit I've had to pull and either redo or outright replace because some fleck of conductive dust got into the ports on the switch, moisture got in and corroded them, or the cables got crossed together because they weren't crimped down correctly and something physically bumped the ends at some point.

File: C6AS-3085BL_01__22096.jpg (113 KB, 1280x1049)

113 KB JPG

>>107007838
Additionally, spend your money on good cable, do it once. Plenum rated, insulated, and shielded. You'll need to make sure your connectors accommodate this type of cable. Shielding isn't as important as good insulation.

File: 1744891101352721.jpg (6 KB, 241x209)

6 KB JPG

>>107006087
>he doesn't have a fiber optic fusion splicer

>>107007838
Any good crimper has a strip-trim combo cutter that leaves exact right length of wires.

>>107007871
If they aren't cleanly cut at the end they can short the wires too. I've installed thousands of rj45 connectors professionally. Pass through consistently cause problems. Most troubleshooting of cables involved replacing somebodies passthrough hackery.

>>107007892
>shielded
retard
unshielded twisted pairs are shaped like this to easily filter out external, uiform emi.
shielded cable without grounded connectros on both ends only makes internal emi bounce inside the cable creating distortions that are harder to filter out.

>>107007938
I have yet to see one that doesn't knick or smash insulation, or cut clean. I'm saying this as having worked professionally in low voltage installation. There's too many variables between manufacturers and cable types. You might be able to find high end German or Japanese tools that can do it, but electricians scissors will consistently cut clean.

>>107007838
Seconding avoiding pass through connectors, I've had more cables fail with pass throughs than any other means.
>>107008006
Never had to deal with NEXT failures have you?

>>107006207
You can buy cables made out of very nice thin and flexible braided copper for patch panels, and you can get them in a huge variety of lengths. If you're doing stuff in an enterprise setting, it's quite literally cheaper to have high quality prefab cables than to pay a guy in operations to make them. Organizationally it's better to have different styles/color codes for things because it saves a lot of time, and that's easier when you're buying pre color coded crap and storing thousands of them in dozens of different colors/sizes on a cart.

One exception can be crimping your own male ends on stuff that goes into permanent fixtures like WAPs, cameras and such. If it's outside, in a dusty area, or just plain difficult to access without shutting down part of a factory, it's better to go straight to a male connector and bury that inside the waterproof housing on the device than to have a female port you have to use a stubby patch cable for. Fewer points of failure in that case. For basic patch panels though, it's not worth crimping your own unless you need very specific lengths that are more than 5-10 feet.

If you need a bunch of 13ft cables with some 13ft 6 inchers to cleanly strap a bunch of stuff down between two racks, then yeah, it's easier to crimp them yourself, but that's a rare situation, especially these days because those situations are often using fiber. We cut our own fiber, but there's a lot less that can go partially wrong with fiber. It either works, and will work until you physically damage it years/decades later, or it doesn't. RJ45 has way too many states where it degrades without hard failing, and it can be difficult to track down what is actually causing the issue.

>>107008006
>You'll need to make sure your connectors accommodate this type of cable.

Thanks for your braindead reply and fauxscience.

>>107006062
Yes, it takes me 15 seconds to crimp a cat6 cable
Well worth it, especially since you can drill smaller holes through shit.
Unfortunately, now I am moving to fiber and crimping that as a fucking nightmare. So I buy cables.

just put clear varnish on the end of your passthrough connectors
no more issues

>>107008067
Sounds like you grew up on a farm.

>>107007892
Plenum ratings are not important in a home environment. Plenum ratings are mostly about adding fire retardants to the plastic to slow the burn rates in stuff like drop ceilings. Most stuff you do in residential setups doesn't require plenum rated cabling, and since it's substantially more expensive, it's a waste of money.

>>107008011
I'm just a lowly sysadmin that prefers to waste employers money on premade cables rather than my precious time criming when I could be shitposting instead, but when I do need to crimp something I trust very few brands not to fuck it up. Paying 3-5x more than bottom of the barrel aliexpress shit will be worth it.

>>107008006
>STP exists for decades
>random 4channer comes along and calls it retarded

>>107005973
What's up with the wire sequence in that pic?

>>107005996
>without the cripping tool.
Didn't know that was even possible.

