/hsg/ Home Server General

>why doesnt my proprietary black-boxed garbage not work

Attached: 1515309052501.jpg (1257x1790, 2.38M)

well at least now i know why it was dirt cheap

honestly, i think its too bad it doesnt work properly
i would assume the reason for ip and app access being different is because they want you to use the app so they can steal and sell your info

Attached: 1512049520231.jpg (953x960, 106K)

correct, and the article mentions neither MS not Google found correlations across all models and brands

what protocols does it use?

based

Attached: 1556796013386.jpg (1732x1200, 1.48M)

Got a crappy working computer from my professor for free. Has Windows 10 and about 500GB of storage. Looking to turn it into a storage for the house, is there any recommended guides for this?

install debian linux
install samba
add as a share in explorer

Then you can have a centralized network share for all Windows and *NIX devices

i'm honestly not gonna waste time finding out, i just found out this article that says shit like:
>If you're expecting any NAS or network appliance functionality, you can forget about it. You can't even set a static IP address for this device. For some reason, devices plugged into the My Cloud Home's USB port are not mounted as network resources. You're given the option to import content from them but not the other way around - and bizarrely, this can only be done through the mobile apps. There's no FTP or SSH access, no downtime schedule to save power, no per-user permissions, and perhaps worst of all, no easy way to either back up the contents of this drive or use it as a target for third-party backup software.

The Public folder is also for some reason hidden from all the apps, and even the virtual Z: drive in Windows Explorer. There's no storage quota and no security, and the apps can't show you how much space its contents are occupying, all of which make it a poor backup destination. At least it's a standard network share which could come in handy - and yes, it worked just fine without an Internet connection, exactly as it's supposed to. We just wish we could create more folders and manage permissions, bypassing the apps altogether. Clearly, since WD allowed these two exceptions, it could have gone all the way.

Sounds easy enough. Gonna get to work on that.

sup hsg
had a break in yesterday but they didn't take my shitty old servers of course, just a couple guns and laptops

Attached: hsg-20190507.jpg (2962x1500, 1.94M)

Why is there a person in your server and what is she doing?

>The Public folder is also for some reason hidden from all the apps

Not trying to insult you but what do Zoomers mean when they write stuff like this? Do you mean that programs cannot access any directory on your network share? Since you write virtual drive I expect it to look as pic related? By quota do you mean space used? I understand in quota to allocate limited amounts of storage or traffic/time to a user account.

Attached: 363426436.png (275x210, 10K)

I'm sorry user, what guns did you lose?

Attached: 1556677113590.gif (480x480, 3.07M)

Stand your ground. It's bad to think what the thieves might do with them.

Attached: Georgi-Zim.jpg (460x276, 18K)

So, I'm making my first server, and I wanted a Media/Emby/NAS box with optional SSH Socks proxy, DDNS and maybe some other use-case stuff as I go on.

I went with CentOS 7, installed on an old Phenom II x4 945 rig I had in my closet and loaded it up with 3x4TB hard drives and a 64GB SSD for the boot drive.

I made a Raid5 array with mdadm and formatted it ext4.

Everything was going great, I had set up emby and samba sharing, worked out the permissions and set emby to cache and run transcoding on the RAID array.

Then, last night, I got a kernel panic and my the RAID5 shit the bed.

No superclusters on any of the drives.
after spending an hour looking it up, I decided to force re-create the array with --assume-clean.

It has been re-syncing for the past 15 hours.

What are the chances I nuked the contents of the array and I'm gonna be spending my weekend re-downloading all my anime and ripping all my blu-rays and DVDs?

The drives are all mostly new. Less than 200 hours according to SMART tests, and they're not all from the same run, so I doubt they all failed.

Is software raid just plain unreliable, or did I fuck up?

I'm not very good at linux, but I'm learning enough to do what I've done so far.

Any recommendations?

Attached: rustled hard.jpg (569x596, 125K)

No idea friend, but I feel your pain. I always wonder when shit like this happens to what degree it is either the hardwares fault or the shitty support of Linux for certain hardware.

Could you at least get some info out of the kernel panic what may have caused it? More important than fixing this mess is to make sure it never happens again.

