publication croisée depuis : https://lemmy.world/post/1474932

Hi there.

I wanted to run LLMs locally on my server (for better privacy), and was wondering if:

1. I could use Intel ARC/AMD GPUs - these are often less expensive and AMD has open source drivers, which is something I like.
2. If a PCIe x4 Gen 3 slot would be enough (it's an x16 slot with x4 speeds) - this is an important consideration.
3. Would 8GB of RAM (in the GPU, I believe it's called VRAM?) be enough?

I'm looking at language models to train on my Reddit and Lemmy content, in an aim to make it write like me (and maybe even better than me? Who knows). I don't quite know which models I will train, or how I will do so (I certainly won't be writing anything from scratch), but I was wondering; with the explosion of FOSS AI models, maybe something like this would be possible with the hardware constraints I mentioned above?

Does the speed of the connection between the GPU and the CPU really matter in such applications?

Thanks!

A while ago I made an app for tracking baby activities because I became a parent and was horrified at how many permissions the existing apps required and how much tracking they contained. Both the app and the server are open source.

This is a web-app which also has an Android version in the Play Store (F-Droid didn't accept it because they don't feel like web-apps should be welcome in their store). On iPhones it can be installed as a PWA to the home screen.

Features:

• No tracking whatsoever
• End-to-end encrypted, no personal information is stored on the server unencrypted
• Track baby's feeding, diaper changes, breast pumping and sleeping (more to come)

Links:

• https://baby-journal.app - the PWA itself
• https://play.google.com/store/apps/details?id=app.baby_journal.twa - the Play Store Android wrapper
• https://github.com/RikudouSage/BabyJournalApi - the server source code
• https://github.com/RikudouSage/BabyJournalUi - the PWA source code

Hey All,

Newbie selfhoster here trying to figure out next best steps for my UnRAID install. Essentially, I got a little overzealous in my setup and didn't quite provision things optimally, so I'm hoping to avoid having to start completely from scratch.

My drives: 2x8tb HDD, 2x14tb HDD, 1x1tb Nvme SSD, 1x2tb Nvme SSD

Array: all 4 HDDs, 1 parity, all formatted in ZFS

ZFS Pools: 1tb is standard UnRAID cache drive, 2tb is a slog (not certain if I'm using this term right; it's used as temp storage for downloads before data gets moved to the array)

My issue: didn't realize that the array in UnRAID doesn't allow for actual zpool/raid configuration, so writes to the array are pretty slow. I want to speed it up.

What I want to do: move the 2x8tb HDDs off the array, wipe them, reconfigure them into a new pool in raid0 (insert equivalent z-term for ZFS here). Leave the 2x14tb HDDs in the array, keeping the 1 that's there as parity.

My existing data: currently in the array, solely on the 14tb HDD not being used for parity. Nothing incredibly important, just media files and about 200gbs worth, but I'd like to keep that in tact.

My question(s): Because I have no data on the existing 2x8tb HDDs that are in the array, can I safely move them off the array and wipe them without concern for my data that currently exists on the array?

Will I have to rebuild the parity drive?

Should I preclear the drives I am looking to wipe?

Any other potential concerns I'm overlooking?

Sorry for going full noob on you all. I've been reading docs for what feels like a better part of a week, watching the guides, and I guess I'm just slow on the pickup and looking for some layman guidance. Also, sorry if this isn't the right place to ask. I've got a secondary backup solution available, so I'm not worried about ZFS and it's relatively new intro to UnRAID.

I self host all of my services but utilize a VPS as a gateway for access. Primarily to allow access to a media server and file storage for friends and family.

Recently I’ve been shut down by my VPS provider on multiple occasions because they claim my server was DDoS’d at 2gigabits/s. I don’t see any evidence of this in my logs.

Regardless, I set up Traefik proxy to geoblock any IPs outside of my country. Literally a few mins after doing so and confirming via VPN that it was working I got shut down and received an email that my network was severed temporarily due to a DDoS Blackhole event.

The questionable nature of their detection system aside, it’s got me wondering…does ip blocking actually help mitigate DDoS attacks?

The server still needs to process the incoming connection before it filters it, so I’m assuming the attack is still accomplishing it’s intent which is to overload the server. Can somebody more knowledgeable provide some insight?

My dad also used to self-host. Now I run all of the services he used to, and more.

Anyway, his server is still around so I thought I might as well use it for an offsite backup box, I run a matrix instance and nextcloud, as well as other things. But those two are the contents of which matter most.

How would you set this up?

Just a nightly rsync over sftp? That seems ineffecient. Is there a best way ro do this?

cross-posted from: https://lemmy.lucitt.social/post/77727

I made a guide on how to go from zero to hero: Self hosting a lemmy server. All you need is an old pc, a thumb drive, and some time.

Please let me know if there are errors. This is the first draft. Thank you!!!

What the title says. Are there any good ChatGPT alts that can be self hosted?