>>107008067
>wait for shit to dry before using cables
Why make jobs take 3 time as long because you have to wait for shit to dry before you can even test a cable? Not to mention how that material is going to flake off after a few years and potentially gum up random ports.

Learn to use the traditional non-passthrough connectors. You need to trim the outer wrap the correct length, but if you can't learn how to do that correctly consistently within a day you so retarded that you shouldn't have graduated high school.

>>107006124
If you're the kind of person that buys Apple and Tesla products, then yes it is for you.
It's shit but it looks good.

>>107006062
Long cables that go through floors/walls you definitely want to crimp connectors on yourself.
For short cables that just connect two devices close to each other you can use either.
I only ever made short custom cables to reduce clutter in my cramped server cabinet, for anything else I just grab something from my box of assorted cables.
But if you want to put in the effort to make everything super neat that's fine too.

Ali Express should be fine but I would buy from a local online store because cables and tools are pretty cheap.

>>107008102
Plenum rated cable has thicker insulation. You guys ever wonder if you know enough to not put your foot in your mouth? Don't, because you don't know enough. At the very least preform a cursory internet search, ask ai, whatever.

>>107008067
>>107008167
>Learn to use the traditional non-passthrough connectors. You need to trim the outer wrap the correct length

A good compromise are the type of connectors that have a tiny separate block to pass the wires through before trimming the excess length and shoving it all up the connector.

>>107008130
Nothing in particular. It's not type a or b, just a technical mistake on a store picture. Certain equipment does have certain wiring arrangements, you know manufacturer spec for install and the like.

>>107008006
I prefer shielded because mice don't like chewing through aluminium foil.

>>107008139
Pliers can start, then you can finish pushing in the pins with a screwdriver.

>>107008277
Those are a hassle to use. I don't know how that would be easier.

>>107008277
The problem with those is that they're physically larger. You can get punchdown style male connectors, but they're also generally larger.

non-passthroughs are fast, and outside of a small initial learning hump, are very easy to use.

>>107008325
>I don't know how that would be easier.
Because you can expose more of the wires which makes them easier to manipulate.
It's just like pass-through but without leaving the ends exposed.

>>107008332
>The problem with those is that they're physically larger.
Never noticed that.
But even if they are that's generally not an issue for home use.

>>107006124
it cant run an open source operating system but neither can the china-link or the asus
pick your poison

>>107008062
uhm ackshually you TERMINATE fibre optic cables you dont CRIMP them

>>107008501
Sry sir

>>107008501
erm acktually you have to crimp the strain relief of the jacket/kevlar so you are crimping the cable.

File: 6NbQD[1].png (106 KB, 1178x456)

106 KB PNG

Do you still use WPA2 (most likely in a WPA2/3 mixed mode)?
Or have you decided it's too much of a security risk and gone pure WPA3?

>>107006234
is that supposed to be impressive?
those are kabylake Xeon numbers
i was expecting way less

>>107009199
I don't have anything that supports wpa3

>>107009199
why would i encrypt my wifi when unencrypted is 3x faster

>>107006234
Congrats, it's the same as a 2012 Synology with zero powersaving features.

>>107009199
How is this even a question
Even shit made in the last few years don't support wpa3

>>107009199
I don't run a wireless access point anymore since I so rarely used it for anything. Everything I care about has an ethernet cable plugged into it.

>>107009739
really? like what?

>>107009199
>WiFi
>secure
I don't even put a password, literally useless

>>107008501
Splice life only

>>107009199
I have to use mixed. The raspberry pi shit I own doesn't do WPA3 I guess.

>>106998011
Thanks for the reply. I have a discount on my electricity, so that's not much of a concern. Ideally it'll be 24/7, probably no more than 20 people at a time for different games. I hadn't thought about the router/modem part, do I'll have to look into that.

>captcha 44444

What do you guys do for backups? Do I just get a spare drive and store it in a bank vault?

>>107007871
>some fleck of conductive dust got into the ports on the switch
>moisture got in and corroded them
Seems like the pins on a non-passthrough connector would be just as susceptible to these.