Good luck, I hope re-sync finishes soon

oh, i know exactly what they meant in the article (sorry, i just pasted everything after the first line) there really is a public folder that you can only see if you access through the ip, as i said, there's previously made folders that you can see only through the ip, but not in the app, and the other way around, some family/something/something folders you can only see through the app, it fucking baffled me, so i fixed an IP to the MAC address of the thing (you can't configure the piece of shit for real) and have used it like that so far, but i really got tired of trying to work it like other NAS i've used before, it has access to the files through internet browser but not to configuration, like an EX2 or whatever it was called

> 3x4TB hard drives
> RAID5
Bad idea because it'll sync forever.
> I decided to force re-create the array with --assume-clean
Assume-clean means that mdadm will assume disks are empty and block any attemtps to sync from that disk.
> What are the chances I nuked the contents of the array
Depends, you did it for all disks? It will be empty. You did it one one disk which was problematic? It should be OK, but then again the array should be mountable even if degraded, if it's not - sorry.
> No superclusters on any of the drives.
You mean superblocks? Now I wonder how you managed to create it. If you happened to have the output of mdadm --detail /dev/md*, it should state right there "Superblock is persistent".
In the meantime, prepare for the worst. Software RAID isn't unreliable per se, some prod servers I manage use it. But - it's always good to learn about failures, they mean more than success stories.
Do you have any output on dmesg or /var/log/messages which could explain what has happened?

It's still syncing, so no.
Oh well, I have backups of almost everything, and what I did move instead of copy was off of external drives with no other activity, so I should be able to recover them.

Actually, I'm not sure if I did assume clean...

I did a lot of things that didn't work before finally giving up and re-creating the array.

I do remember it said "superblock is persistant" on all drives, but mdadm couldn't find them.

When I tried to re-assemble, it said each drive was busy.

Oh well, cross my fingers and pray to the machine gods.

this makes me want to share this so i'll just do it
>this shit happens on the weekend at the office
>they left very quickly as the alarm went off
>left pretty much everything intact
>took my boss' laptop that was sitting on his desk... because he didn't close his fucking door despite having good locks
>they broke open my boss desk and two others, nothing was in there, as everyone else had work left to do and took their laptops that day (fucking lucky)
>there's two new guys at the office, one got a band new laptop and the other a shitty 4GB ram hand-me-down that i had just repaired (broken screen)
>new laptop was left on the desk, old was left inside, but they didn't give the guy the keys yet
>one of the girls put the new laptop inside the desk, this guy DID get his key, but he just left the laptop out there... and the key perfecty stcuk on the keyhole
>they totally skipped his desk because of this and stole the shitty laptop instead
why are the uncaring ones the lucky ones? motherfucker

Attached: FedUp.png (520x596, 215K)

Rapid fluctuation kills drives, but only with regards to the drive temp itself, the drives probably don't cross below 30 at night, and having hours and hours of time to change ambients shouldn't be too bad. The pathological temp fluctuation example is when a drive is kept at 38C in a server rack for two years of nonstop use, gets pulled and cools down to 20C in about an hour at which point it never runs again. A buddy of mine jammed some drives in the back of a machine learning server, and the wild swings in temp as the processors received and completed workloads cooked em all in a few months. If you are worried, write a script that polls the SMART data of the drive every 15 minutes and graphs the temps out. There hasn't been a whole lot of good research done of drive temp fluctuations so most recommendations are apocryphal at best, but there is no question that fluctuations can and do kill drives.
Regarding your second questions, HDDs use either SATA or SAS to interface with a controller. Since your laptop probably doesn't have a SAS controller that leaves direct SATA (eSATA), or SATA over USB. Both are fine, you don't lose meaningful performance in sequential reads over USB3 unless the drive already sucks. Random 4K benefit immensely from eSATA's lower latency.

