Pi-Hole vs AdGuard for casual users
I know it's a recurring question, but everytime you see opinions from users that have particular needs or setups.
My question is: from a "casual user" perspective, that doesn't need any strange setup but just and Ad blocker setup and just want a little home security, what's the best between those two?
Also, for user that don't want to waste time to setup the whole system, just put Linux on a PC and install one of those, a couple of configuration and done.
I have mine on an old Asus VivoStick that I almost never used and was sitting there alone, put Lubuntu on it, connected via USB to LAN and done.
For example I just saw that AdGuard has built in encryped DNS that Pi-Hole hasn't and a "web interface" that I don't think I ever use it but it's a nice perk.
https://redd.it/1ryox87
@r_SelfHosted
**Batch DELETE 1000**: MinIO 165ms, SeaweedFS 189ms, Garage 385ms — Garage is 2.3× slower
- **SeaweedFS S3 bug**: `mc mirror` fails with ContentLength mismatch on some files. boto3 works fine.
- **Garage deduplicates**: Content-addressed blocks mean identical files are stored once. Didn't realize this until I noticed 1000 "different" files stored as 1 block (my benchmark initially used the same random payload per size class 🤦)
## My Verdict
| If you care about... | Pick |
|---|---|
| **Upload speed** (small files, ingestion pipelines) | **SeaweedFS** |
| **Memory efficiency** (low RAM VPS, edge, RPi) | **Garage** |
| **Simplicity + S3 compatibility** | **MinIO** |
| **Disk efficiency** | **SeaweedFS** (1.02×) or **MinIO** (1.00×) |
| **Lowest per-request latency** (HEAD, DELETE) | **Garage** |
| **Large files** (10MB+) | **MinIO** |
For my workload (326K small files, write-heavy burst ingestion, write-once filings), **SeaweedFS has the strongest case** — 3.8× faster sequential uploads, best LIST performance, best disk efficiency. The tradeoff is operational complexity (3 containers vs 1) and an S3 compatibility edge case.
**Garage is the dark horse** — incredibly memory-efficient (~4 MB heap!), fastest HEAD/DELETE, competitive uploads. But LMDB metadata scaling (~2 KB/object = 675 MB at my scale) and slow batch deletes are real concerns.
**MinIO remains the safe default** — simplest to run, best S3 compatibility, lowest CPU usage, but noticeably slower for small-file ingestion.
---
*Benchmark tool: Python/boto3 with `time.perf_counter()`, 20 iterations per test, all on localhost.
https://redd.it/1ryfzw4
@r_SelfHosted
MinIO vs SeaweedFS vs Garage
I'm building a new server to ramp up my workload ~10×, so I re-evaluate my storage backend. I've been running MinIO, but since it's been archived, I wanted to explore
alternatives. SeaweedFS and Garage both looked promising, so I ran all three side-by-side on the same machine and benchmarked them.
I'm debating between SeaweedFS and Garage. SeaweedFS has better storage efficiency and raw throughput, but Garage's tiny memory footprint is tempting — my 64 GB server is shared with databases, app servers, and other services. Note: I only need a single server, no clustering. I'd love feedback from the community — especially anyone running SeaweedFS or Garage in production.
Full disclosure: The benchmark scripts and this write-up below this point were generated with AI. I caught and corrected
several issues along the way (dedup skewing Garage's storage numbers, docker stats under-reporting Garage's memory due
to mmap, AWS CLI adding 500ms overhead masking real latency). I think the numbers are reasonable now.
My workload: A self-hosted project storing Japanese stock filing documents (EDINET/TDnet) — 326K files, 20 GB
total. Most files are tiny: 84% of one bucket is under 10KB (daily JSON/CSV stock snapshots), another has 91K filing
images (PNG/PDF, 100KB–1MB). Write-once, read-occasionally. Burst ingestion pattern. Single server only — no clustering
needed.