>>107010476
I have everything on my server synced to a "NAS", which is really just "an old machine with as many old HDDs as I could stuff into it". I also sync the more important stuff to bare drives (they sell HDD-sized plastic protection boxes, get some) and when I go visit my dad I leave a drive at his place. I figure if a natural disaster destroys both my house and my dad's place 20 miles from me, I have bigger problems.

I'm trying to add a hardware video transcoding device to my Dell T610. It has four pcie 2.0x8 slots (mechanically x8) that can each deliver 25 watts. I can't seem to find any Nvidia Intel, or AMD cards that will transcode in this system via NVENC, QuickSync, or VCE. Anyone got recommendations that will be able to transcode videos in Nextcloud Memories?

>>107006092
>SATA expander
why do people do this

>>107011302
Because they need more SATA ports, dumbass

>>107006092
Should have put the SATA card in the PCIe 1x slot so you'd have the 16x slot free for something better (I know you can switch it, but weird addressing shit can happen when you do).

>>107011355
use case for sata expander cards when sas hbas exist?

>>107011380
I have found only expanders with 2 SATA ports on PCIe 1x. This one takes PCIe 4x, but has 5 ports. I have no ideas what else to put inside that large slot. The only other option for me would be 2.5G ethernet, but that apparently needs only PCIe 1x, so I thought I'd save the little slot for that (if I ever need it).
>>107011430
Too expensive.
>>107009466
>>107009735
It's not great, but I blame it on the ancient PSU. I guess I could get PicoPSU, but my goal was to go low-cost and use as much of my old hardware as possible. This PC will replace 3 other devices on my network, so I'll likely save on total power draw anyway.

>>107011430
it's really quite simple
sata expander = more sata ports
sas hba = not more sata ports

>>107011684
Aren't there special cables that allow you to connect SATA drives to SAS HBA? I was considering it, but the price drove me away.

>>107011696
yeah but why do all this retarded extra shit to use sas when none of my hardware uses sas instead of just having more sata ports
fact is the sata expanders jestwerk

>>107008167
>Learn to use the traditional non-passthrough connectors
the thing is most people never will because normies (like me) crimp like 4 connectors a decade, and you forget how to do it in between. for me, passthrough is very comfy.

>>107006234
better invest some time into why your system isn't entering deep sleep my nigg, my n100 idles at 9-10W with a bunch of SSD drives connected. Take a look at your powertop output and go from there

>>107011845
and by that I mean deep C-states, excuse my brain fart

>>107011845
What PSU do you use? From what I've read, it mostly boils down to this. Apparently, ATX PSUs aren't designed for low-power idle use. This setup draws almost 2W with system completely off, lol.

>>107011887
some small seasonic 80 plus gold for itx cases that cost me around 60 bucks

>>107011430
Costs less
Consoomers don't need SAS controllers and they're scary

>>107011696
there's no special cables, a sas cable will connect to and work with a sata drive.

>>107011659
>>107011684
>>107012012
>>107011712
i'm looking at sata spanders for about 25 bucks, which is what lsi sas2008 cards go for on ebay. you can use both sas and sata drives with those. they have a reputation for being rock solid (aside from tendency to overheat), while ive read too many comments with people having issues with sata expanders, including data loss. not to mention the ridiculous price increases of sata drives lately, while sas has remained largely unaffected by it. seriously, am i missing something or is this a classic case of consoomer "sticking to what he knows" because sas too scary?

Is there any risk involved with running Linux on Dell rack mounts for WAN-facing services? I had a bunch of Optiplex's and the amount of sneaky bullshit the BIOS/firmware/remote admin tried to do was concerning.

>>107006234
What size is the PSU? I'd guess at least 200W. Below 5-10% load, the PSU just kind of spins its wheels to keep the power on and you lose all efficiency. It's not anything to care about. The cost difference between you and >>107011845 is about $15 per year. Not really worth it to get a newer PSU just to save 10W.

>>107012439
It's 500W.
>It's not anything to care about.
That's what I figured. I also like putting old stuff to work as long as it makes sense, so this is another benefit for me. Initially, I wanted to use my gf's old PC as HTPC, but it didn't have support for modern codecs, and was idling at over 60W, so I decided against it, and upgraded slightly. Overall, I am pretty happy with this setup.

holy crap, I *finally* managed to get local Vaultwarden going. it wasn't too hard to run generally, except for Bitwarden Firefox extension on Android. turns out Android Firefox is super picky about the way you should handle your SSL certificates. I spent waaay too much time on that, but at least it finally works.