Thank you
I monitor it with Crystaldisk and hwmonitor for real time graphs of all the hardware.
The problem is that is an old MSI CR620 (i3-330m 4GB ram) but is enough for Tixati. I have lots of 1TB 3,5 hdd and would love to connect one trough sata on the unused optical drive port
>pic related

Attached: Screenshot_20190506-193626_Chrome Beta.jpg (1437x837, 195K)

There are a bunch of optical drive (Slimline SATA) -> SSD adapters that accept any 2.5in drive that's thin enough, you could probably get that and a SATA power and SATA data extender cable and throw together something that works.

How does Jow Forums feel about the Dell 12th gen servers now that they are starting to get cheap. I got an R320 a few months that I am in love with, it is quiet and doesn't run up my energy bill while still handling everything I throw at it. I also got a 620 from work that is pretty fast, but makes my apartment way too hot so I don't run it too often. Both absolutely dumpster my old 710 in terms of performance, power draw, and noise.

is there a way to have a vm natively write to a hdd (sans a 2nd machine) so that you can move the hdd between systems hassle free? i don't want:
vm files>fs>datastore>host fs>disk
i want:
vm files>fs>disk

Add a SAS/SATA controller and pass it through to the VM. There is also a way to directly pass disks through in ESXi but I don't remember the steps off the top of my head, and it's difficult and inconsistent.

> There is also a way to directly pass disks through in ESXi
Its RDM (Raw Disk Mapping) for anyone interested. You have to SSH into your ESXi server in order to find the drive addresses and then create a type of hardlink which maps directly to the drive you can then mount as a SATA/SAS drive in your Virtual Machine settings.

any good cheap lowpower nas parts recipe out there?
i dont mind spending a bit of money but i want it to be really low power so i can have it 24/7

neat, thanks

How low power is "low power"? You can push Dell servers below 60W with the right settings, and those Synology guys are as low as 15W. An RPi or some other SBC (odroid or whatever) would be a mediocre server but a passable NAS for less than 10W power consumption. If your electricity is

Yes. The Rock64Pro or Odroid XU4/HC1/HC2/NC2 are all low power and great for 1-2 drives each (still useful with 4 UASP USB3 attached ones, semi-messy as that may look depending on how you set it up).

If you prefer to stick with Intel or want moar drives, grab something like a AMD 200GE or Ryzen 2200/2400GB or low power Xeon. Up to a certain drive count (...and with no fat GPU) you can save some power by using a PicoPSU or Chinese clone thereof.

The earlier ARM SBC will run at like 5W; the later setups can be something like 15-25W. That's without spun up HDD.

PS: I only listed the pretty ez versatile setups that will work fine to saturate a GBE line when acting as NAS and that also give you a lot of options what to run.

You could further reduce power consumption by using a smartphone or one of the more suitable competitors to a RPi Zero or by just replacing your WLAN AP with one that also takes over the NAS functionality. They just won't necessarily saturate GBE, look physically pretty and/or serve additional roles as-needed.

Pieced this piece of shit together with mostly free stuff salvaged from deserted office buildings.
Top to bottom:
ISP supplied modem
asus rt-n66u - wap running dd-wrt
random d-link switch
shitty tn panel
dell optiplex 980 - pfsense router (lol)
dell optiplex 980 unplugged
colour laser printer
openbsd box 4x 4TB zfs; i3, 32gb
APC smartups 1500va
2-post rack w. the panduit cable things

salvaged everything except the openbsd box i built. My electricity bill is included in my rent.

Attached: shit_server.jpg (2988x5312, 3.02M)

So my new case arrived; can someone please explain how 3-drive hotswap bays work? Why are there only 2 power connectors when it fits 3 drives? Also what's the connector with all the pins? Is that for the fan? Why isn't it a regular fan header?

>dude just read the manual
It's a used case. I didn't get a manual.

Attached: IMG_20190508_143104.jpg (2852x1667, 1.76M)

it's not a connector, it's an array of GPIO pins
There's a jumper on the second set, you move that around to change whichever settings. Right now it looks like it turns the fans on when it hits 60C. You'd need to find the manual for anything more than that.

Thanks.

>It's a used case. I didn't get a manual.

have a look for serial number/manufacturers name and give it a google - lots of things have manuals online.

2 for 3 power connector isn't that bad, just each will draw a bit more than normal from the psu, no great concern - if it was 1 for 5 perhaps it would be.