## The Setup
- MinIO: Fresh empty instance, single container, default config
- SeaweedFS: Master + Volume + Filer (3 containers), -volumePreallocate=false
- Garage v2.2: Single container, LMDB backend, block_size=1MB, compression off
- All on localhost, same Docker host, 32 GB RAM
## Upload Latency (p50, 20 iterations each)
| File Size | MinIO | SeaweedFS | Garage |
|-----------|-------|-----------|--------|
| 1 KB | 7.4 ms | 2.3 ms | 1.8 ms |
| 5 KB | 9.3 ms | 2.3 ms | 3.0 ms |
| 100 KB | 9.5 ms | 3.2 ms | 3.6 ms |
| 1 MB | 19.7 ms | 10.7 ms | 8.4 ms |
| 10 MB | 48 ms | 65 ms | 63 ms |
SeaweedFS and Garage are 3–4× faster than MinIO for small file uploads. MinIO only wins at 10MB+.
## Sequential Batch Upload (1 worker, 1000 × 5KB files)
| Backend | ops/s |
|---------|-------|
| SeaweedFS | 431 |
| Garage | 376 |
| MinIO | 113 |
MinIO is 3.8× slower at sequential small-file ingestion. At 50 concurrent workers they all converge to ~570 ops/s.
## Downloads — Basically a Tie
All three deliver sub-2ms p50 for files under 1MB. Garage has a slight edge at 1–5KB (~1.1ms), SeaweedFS wins at 10MB (12.9ms vs 35ms Garage). No dramatic differences — downloads aren't the differentiator.
## Resource Consumption (the surprise)
Measured with a fresh MinIO instance (no production data) for fair comparison.
| | MinIO | SeaweedFS (3 containers) | Garage |
|---|---|---|---|
| Idle memory | 269 MB | 247 MB | ~90 MB (cgroup) |
| Memory under write load | 441 MB | 421 MB | ~13 MB (docker stats) |
| CPU under write load | 53% | 101% | 134% |
| Containers | 1 | 3 | 1 |
\Garage uses LMDB (mmap), so docker stats under-reports. Cgroup memory is ~90 MB idle. The OS manages which LMDB pages stay in RAM via page cache — cold pages get evicted automatically under memory pressure. Actual non-evictable heap is just ~4 MB.
Tradeoff: SeaweedFS/Garage burn more CPU for faster uploads. MinIO is the most CPU-efficient.
## Storage Efficiency (tested with unique random data per file)
| Backend | 24.65 MB logical → disk | Ratio |
|---------|------------------------|-------|
| SeaweedFS | 25.29 MB | 1.02× |
| MinIO | ~24.7 MB | 1.00× |
| Garage | 31.79 MB | 1.29× |
Garage has ~6 MB LMDB metadata overhead per 1K objects (~2 KB/object). At my 326K objects that's ~675 MB just for metadata on disk. SeaweedFS and MinIO are essentially 1:1.
## Other Notable Findings
- LIST 1000 objects: SeaweedFS 68ms, Garage 72ms, MinIO 86ms
- HEAD latency: Garage wins at ~1.1ms across all sizes (vs ~1.25ms for both others)
-
Foldergram: Self-hosted local photo gallery with an Instagram-style feed and layout
https://redd.it/1ryjc9e
@r_SelfHosted
What's your 'I can't believe I self-hosted that' service?
Curious what services surprised you by being worth self-hosting. Not the obvious stuff like Plex or Pi-hole, but things you didn't expect to work well or didn't think were worth the effort until you tried. What's running on your setup that you'd never go back to a hosted version of?
https://redd.it/1rykzo4
@r_SelfHosted
PSA: Think hard before you deploy BookLore
Wanted to flag some stuff about BookLore that I think people need to hear before they commit to it.
**The code quality issue**
There's been speculation for a while that BookLore is mostly AI-generated. The dev denied it. Then v2.0 landed and, well: crashes, data not saving, UI requiring Ctrl+F5 to show changes, the works. These are the kinds of bugs you get when nobody actually understands the codebase they're shipping.
The dev is merging 20k-line PRs almost daily, each one bolting on some new feature while bugs from the last one go unfixed. And the code itself is a giveaway: it uses Spring JPA and Hibernate but is full of raw SQL everywhere. Anyone who actually built this by hand would keep the data layer generic. Instead, something like adding Postgres support is now a huge lift because of all the hardcoded shortcuts. That's not a style preference, that's what AI-generated code looks like when nobody's steering.