File: file.png (183 KB, 960x1510)

183 KB PNG

what fallback dns services do you use?

>>107012288
>rack mounts
>Optiplex's
I rac mounted a dell optiplex and use it 24/7 as a streaming server

>>107009739
Quit buying chinkshit. Even my 8 year old phone support WPA3. I only keep around WPA2 for my 3DS.

>>107012589
In descending order:
hosts
name cache
ISP DNS #1
ISP DNS #2
4.2.2.1 (Google)

I don't think I've ever had to query the Google DNS

>>107008171
Why is it shit exactly and what's less shit for the same money?
The problem is comparable ASUS/TP-Link routers are not that different in price and I don't trust them with long term software support.

>>107008492
Well, is there a better WiFi 7 option for the money, I'm not set on buying any specific router, but I'm also not going to bother repurposing an old 1L PC into a router.

>>107012589
>>107012657
mullvad via DoT
digitale gesellschaft via DoT

because i value my privacy

>>107012810
wifi 7 is just not available on openwrt yet
banana pi has a model that runs "openwrt" but youre forced to use their custom binary so thats still not open source
if you want free as in freedom your best bet right now is wifi 6/6e, which works perfectly fine

hence the pick your poison, the ubiquiti will probably be the nicest experience

im waiting for proper wifi 7 support under linux so i can get a bpi-r4 and an sfp fibre module and do away with the isp ont to plug the main fibre cable straight into my router
it will likely take some revisions though, the current model has pretty poor rf performance with terrible interference due to a lack of shielding

>>107012589
root

>>107012657
4.2.2.1 is level3, not google. google is 8.8.8.8.
>>107012810
I have a w1700k. It will likely be the first well supported wifi7 router, but right now it requires more tinkering than you want to do. I'm enjoying 6e speeds on it though.

File: 71s750wXhmL.jpg (168 KB, 1600x1600)

168 KB JPG

>>107011659
>I have found only expanders with 2 SATA ports on PCIe 1x
You haven't looked very hard

>>107012269
>sas2008
Get a 9300 for $30 or 9305 for 60. 05 doesn't need a fan, just a bit of passive airflow, while the 9300 does.

>>107013164
picrel can't possibly work, can it?

>>107013349
They do, just v e r y s l o w l y if you have all disks running (20/1gb=5mb per disk). I used a 1x to 10 sata for a few years without issue besides speed before I learned of hbas.

>>107013164
is 9305 the lowest powered one overall / per number of ports? i'm guessing none of the Pcie2 options are power efficient, right?

>>107005973
Is it possible to have a an ethernet cable where the plastic tab doesn't break off within a year?

>>107014134
yes, that's basically all of my cables. try not being a gorilla.

>>107014146
>try not being who you are
Yeah, real useful advice there, dumbass.

>>107014134
no, how would they sell you more cables if the tabs didn't break?

>>107014091
9200/9201-8 and -16 both use 13-16w
9300-8 is 13w, -16 is 26w (it is 2 sas3008 chips on one board with a plx switcher)
9305-16 and -24 are 16w

>>107014495
thanks

>>107007871
put some transparent nail polish in the end, or if in a professional environment conformant coating

>>107010669
Dust gets pushed to the back of the port and off of the contact points when you insert a cable. Push through connectors have the cables flush with the end, so anything caught in the void back there can contact them.

Moisture is harder to get into a standard connector because the only way for it to do so is to creep through the crimp, which on a decent one is going to be largely water tight. It's possible, but you can dunk normal connectors in water and they'll come out fine. Pushthroughs expose the ends of the wire to the air, which is inviting corrosion over time. The port of the switch doesn't necessarily have decent airflow either, so if you have moisture going in, it might not escape in a timely fashion and just sit there.

It's not that normal connectors are ideal. No connection is, but the actual connections on better connectors should be corrosion resistant and the connector should properly shield the cable from the environment. Pushthroughs aren't merely non-ideal, they actively expose the cable ends.

>>107016083
see
>>107008167
Why go through workarounds that take more time, money, and don't even truly eliminate the additional points of failure when you could spend a few dollars and teach yourself to use normal connectors correctly on some scrap cable while watching sportsball games or some other idle time activity of your choosing? The crimp tools for normal ones are cheaper too because they don't need blades.