Speaking of PSUs, is it generally a good idea to go gold for an always-on server?

ideally you want the most efficient you can get, but in reality it depends - idling at high efficiencies don't net you much savings in cost/heat, under load they definitely do.

Reliability is much more important - it may well be running for years without much cycling.

If you can make sure that you also put the PSU under 50% stress during server loads, savings would be even more.

Most PSUs suck during high and low loads.

>You can push Dell servers below 60W
quoted as example where the PSU would have hard times to save power.

Attached: 1.png (492x295, 92K)

But of course I exaggerate like the graph exaggerated:

You probably only lose 4% efficiency if just putting it on 20% load.

>SAS
Random browsing through, I always found SAS to be either expensive or too enterprise for home use. What are pro and cons?

I want to setup a little service for streaming my collection of MOD tracker so i can listen to it while i'm at the job. I also want to be able to skip tracks. Which software stack should i use for this?

Attached: 1555690579339.jpg (600x600, 113K)

Thoughts on this build? Mainly for media streaming and VMs.

CPU: I am currently deciding between the Ryzen 2400G and the Ryzen 2600. I am still trying to gauge what kind of advantages the more expensive 6-core, 12-thread CPU entails, and whether or not it's worth the slight increase in price for me.

Motherboard: I'm going to choose a B450 Motherboard to avoid potential compatibility issues with 2XXX-series Ryzen CPUs. I'm choosing between the MSI B450M-PRO-M2-V2, the ASUS Prime B450M-A and the Asus TUF B450-PLUS. The first is the cheapest, most affordable option, the latter two both have 6 SATA ports whereas the MSI only had 4, and the B450M-A has ECC compatibility, which is something I'm still not sure if I want. I hear a lot of people expand their SATA ports with PCI-E Controller Cards like this.

Ram: G.Skill Ripjaws V 8GB (2x4GB) 2400MHz DDR4 - Nothing too fancy. Cheap DDR4 server solution.

PSU: Corsair - VS 450 W 80+ Certified ATX Power Supply - I'm still unsure as to how big of a difference going gold would make to long-term cost. If it's significant, I could always bump up to a SeaSonic - FOCUS Gold 450 W.

2400G
ECC-RAM & board
>hich is something I'm still not sure if I want
Not wanting extra features you can use or just leave turned off in BIOS? It is more, can you afford it?

Use ffmpeg. It has support for the external libraries libopenmpt, or the older libmodplug, which are needed for module files. You'll need to get or compile ffmpeg with --enable-libopenmpt or --enable-libmodplug.
lib.openmpt.org/libopenmpt/
You could convert a track to a virtual file system in RAM, like 64MB RAM reserved, then stream the OGG/mp3 from there by standard means of creating a m3u stream for browsers.

As far as price difference between motherboards goes, it's negligible.
According to pcpartpicker though, none of the ECC ram modules on that site are compatible with the 2400G or the Motherboard. Despite that, the official specs of that motherboard list ECC mode (depending on the CPU): asus.com/us/Motherboards/PRIME-B450M-A/specifications/

I just don't get it, and not sure I should be sticking my head here.

What, self shot not permitted where you are??

>none of the ECC ram modules on that site are compatible with the 2400G
Damn, that's a fucking buzzkill. Alright just get a compatible board then. Remember it as future upgrade option and you already have peace of mind. When you need it, it will be there to use. Should you ever find DRAM modules supported.

>Alright just get a compatible board then
For what? The CPU or ECC?

If you're referring to ECC, if pcpartpicker is anything to go by, it's already out of the question. The cheapest ECC compatible motherboard is way beyond what I want to spend on a mobo, like over 4 times. Plus, there's a change pcpartpicker just made a mistake, the Asus website literally lists the B450M as having ECC mode. As for pcpartpicker, the Ryzen 2600 is listed as a compatible CPU, but the 2400G isn't.

The 2400G should support ECC if the mainboard does.