**How contributors get treated**
This part is what really bothers me.
People submit real PRs. They sit for weeks, sometimes months. Then the dev uses AI to reimplement the same feature and merges his own version instead. Predictably, this pisses people off. At the time of writing this, the main dev has alienated almost all of the contributors that were regularly supporting, triaging issues and doing good work on features and bugfixes.
When called out, he apologizes. Except the apologies are also AI-generated. And more than once he forgot to strip the prompt, so contributors got messages starting with something like "Here's how you could apologize—"
One example I'm familiar with, because I was following for this feature for a while (over 2 months?): someone spent serious time building KOReader integration. There was an open PR, 500+ messages of community discussion around it. The dev ignored it across multiple releases, then deleted the entire thread and kicked the contributor from the Discord. What shipped in that release instead? "I overhauled OIDC today!" Cool.
Every time criticism picks up in the Discord, the channel gets wiped and new rules appear. This has happened multiple times now.
**The licensing bait-and-switch**
This is the part that should actually scare you if you're thinking about deploying this.
BookLore is AGPL right now. The dev is planning to switch to BSL (Business Source License), which is explicitly *not* an open source license. He also plans to strip out code from contributors he's had falling-outs with. Everyone who contributed did so under AGPL terms. Changing that out from under them is a betrayal, full stop.
The main dev had a full on crashout on another discord, accusing people of betrayal etc because they were....forking his code? I am not going to paste the screenshots of the crashout because it is honestly just unhinged and reflects badly on him, maybe its something he'll regret and walk back on - hopefully.
It gets worse. There's a paid iOS app coming with a subscription model. What does that mean concretely? You'll be paying a subscription to download your own books offline to your phone. Books you host yourself. On your own hardware.
The OIDC implementation, which should be a standard security feature, is being locked down specifically to block third-party apps from connecting, so the only mobile option is the paid one. Features the community helped build are being turned into a paywall funnel.
The dev has said publicly that he considers forking to be "stealing" and wants to prevent it. He's also called community contributions "AI slop." From the guy merging AI-written 20k-line PRs daily. Make of that what you will.
**Bottom line**
* Contributors get ignored, reimplemented over, and kicked out
* AGPL → BSL relicense is coming, with contributor code being stripped
* Paid iOS app will charge you a subscription to access your own self-hosted books offline
* OIDC is being locked down to kill third-party app access
* The dev thinks forking is theft and has open contempt for OSS
Self hosted options for Youtube videos & clips
So, obviously there is Plex for videos and I self host an instance of that for our Movies & TV shows.
But recently was wondering if there could be an option for some sort of filtered YouTube.
I came across one project on Github that um... advertises itself for Gooners and organizing your goon collection, lol. It looked functional for SFW content, but maybe not ideal.
Just wondering if anyone has any suggestions or if I should just consider sticking with videos inside Plex.
https://redd.it/1rrwv91
@r_SelfHosted
Offline VIN decoder - no API keys, works locally
If you're building anything automotive-related and want VIN decoding without external API dependencies:
**@cardog/corgi** \- offline VIN decoder
* 23MB SQLite database bundled
* No network requests needed
* Works in Node.js, browser, CLI
# CLI
\`npx u/cardog/corgi decode 1HGCM82633A123456\`
# Node.js
import { createDecoder } from '@cardog/corgi' const decoder = await createDecoder() const result = await decoder.decode('1HGCM82633A123456')
Just shipped v2.0 with community-contributed patterns for international vehicles (Tesla China/Berlin).
GitHub: [https://github.com/cardog-ai/corgi](https://github.com/cardog-ai/corgi) npm: [cardog/corgi" rel="nofollow">https://www.npmjs.com/package/@cardog/corgi](cardog/corgi" rel="nofollow">https://www.npmjs.com/package/@cardog/corgi)
https://redd.it/1rruhl8
@r_SelfHosted
Sigh more Vibe Coded Apps....
Beware folks, why would you give your entire network access to vibe coded stuff. https://github.com/BlkLeg/CircuitBreaker. These apps are popping up everywhere and potentially dangerous.