>>107016424
https://store.ui.com/us/en/products/uacc-rj45-cover

Should I get a 9211-8I or can I cheap out and buy an m.2 ASM1166

>>107016999
get the 9211 off that one chink off ebay and you can have best of both

>>107016961
That doesn't address anything I said. At home you can typically just blow the dust out of ports, but in various enterprise settings you cannot rely on a switch being in a reasonably clean or dry environment. All it takes is a bit of graphite dust or stray metal shavings to be sitting on something and to get knocked into the air during a cable insertion and you can create a failure that might trigger months or even years in the future.

Back in my days doing network installs I saw degradations and outright failures because of pushthroughs multiple times a month. Pushthroughs offer no advantages beyond letting complete retards make patch cables for things they probably shouldn't be making them for in the first place, and in the right conditions the results can get very expensive, very fast. We had multiple mid 5 figure jobs where the main problem wound up being one or two bad patch cables in a wiring closet that took several days to cleanup enough to properly troubleshoot and break apart into workable subsections.

>>107008256
>Plenum rated cable has thicker insulation.
Which doesn't matter for a home use setup. If you need shielded cables or want thicker insulation, get 6A, 8, or run fiber. Plenum 6 unshielded costs about as much per 1000ft spool as 6a shielded riser.
>Don't, because you don't know enough. At the very least preform a cursory internet search, ask ai, whatever.
The fucking irony.

>>107010476
I had a NAS that had my data mirrored, when I got my backup installed my important data was backed up to a 3 way mirror, and my NAS became RAIDZ2 box.

>>107007725
>>107007753
>>107007771
it's a lot better now thanks :)

i previously thought band steering was the router steering each device to the most appropriate band for the moment
turns out it's not, it's "try and get everything on 5GHz as much as possible!!!!!"
fucking retarded shit, can't believe i suffered over a month with this

File: Hang in there.jpg (64 KB, 736x736)

64 KB JPG

>>107018404
Short of an AP hard dropping a client, it's always up to clients when to switch bands or roam to other APs. In theory this is a good thing, but some clients are retarded and have really idiotic weights attached to things. iPhones in particular cause all kinds of issues with roaming because they absolutely hate swapping to lower frequencies, so mixed wifi networks (or even just walking far enough away from an AP on the periphery) can cause hiccups and momentary drops before the phone gives up and rejoins 2.4.

Wifi 7 partially fixes this by allowing device to connect to multiple frequencies at once, and that in turn keeps things stable in partially degraded states where one of the frequencies is trash for whatever reason, which provides more leeway on roaming.

>i previously thought band steering was the router steering each device to the most appropriate band for the moment
>turns out it's not, it's "try and get everything on 5GHz as much as possible!!!!!"
2 points here. Actual enterprise WAPs will nudge devices in both directions. A lot of cheap shit does turn it into "try and get everything on 5GHz". The other point here is that if you have a "proper" wifi deployment with multiple APs so you have good 5GHz coverage, you do actually typically want things on 5GHz (or even 6).

The problem here is probably a combination of you not having excellent wifi coverage, and your equipment being retarded. If one or the other were untrue it probably wouldn't be an issue. It might still be an issue if your clients suck (see above about iphones), but if you had multiple APs and had roaming setup correctly it would most likely be completely fine.

Shit like this is why I tell people to get a cheap enterprise WAP and run their wifi off of that. It's honestly fucking astonishing how many problems magically go away with a decent piece of hardware.

Welcome to the hell that is wifi, anon. Look into if your router can be flashed with openwrt or something.

>>107018971
>but if you had multiple APs and had roaming setup correctly it would most likely be completely fine
I had multiple APs (ubiquiti) at my house and for some reason all 2.4ghz devices ended up on one and all 5ghz devices ended up on the other.
No settings I did changed that behavior

I use one AP now and still get almost gigabit on my phone in the driveway

OpenWRT + DAWN
https://github.com/berlin-open-wireless-lab/DAWN for band steering.

>>107019001
I vaguely recall some bug with unifi APs that could cause behavior like that a while back.