Fix the issue with no available RAM if you ever should bother. I just don't want you be in a dead end when you later really want that. Switching a board or CPU alter on is fucky.

en.wikichip.org/wiki/amd/ryzen_5/2400g#Memory_controller

>The cheapest ECC compatible motherboard is way beyond what I want to spend on a mobo, like over 4 times.
Alright didn't read this far. It's fucked then.

bixnood has never once posted a photo of his actual hardware

No problem!

I don't know what I'd do if someone stole my server. I can't afford to backup everything online.
How could I prevent someone from physically taking my server?

I was thinking maybe I could assemble a giant rack tower inside my room. It'll be too big to fit through the door, so the only way someone could take it would be to disassemble it, which would take hours. They won't be able to remove the server from inside it because it'll behind a locked door within the rack. Sound like a plan?

no idea, i've considered locking my case to a nearby radiator

Encrypted drive in a safe deposit box at a federal or private bank?
Gonna cost a yearly fee though. But unless you don't have friends or family for an off-site backup, don't blame my idea. You'll figure a place for stashing, like, a shitty 8TB Seagate Archive HDD. Or any HGST / WD competitor.

Attached: 3463462346.jpg (480x360, 26K)

Would an old low power AMD APU good for a fileserver and won't have any bottlenecks as long as it has gigabit ethernet? Will probably be only streaming to myself and maybe have a shared folder for myself. Those CPU doesn't support ECC ram but would that only be useful for repairing RAID?

Continuing on the idea, since I work literally in security. There is hardly any way to secure something perfectly in a private home.

The basic of security is to waste the time of a thief, as they usually spend less than 30mins inside a home:
1) Steel door to server room
2) A metal cage around the server with a stronger padlock than in the image
3) concrete strike anchors for the server rack
4) Special screws that require different tools

Over an hour to get past all.
Pretty useless to install all this, but I enjoyed thinking about a possible solution.

Attached: 34632462346.png (1166x934, 1.82M)

How about a killer Roomba?

>no idea, i've considered locking my case to a nearby radiator
If they rip it out you get a nice water damage in your home.

>How about a killer Roomba?
Dunno about burger laws, but I'm all for devices that cripple and maim thieves.

Hello /g entleman
i like to Self host my self and some friends
can you tell me what do you think about the HP Proliant ML330 G6 ?
Plan to install proxmox and LXC container
i can get on for less than 100$

Thanks for your answer =)

Attached: serveur-hp-proliant-ml330-g6-1-x-xeon-six-core-x5650-16-gigas-35-tour.jpg (800x800, 199K)

Is your waifu ok?

ebay.fr/itm/HP-ProLiant-ML330-G6-Server-Quad-Core-XEON-E5504-2-0-GHz-500GB-8-GB-DVD-26036/202636873965
Mon ami, I dunno.

Thanks

>Mon ami
i look so french ?

For me it look fine,
but i don't know how i can plug Harddrive
seem i can use 8 HDD
but do i need HP hdd case array ?

Cannot look behind the door

Attached: 51tq07a2DnL.jpg (500x500, 35K)

oups sry
thanks for picture ^^
is it easy/expensive to find this hp array HDD ?

Specs doc say it takes 8 LFF 3,5" regular sata drives. just search this on google:
HP Proliant ML330 G6 filetype:pdf

>HP Proliant ML330 G6 filetype:pdf
Thanks you very much !

good VPN service to add to my SBC (odroid xu4)?
i dont want to overload the thing and the vpn will only have 1 user (me) maybe 2 at most.

get the 2600, ecc is at best hit or miss on the apus; the v1000 boards do support it

>Expensive and too Enterprise for home use
Correct on both counts. Also they use twice the power and run very hot. Upsides is you can get 15K RPM drives, you can jam a huge amount of drives in to one controller, they tend to have lower access latency and move more data. It is the extreme of "we will make it as fast as possible to matter the cost". I think there is also something inherent to the SAS protocol that gives it benefits is a system where you can expect your drives to perform well, but I don't know too much theory.

>Upsides is you can get 15K RPM drives
Thanks but I already have hearing loss, haha.

>they tend to have lower access latency and move more data
>there is also something inherent to the SAS protocol that gives it benefits
Might be nice then to have just one SAS drive in that case.