From the creator:
"Between classes (cybersec) and my job, I'm working on my pet project below. The backend is Python, frontend is JS. Primarily deployed via Docker for cross-compatibility, but I plan on binaries for Linux and Windows to maintain low overhead. Its currently at 140MiB RAM with 0 optimization passes run (pending)."
and
"Is this vibe coded ?"
"Yes. I'll always be upfront about this, as I'm not a programmer by trade or hobby. Since my particular dept of cyber security will likely involving reviewing code for vulnerabilities regularly, that's where my brain power goes. That's why I chose Python, as I've had the most exposure to it so far. Security, performance, and function are priorities. Before full release, it will be refactored for maintainability and stability so I can take on volunteer contributors. Circuit Breaker will be the people's app through and through."
https://redd.it/1rrrj69
@r_SelfHosted
Any project management software with no limitations for self-hosted version?
I'm trying to find a good project management tool to replace ClickUp or Asana.
What I need is to create tasks and visualize them in different ways, like table view, calendar view, and boards. It also important to have custom fields.
The main issue I'm facing is that most tools seem to limit features in the self-hosted version.
For example, Plane looks really promising, but from what I understand, the free self-hosted version doesn’t support custom fields. I ran into a similar limitation with Leantime.
Does anyone know a project management tool that is fully functional when self-hosted, without feature restrictions compared to the cloud version?
Thanks!
https://redd.it/1rr7lws
@r_SelfHosted
The best security is having it offline, but….
….I’d like to access my stuff remotely. I’m working on locking my network down for improved security. Right now, this involves setting up Guacamole so it will be accessible through a Cloudflare tunnel. The idea is I can access my web applications, and Guacamole will act as a bastion for SSH/RDP/VNC. I’ve got three cloudflared servers - one VM on each Proxmox cluster member and another on an oDroid.
The tricky part is, if this is all configured the way I envision, if Guacamole goes offline for whatever reason, I won’t be able to access any of my workstations or infrastructure. Only the web apps will be available. I realize I don’t have redundant routers or switches so those are single points of failure as well, though the probability of those going offline is lower than the Guacamole VM having problems (especially if I’m doing maintenance remotely and it blows up).
For those of you who have a similar setup, what are you using (some combo of Guacamole, Pacemaker, something else?), and how do you have it set up? How did you work around any single points of failure that could interfere with remote access?
https://redd.it/1rrjd4i
@r_SelfHosted
Should I self host Bitwarden (with Vaultwarden) or am I just paranoid?
Hi!
So, I totally get that sometimes, it makes sense to pay other people to host crucial services. I saw some dude call it the beer test. If a service is important enough that if it went down and you were on vacation enjoying a beer, you'd put your beer down and fix it, you should not self host it.
That makes sense to me and that's why I paid Bitwarden their very fair subscription.
However, with everything that is going wrong in the world right now, I really don't want to put something as important as a password manager into somebody else's hand. If my email provider goes away, I can move my domain somewhere else. That's not that easy with Bitwarden, I feel.
There are two potential issues I see:
1. Enshittification is going to hit Bitwarden as well or they sell the company or whatever. I feel like in the last years almost every single product I used to use turned to garbage.
2. I'm not American and if somebody in the US government realizes that the easiest way to make Europe jump is to just cut that deep sea cable I'm gonna be in real trouble.
I don't consider Bitwarden to be part of the same garbage that Big-Tech is. So I'm not really trying to replace them in the same way I'd want to replace Google for moral or privacy reasons.
But I'm not sure if I'm paranoid or if that is something I should be concerned about. Even though I said not self hosting password managers make sense, emotionally it always feels wrong to have this public.
If I were to self host, I'd only make it accessible via a VPN, having everything in 3-2-1 backups. So I think I can pull it off safely but I'm not sure if I should.
https://redd.it/1rr8bzy
@r_SelfHosted
all responses.
# Configuration and Environment Variables
* **USE\_HEROUI\_THEME:** Toggle the HeroUI/Aceternity interface (Default: TRUE).
* **SHOW\_NOW\_WATCHING\_CARD:** Toggle the main page playback card (Default: TRUE).
* **SEERR\_URL / SEERR\_API\_KEY:** Unified variables for Seerr-compatible services.