File: 20251026_234733.jpg (1.98 MB, 2992x2992)

1.98 MB JPG

my network is 100% sane, stable and reliable
meanwhile uniplebs will have their entire network drop from flow control

>>107017409
>uhm akshually if I ignore some things you said, their context, and add some others you didn't, I can pretend I'm winning an argument that was only happening in my mind
>the fucking irony
You're telling me.

>>107008256
>>107019694
M8, it being thicker doesn't make it magically block EMI. They support the same data rates. It's a fire spec for how much toxic shit it can release when burnt.

Did you give yourself brain damage huffing the smoke released when you burnt the riser cables you stole for copper?

i have this extremely annoying DNS problem that i cannot figure out. it seemingly resolves itself and is fine for long periods of time but eventually i find myself in a situation where it just stops working for a long time. it's because my setup is a weird combination of pieces which i have to explain because you might be able to help me if you understand.

for reasons unsaid i cant directly access the main router where i live, rather it's a wifi situation. so i have an asus nighthawk in bridge mode, it then serves a single ethernet to a microtik routerboard which then serves ethernet to all my machines including my main desktop where i'm having these retarded DNS issues.

i need to be pointed in the right direction as to what to look for to figure out which part of my networking chain the issue is happening from, and why. thank you and sorry.

and its beyond turn em off and on again, i have done so, in many different orders and combinations. what is the ideal, first computer, then routerboard, then nighawk in bridge mode to router, vice versa makes more sense to me but i could be wrong.

>>107006062
Are you better at making cables than a $25,000,000 machine purpose built to make cat6 cable? Probably not

>>107020491
>>107020035
Sounds like apartment wifi, or you're leeching off a neighbor. Is everything NATed behind your router, or are you letting the main network assign IPs? Are you encrypting your DNS traffic?

Brute force solution is probably to setup a NAT and run a local DNS server if you can't force things to go through to google or whoever.

>>107020562
Misquote. Meant to quote
>>107020491
>>107020501

>>107020491
Why can’t you access the main router? Sounds like your bridge just goes down because it’s shit

>>107020550
Yes, as a matter of fact

>>107020501
alright so i should drop the nighthawk for something else?
>>107020562
yes apartment wifi, my upstairs neighbor / landlord is the one with the main router and handles the billing. technically i can access it at some point but not randomly in the middle of the night when somethinf stops working. i need to figure out a better solution like installing my own device up there at some point. i mentioned that because ideally i wouldnt need to bridge and could just run cable to the modem/router up there.

anyway so for the time being though i need to figure out how to fix this. i want the main router to assign IPs, as for DNS i dont know, i could maybe setup the routerboard to handle it or use my old raspi but this might be more effort, i can just let DNS be handled by the main router too.

could just be something with w10 locally cucking me, i tried disabling ipv6 momentarily and manually setting everything but it made little difference.

i mean really what i wanted was the nighthawk to just be a strong attenad repeater just to bringe the internet down the pipe with a strong connection (it's mounted in a place closer to where the upstairs main router is) and then i was gonna have my own networking from there, with my own physical firewall and this routerboard serving as the main switch / hub for all my devices ethernet connections.

some variation of turning this or that on and off, resetting the bridge router from scratch, and then turning more stuff on and off, finally a webpage loaded and my connection is presently active. seems about usual. although, the hours spent doing this, not being able go online, having to phone post in desperation (cant even usb tether my 4G connection cause i'm dumbphone maxxing) it filled me with some dread mostly because it created a loop where i couldnt do any of the offline stuff i would normally do because i needed to fix this problem or it would just have to be dealt with at a later time anyway, very frustrating. i need to sort this out proper so it doesn't happen again and stop being retarded. even though it's working right now, frankly have no idea specifically what happened to give it the green light. feel like a fucking cave man right now shaking my magic cubes until they do magic again

>>107007725
>>107018404
but it's not "band steering" between 2.4GHz and 5GHz. it's AP roaming, meaning if you have multiple APs your device will hop between them based on signal strength, without losing connection.

>>107009199
MY SSID is my password anyways

>>107018971
Wifi is hell, fuck trying to even figure that shit out.
One phone will drop connection right after screen wake, then take about 30 seconds to reconnect, happens about half the time. Another phone with identical OS doesn't have this issue. Then there's an iPhone that will just disconnect mid use and take minutes to be able to reconnect successfully. I'm using OpenWRT, but I *always* had similar issues no matter what router/AP/device combo I used. Nothing ever just works and I've pretty much accepted that as a fact of life.