I'm reaching you all asking for advice. Given the recent news about some social networks attempting to rule my/our content I want all of it to be free forever. To avoid vendor lockout I've decided to host all my tweets and Reddit posts in my own site. Probably will delete original posts too, but not thought about it much yet

Do you think this is a good solution to achieve the goal? What issues are present in doing it this way? Should I do it in a different way? What about deleting the original posts after being published on my site?

All voices are welcome!

Hi,

I want to self-host my own web server for nextcloud, jellyfin, gittea, and a bunch of other things to move away from big tech. I'm planning on having a VM for each of those apps, and running each of them in docker. I could then use Apache or Nginx to access it from outside my network. I've looked into virtual machines and found that QEMU would be the best option, especially for using the CLI. How would your recommend setting it up?

I ask this because I don't want my server being used in some kind of botnet or some shit like that. I don't think that will happen, but I'd prefer to just employ good practices to begin with just in case. Is it even worthwhile having a virtual machine for each of those services anyway?

Keep in mind that my PC I'm using is scrapped from spare parts with an R5 3600 and 16GB of memory. If I need to upgrade it I'm happy to get a bit more, but it shouldn't be an issue.

This is also my first post on programming.dev. I'm not sure if it is a good place to post this on but hopefully there are some people

Thanks!___

I know cloudlfare and other cloud providers often provide geoblocking features if you host with them.

Does anybody know of any similar foss packages for us self-hosters?

I know I don't have any need for somebody outside of my country to access my server, so geoblocking IPs would be a great security feature.

Is it possible/practical to host multiple websites from the same VPS server?

I have a number of domains (for example): bilbobaggins.com, rx4free.com, mypersonalblog.com

I don’t get much traffic and I’m not interested in paying for separate servers for each domain. But I still want https certificates to work properly.

Now I’m familiar with setting up wildcard subdomains with traefik and letsencrypt. But can I do the same or similar with completely different domains such as listed above?

I have a pi zero laying around and I have no use for it as of now. I was wondering if I could host a matrix server on it, I worry it might just not be powerful enough as the title would imply.

Hi, I realise that this might not be a question for this community; that said, this community is fairly big so I'm sure plenty of people here are already doing this.

I have been interested in hacking wireless infrastructure for a while now, but I'm struggling to find motivation in my day-to-day life to actually embark on said journey. Frankly speaking, I don't see a point to do so in a modern homelab. If someone is using WPA3, no unsecured wireless connections like Bluetooth, and uses strong passwords, how would someone realistically hack them without a good amount of time/resources?

One avenue that I came up with, related to wireless hacking, is with IOT. I do not know much about the security of various wireless protocols like Zigbee, or if one can somehow decrypt MQTT messages (they are sent using TLS, yes?) or anything of the sort. Other than this, I'm really struggling to see a practical point in pursuing action in this field (other than the basics like upgrading to the most secure protocol and maintaining digital hygiene) unless one is interested in wireless hacking from pure interest (without any need for motivation stemming from problems in their lab).

Thanks!

Just got my own instance up and running finally, but it doesn't seem to be federating as expected.

I can search for and subscribe to communities, but unless I specifically search for the URL (e.g. selfhost@lemmy.ml) nothing shows up. I'm also making this post from my instance as well (to see if anything happens...)

Is this something I just need to wait a few hours to happen or is something incorrect.

Thanks in advance

Edit: Oh, and when I do subscribe to a community, I can only see new posts. It doesn't retrieve older posts

I reaserached for hours but it looks like since they added typesense the 32-bit armv7 is no longer supporterd with docker.

I have a raspberry pi 4b with Raspbian installed which sadly is 32 bit, even the docked docs say they require a 64 bit, but i heard people usually habe no trouble setting up docker packages in a 32 bit.

Docker is the only way to install Immich right now, i tried to install an older version without typesense but docker can't find the package for those ("unknown") pheraps because it's a pretty old version.

I wonder if someone had the same problem and found a solution outside of that crazy dude in the repo discussion that managed to get it running somehow compiling it on its own.

I am currently using Immich to host my images of my family. Maybe this community can give me some insights of the features and the future of both products, and how you use them. It seems like the comparision chart in LibrePhotos is outdated. Thanks in advance :)

I have a domain and I use a hosted email I use a catchall email address to create an address on the fly for webshops and such. If they get conprimised or spam me I can block that email.

Is there awauy to have something like this for the new Identifier on the web: the phone number

Consider a wireguard network of many clients which all interact with each other through a central hub server on a cloud VPS. One of the clients is a desktop used for SSHing into the other various clients--again, through the central hub. If the "terminal" client connects to another client through the wireguard hub using SSH public/private key authentication, what if any information within that SSH tunnel gets exposed or leaked to the "hub" server?

My threat model is if the VPS was to ever get compromised. I previously SSH'd into the hub VPS server and from there I would SSH into any of the other clients with a password. Horrible security, I know.