* **CORS\_ALLOWED\_ORIGINS:** Define allowed origins for WebSocket connections (Default: \*).
**Full Changelog**: [https://github.com/sahara101/Movie-Roulette/compare/v5.1.2...v5.2.0](https://github.com/sahara101/Movie-Roulette/compare/v5.1.2...v5.2.0)
https://redd.it/1rpahrw
@r_SelfHosted
first attempt at making a diagram
https://preview.redd.it/rym3sf4e71og1.png?width=9159&format=png&auto=webp&s=0ea860e03a5a0b6fde12197edab9ca4ed74670ab
not the most detailed one yet, but I’ll keep iterating on it. Let me know your thoughts :)
also not the best choice of components, but I made do with what I had!
https://redd.it/1rp23oj
@r_SelfHosted
Self-hosting Vaultwarden
With 1Password increasing their prices I'm interested in self-hosting a password manager and Vaultwarden seems to be the choice of many. Hosting it so it is accessible via VPN tunnel only is a fairly safe way to go about it, but since I also like to use a commercial VPN (Mullvad) switching from one to another isn't the most fluid process.
My current plan is to have a Caddy reverse proxy that routes via Tailscale tunnel from my VPS to my home Raspberry Pi 5 that hosts Vaultwarden. My plan for Caddy is to configure it to only accept certain IP ranges as well as have caddy-security. The subdomain that is configured like this would be behind a wildcard subdomain (think pi.domain.tld would have wildcard to any domains under it and vault.pi.domain.tld would forward to my Pi's VW port). I'd also have CrowdSec to block any IPs that hammer my domains.
How secure would this set-up be? Any other things I could/should consider to keep my info secure, or should I accept that I can only access it via Tailscale? I want my partner to also use this as their password manager and they are quite reluctant to turn on Tailscale every time they need access to a password manager or use it constantly either.
https://redd.it/1rozuit
@r_SelfHosted
Do you self-host Matrix?
https://redd.it/1rysyw3
@r_SelfHosted
MinIO vs SeaweedFS vs Garage
I'm building a new server to ramp up my workload ~10×, so I re-evaluate my storage backend. I've been running MinIO, but since it's been [archived](https://github.com/minio/minio), I wanted to explore
alternatives. SeaweedFS and Garage both looked promising, so I ran all three side-by-side on the same machine and benchmarked them.
I'm debating between SeaweedFS and Garage. SeaweedFS has better storage efficiency and raw throughput, but Garage's tiny memory footprint is tempting — my 64 GB server is shared with databases, app servers, and other services. Note: I only need a single server, no clustering. I'd love feedback from the community — especially anyone running SeaweedFS or Garage in production.
**Full disclosure:** The benchmark scripts and this write-up below this point were generated with AI. I caught and corrected
several issues along the way (dedup skewing Garage's storage numbers, `docker stats` under-reporting Garage's memory due
to mmap, AWS CLI adding 500ms overhead masking real latency). I think the numbers are reasonable now.
**My workload:** A self-hosted project storing Japanese stock filing documents (EDINET/TDnet) — **326K files, 20 GB
total**. Most files are tiny: 84% of one bucket is under 10KB (daily JSON/CSV stock snapshots), another has 91K filing
images (PNG/PDF, 100KB–1MB). Write-once, read-occasionally. Burst ingestion pattern. Single server only — no clustering
needed.
## The Setup
- **MinIO**: Fresh empty instance, single container, default config
- **SeaweedFS**: Master + Volume + Filer (3 containers), `-volumePreallocate=false`
- **Garage v2.2**: Single container, LMDB backend, `block_size=1MB`, compression off
- All on localhost, same Docker host, 32 GB RAM
## Upload Latency (p50, 20 iterations each)
| File Size | MinIO | SeaweedFS | Garage |
|-----------|-------|-----------|--------|
| 1 KB | 7.4 ms | 2.3 ms | **1.8 ms** |
| 5 KB | 9.3 ms | **2.3 ms** | 3.0 ms |
| 100 KB | 9.5 ms | **3.2 ms** | 3.6 ms |
| 1 MB | 19.7 ms | 10.7 ms | **8.4 ms** |
| 10 MB | **48 ms** | 65 ms | 63 ms |
**SeaweedFS and Garage are 3–4× faster than MinIO for small file uploads.** MinIO only wins at 10MB+.
## Sequential Batch Upload (1 worker, 1000 × 5KB files)
| Backend | ops/s |
|---------|-------|
| SeaweedFS | **431** |
| Garage | 376 |
| MinIO | 113 |
MinIO is **3.8× slower** at sequential small-file ingestion. At 50 concurrent workers they all converge to ~570 ops/s.