>>107007892
>Shielding isn't as important as good insulation.
What's the difference?

I like good old unshielded Cat5e.
Thin and flexible enough that makes it easy to pull through small holes and around corners, while GbE is low frequency enough that it's not likely to ever have interference issues. You might Need More, but I sure as hell don't.

>>107005973
Is Shimmie2 or Szurubooru really my best bet when it comes to hosting HDR images? Native or gainmapped, most applications seem to strip the necessary metadata, which results in "washed-out" images for native Rec2100-PQ images, or only the SDR version, in the case of gainmapped images. Immich comes close, but you have to zoom in to get the original file first.
Szurubooru (and its fork, Oxibooru, if it would allow for AVIF uploads) is a workable solution so far, but it is very locked-down and uncustomizable.

Is it true you don't want to mix 5400 and 7200 drives in the same jbod/shelf or does it not actually matter?

>>107021963
To elaborate I mean in a physical damage kind of way not performance/speed, as in will the speed disparity cause some kind of vibration or whatever that would cause issues or damage to the drives.

>>107008047
>>107006062
What's wrong with passthroughs?

I have to roll out 80-100 patches from all-weather covers every year outdoors, some of the cables live outside year-round (strapped to fencing, not worth the time to take down versus just re-crimp it if corroded over the winter)
I don't think we've ever had problem that would be explained by them being crimped with passthroughs, and some of these are well over spec (~190m) working at gigabit + poe.

>>107021963
>>107022043
It won't cause any kind of vibration issue, but RAID5 and RAIDZ type topologies are typically constrained by the lowest performing drive in a given array. 5400 and 5900 RPM drives have lower IOPS and (usually) lower linear read/write speeds than 7200 RPM drives, so if you made a RAID5 with 2 5400s and 2 7200s, the 7200s would spend some time waiting for the 5400s during heavy load.

For mirrors a properly implemented RAID controller (software or hardware) will distribute read operations between the two drives based on latency. Write performance will be bottlenecked by the slower drive, but reads will somewhat prefer the faster one, and in a mixed workload reads will fairly strongly prefer the faster one.

>>107021079
Honestly most people don't need more than 5e hell most businesses don't realize their contractor cheaped out and ran 5e riser.

I need to run a LAN cable from floor one (my router) to floor two (a switch).
What cable do I use? Shielded or unshielded?
I am reading mixed opinions on this. Some say this, some say that. Some say shielded, but connect the shielding to the ground on only ONE floor, never both.

You guys seem very knowledgeable. Do you have some advice for me?

What do you guys use to make sure you're aware of the versions of services you run that also notifies you of updates you should probably get to? I'm thinking of opening 2 services, but I also don't wanna get clapped by CVEs.

>>107022713
Cool, yeah they will be on different pools so no worries about performance like I said I was just worried about physical damage cuz I could have swore at some point I read that it was possible when mixing speeds in the same enclosure. Though it could have been whoever wrote that misunderstood the reasoning, or was even just making shit up (on the internet no less, who would do that?!), or maybe I just hallucinated the whole thing, who knows.

>>107023208
My important and exposed stuff is setup to auto update every 2-4 hours but obviously that depends on what software we're talking about and how updates are distributed. apt/aur/git are super easy anything else depends.

>>107023208
that's a bit vague. i usually deploy behind a proxy and inside containers to limit the attack surface. if it's specialized complex software i'll be on the mailing lists etc and will just see the CVE discussion there. More corporate oriented distros will also publish CVE stuff in same way or another.

>>107023430
Vibrations can negatively impact drives, but modern ones have all manner of systems to compensate for that.

https://www.youtube.com/watch?v=tDacjrSCeq4

File: 1730958594603689.png (187 KB, 647x674)

187 KB PNG

I don't know what any of these words mean, or what any of you are talking about. I just have my mother in law's old office Dell with some 3tbh HDDs added, and I wiped + debloated windows and then made a mirrored drive with the HDDs, and shared it to the network. I ran a ethernet cable between my server and my main PC to make it faster when I move media to the shared network drive. I also run a minecraft server on it and I installed jellyfin and watch my movies on the smart TV with it.