My new setup is as mentioned above. Only the single desktop client has key authentication to SSH into the various clients. But I want to be sure none of that data gets exposed to the VPS hub just in case.

Hi! I'm still learning a lot. I have a raspberry pie (on raspian) with pi-hole, pivpn and outbound as a dns resolver. The next step will be for me to set up nextcloud. But I'm reading a lot on docker and containers these days and want to deploy docker now. Should I start again from the begining setting up docker and after pi-hole, pivpn and everything or could I keep everything the way it is now and then installing docker? The latter would be the best for me, but I don't know what to look for when installing docker after pivpn, pie hole etc. Thanks you a lot for the help

EDIT: Thanks everyone for your response and help!! You gave me the confidence to just try it and learn by doing it

To those self-hosters who are out there curious if there exists a faster Nextcloud-like alternative without all the bells and whistles. I present to you, OCIS! This howto assumes that you're running a Linux OS and that you have a reverse proxy like caddy running. To the guy who I suggested checking out OCIS, I wrote this guide for you ;)

1. Download the OCIS binary:

sudo wget -O /usr/bin/ocis https://download.owncloud.com/ocis/ocis/stable/3.0.0/ocis-3.0.0-linux-amd64

The above downloads the latest stable version.

2. Make the downloaded binary executable:

sudo chmod +x /usr/bin/ocis

3. Create a systemd unit file to start OCIS automatically:

Description=OCIS server

[Service]
Type=simple
User=ocis
Group=ocis
EnvironmentFile=/etc/ocis/ocis.env
ExecStart=ocis server
Restart=always

[Install]
WantedBy=multi-user.target

4. Create the ocis user:

sudo useradd --system --no-create-home --shell=/sbin/nologin ocis

5. Create the necessary directories:

sudo mkdir -p /etc/ocis /var/lib/ocis

6. Grant ownership of created directories above to the ocis user:

sudo chown -R ocis:ocis /etc/ocis/ /var/lib/ocis

7. Create the environment file for running OCIS:

sudo nano /etc/ocis/ocis.env and paste the lines below - edit the domain accordingly:

OCIS_URL=https://ocis.your.domain.org
PROXY_TLS=false
OCIS_INSECURE=true
OCIS_LOG_LEVEL=error
OCIS_CONFIG_DIR=/etc/ocis
OCIS_BASE_DATA_PATH=/var/lib/ocis
PROXY_HTTP_ADDR=0.0.0.0:9200
PROXY_USER_OIDC_CLAIM=preferred_username
PROXY_USER_CS3_CLAIM=username
PROXY_ENABLE_BASIC_AUTH=true

8. Initialize your OCIS install:

sudo -u ocis ocis init --config-path /etc/ocis

That's it! Visiting the url https://ocis.your.domain.org should present you with OCIS login page. To access the webdav endpoint via something like Cx Explorer or Gnome files, use: https://ocis.your.domain.org/remote.php/webdav

Depending on your setup, you might have to add the ocis domain above to your hosts file. A line like 127.0.0.1 ocis.your.domain.org to the /etc/hosts file.

If you don't have a reverse proxy but would still like to check out OCIS, use the environment file below:

OCIS_URL=https://192.168.x.x:9200
PROXY_TLS=true
OCIS_INSECURE=true
OCIS_LOG_LEVEL=error
OCIS_CONFIG_DIR=/etc/ocis
OCIS_BASE_DATA_PATH=/var/lib/ocis
PROXY_HTTP_ADDR=0.0.0.0:9200
PROXY_USER_OIDC_CLAIM=preferred_username
PROXY_USER_CS3_CLAIM=username
PROXY_ENABLE_BASIC_AUTH=true

I tried mine with keli.local:9200 and it worked just fine- make sure you have avahi-daemon installed!

Edit: If you're using a reverse proxy, your proxy address should be localhost:9200. For caddy, this would mean an entry like:

ocis.your.domain.org {
    reverse_proxy localhost:9200
}

Hey everyone! I run a server with Podman instead of Docker (and on an arm64 processor), so when I wanted to host a Lemmy instance on it, I faced a bit of challenges. I ended up writing this guide so if anyone else is in the same pickle, this should make installing Lemmy closer to headache-free.
If you're reading this post that means the guide works!

A common Google "hack" to get actual results it to add "site:reddit.com" after a search term--Lemmy, however is spread across a ton of instances. I was wondering whether there's a selfhosted search engine that is able to index lemmy posts from a certain number of instances (set manually) and that can be used to search for certain terms--is such a program being worked on? Or is it best to simply use the lemmy search bar on my instance?

Hi, I have an old ThinkPad that's running Arch for some self-hosted things, but I want to repurpose it to be dedicated to NextCloud with file sync and some apps (mainly Photos). Setting it up from scratch with Arch packages looks like way too much work. Should I use the Docker image on the existing OS, or a different package type, or wipe and start with a new OS?

I'm looking for the easiest solution that still has good performance and isn't going to break when it updates.