It's kind of an all or nothing deal. Most SAS controllers eat a PCIe 8x port and can either attach to a bunch of SAS drives or protocol swap to a bunch of SATA drives, but I don't think you can mix them. That may only be true per SAS port. SAS controllers tend to run hot, especially if they are a RAID controller (most are). Unless you have a rackmount server, I would advocate against getting any SAS drives, and if you do have a rackmount, consider whether 15K SAS drives or SSDs will serve you better. My use case was jamming 4 15K 150GB SAS drives in RAID 10 and running my Minecraft server and ELK stack off it. The main benefit is that they do not die from writes like SSDs while still performing well, so using them for write heavy applications is great.

>Is software raid just plain unreliable
No, it is very reliable. I've been using it a long time now and basically all the NAS boxes you can buy in stores (other than maybe drobo?) also use the Linux md/dm RAID stuff.

I have no idea how on earth you could remove mdadm metadata superblocks from all drives unintentionally; the only thing I can think of is that maybe you used an older version of mdadm with the newer SNIA DDF or Intel Matrix format metadata?

I fucked already up, could ahve looted a SAS controller with memory and backup battery from my company with their OK, but the drives seemed too pricey at the point.

We are still getting fucked over hard since the convenient "flood". A 2TB was $45 before it happened. Fucking price agreement between DRAM and drive "cartels". Fine them billions for screwing customers.

>read news this week
>brushless DC motors for HDDs will go down in production by 50%
how convenient, get ready for the next price raise. they will blame SSDs for this. but last time I checked people just them in tandem with HDDs.

Attached: 1557237909757.jpg (750x1076, 91K)

for accuracy
techspot.com/news/79938-pc-hard-drive-shipments-could-fall-50-percent.html

My shitty HP microserver just died. I'll have to check if the hard drives are OK.

Did it power down for no reason?

What do you mean died. Like doesn't power on? I guess check each component one by one. Probably PSU related. It's rare for a CPU/Mobo to just stop working.

>when user disappears with a cliffhanger
Will we ever know if they restored their RAIDs?

AR-15 and Glock 19
I was at work. had they come when I was home they'd be in a holding cell in God's prison

Still they must know you from somewhere. They came with clear intentions to get that shit. Or someone talks to much about you.

i was thinking this at first but i've been here less than 2 months and a couple other apartments in the complex got burglarized, so if they were targeting my stuff specifically they must've been watching me move in. or they're CIA glowies

Pic related is my setup. I've been trying to setup LACP between the switch and freenas and it works for like 10min. I get increased bandwidth but then the whole network starts to slow down, things start timing out, even browsing on the desktop machine becomes super slow and devices fail to connect to the Routers WLAN.
How do I start diagnosing what is going on here? Something is obviously very wrong but I'm really stumped as to why it takes 10-30min for it to fuck up.
The switch ports are set to passive LACP and the freenas server has two intel NICs aggregated as a LACP too.

Attached: Untitled Diagram.png (548x361, 29K)

They at least watched your shedule. If your field of work allows, return at irregular times. But god forbid they target your again.

You now definitely need IP cameras, rig them to upload pictures into the "cloud". I was thinking of a self hosted cheap virtual server, so only you have access.

Or should I hope for them to come back when you have the advantage?

pyimagesearch.com/2015/06/01/home-surveillance-and-motion-detection-with-the-raspberry-pi-python-and-opencv/

Attached: pi_home_reaching_for_beer.jpg (500x375, 52K)

so do you guys have any idea if there are any games that aren't minecraft I could put on my server, give passes to a couple friends and play over the internet together or alone whenever convenient.
anything goes, rpg, shooters etc.

I assume you set up LACP using the *same* endpoints? So you have two cables from your switch to two NICs on the *same* server (point-to-point connection). AFAIK running LACP between different endpoints (point-to-multipoint connection) isn't possible. Also, if I remember correctly, on Cisco switches it's possible to set up link aggregation (EtherChannel) manually, so no need for protocols like LACP. You might want to look into that.

Any GoldSource engine game, Sven Coop adn the likes.

Mod your server with voting and a HL DJ for muisc. Should be comfy then.

Attached: 1557225906268.gif (320x294, 1.7M)