File: EqRWOKiXIAAVIJp.jpg (27 KB, 464x340)

27 KB JPG

>>107024746
I kneel...

>>107008115
Spanning Tree is a dogshit protocol and mechanism to fix a dogshit protocol (ethernet) broken by retarded mechanisms (flood and learn). People who deal with 60 device networks don't get an opinion on this, for the vast majority of enterprise the hardly ccna level sysadmin shitters could set everything up with static entries and it would be perfectly fine. No shit they all think there couldn't possibly be problems with STP, or multi chassis LAG or any other number of niggerlicious Ethernet copes.

>>107024746
You're 10x as based as anyone here.

My server (thin client mini PC) still runs Debian 12 I think. Can I just upgrade to 13 without fucking things up?

File: 1305257774668.jpg (19 KB, 275x250)

19 KB JPG

>>107024746
based
enjoy the digital backups of your legally owned media.

>>107025425
you're the nokia architect guy, right? i was gonna ask what kind of non niggerlicious alternative do you envision for ethernet?

>>107024594
I just want a dashboard that tells me service w is running within parameters, service x is down, service y has an update.

>>107023062
pls reply

>>107023062
it doesn't fucking matter

File: image.png (1016 KB, 1440x1080)

1016 KB PNG

>>107005973
>RJ45
What century is this?

>>107023062
Doesn't matter, just run some cat6 and call it good. If you can do some 6A I would say go for it as it might be more future proof but you're probably fine without it

File: banner.jpg (183 KB, 1200x797)

183 KB JPG

>>107026445
>CS connector
Huh, never seen that one before. Anyway, twisted pair still has its place for anything that can use PoE or only needs 1G (or possibly 10G) speeds.

>>107025425
>FUCK I JUST CAUSED A BPDU STORM IN THE BRIDGE DOMAIN MY MOM IS GONNA KILL ME
remember to configure your priorities and properly protect your interfaces and practice good operational hygiene. go ahead and shill your ethernet replacement, we can all set the setup from a mile away...

> be me
> have a self hosted Nextcloud server running on a Dell T610 for storing documents and terabytes of family photos/vids
> Streaming videos from server to mobile client devices sucks ass over cellular data because Nextcloud wants to stream the full quality video
> look around for compatible GPUs for video transcoding so that videos can be streamed at lower bitrates without buffering
> Find nothing from Nvidia or Intel, only option is the AMD Radeon Pro W6300
> Runs on PCIe slot's 25 watts of power
> Has hardware for VCN 3.0 encoding, which is supported by Nextcloud Memories VA-API
> Only comes in x16 length form factor, and the T610 only has x8 long PCIe slots
> Fuck
What are your thoughts on the Radeon Pro W6300 for hardware video transcoding in Nextcloud Memories? Should I just use a dremel on the PCIe slot to open it up?

>>107007892
how the actual fuck do you know what a cable listing is actually going to send you. how do you know what any cable you pickup is made of. I just trust the chinks on the other end of amazon made an ok cable.

>>107027168
you buy from a cable company or find a brand you trust. LTT shills infinite cables sometimes and the one time I bought something from them I was happy.

>>107027168
Buy from reputable manufacturer such as truecable, cable matters, monoprice, and a few others. Go with whatever is cheapest for your spec needs unless you have familiarity with them. Last I checked, unifi actually had decent prices on 1000ft boxes of basic cat6 cable.

Do not buy stuff on amazon unless it comes directly from a reputable manufacturer. Amazon is full of chinkshit copper clad aluminum. CCA is "fine" for data transmission because high frequency signals will tend to wrap around the outer layer of the conductor anyways, but it has a higher resistance per foot, so it's awful for any kind of power delivery. On longer runs the voltage sag can be significant enough that many PoE devices will not function. CCA also corrodes much more easily because any minor scratch in the copper exposes the aluminum underneath which creates attenuation issues with the signal, and in some cases can allow corrosion to spread underneath the copper layer.

File: Snapshot_3.png (499 KB, 1920x1080)

499 KB PNG

is this fun to mess around with? i want to learn enterprise networking but i don't have a job in enterprise networking.