## Downloads — Basically a Tie
All three deliver sub-2ms p50 for files under 1MB. Garage has a slight edge at 1–5KB (~1.1ms), SeaweedFS wins at 10MB (12.9ms vs 35ms Garage). No dramatic differences — downloads aren't the differentiator.
## Resource Consumption (the surprise)
Measured with a fresh MinIO instance (no production data) for fair comparison.
| | MinIO | SeaweedFS (3 containers) | Garage |
|---|---|---|---|
| **Idle memory** | 269 MB | 247 MB | **~90 MB** (cgroup) |
| **Memory under write load** | 441 MB | 421 MB | **~13 MB** (docker stats*) |
| **CPU under write load** | **53%** | 101% | 134% |
| **Containers** | 1 | 3 | 1 |
\*Garage uses LMDB (mmap), so `docker stats` under-reports. Cgroup memory is ~90 MB idle. The OS manages which LMDB pages stay in RAM via page cache — cold pages get evicted automatically under memory pressure. Actual non-evictable heap is just **~4 MB**.
**Tradeoff: SeaweedFS/Garage burn more CPU for faster uploads. MinIO is the most CPU-efficient.**
## Storage Efficiency (tested with unique random data per file)
| Backend | 24.65 MB logical → disk | Ratio |
|---------|------------------------|-------|
| SeaweedFS | 25.29 MB | **1.02×** |
| MinIO | ~24.7 MB | 1.00× |
| Garage | 31.79 MB | 1.29× |
Garage has ~6 MB LMDB metadata overhead per 1K objects (~2 KB/object). At my 326K objects that's **~675 MB just for metadata** on disk. SeaweedFS and MinIO are essentially 1:1.
## Other Notable Findings
- **LIST 1000 objects**: SeaweedFS 68ms, Garage 72ms, MinIO 86ms
- **HEAD latency**: Garage wins at ~1.1ms across all sizes (vs ~1.25ms for both others)
-
What’s the most “boring reliable” self-hosted stack you’ve used that still scales well?
I’m not looking for the coolest stack or the most “cloud native” answer.
I’m looking for the setup you’d put on a box, sleep well at night with, and not hate 6 months later.
The project is not huge, but it’s also not just a toy. Think somewhere in the middle:
content/site layer,
some app logic,
user accounts,
background jobs,
APIs,
a database,
maybe Redis,
maybe object storage later.
I keep seeing 2 extremes:
one side says “just run everything in Docker Compose and keep life simple”
the other says “you need Traefik, queues, workers, metrics, S3, separated services, external DB, secrets management, and three dashboards before you even launch”
I’m trying to avoid both underbuilding and building myself a tiny DevOps trauma machine.
So I guess my real question is:
If you were self-hosting a small-but-real production app today, what stack would you choose if your main goal was long-term sanity?
Not “maximum scale”
Not “best on paper”
Just stable, maintainable, easy to restore, and not annoying.
Would love answers from people who’ve already gone through the phase of overengineering everything and came back to something simpler.
Thanks :)
https://redd.it/1ryr1en
@r_SelfHosted
Proud of my little self hosted server
https://redd.it/1ryflmp
@r_SelfHosted
norms
[https://postimg.cc/gallery/R3WJKVC](https://postimg.cc/gallery/R3WJKVC) \- some examples. I couldn’t grab some from the official discord, seeing as how ACX has a habit of wiping that one whenever some pushback is posted.
This is the huntarr situation all over again. Deploy with caution, or honestly, wait and see if a community fork shows up under a license that actually holds.
https://redd.it/1rs275q
@r_SelfHosted
Looking to self host a word processor for creative writing
My sister is a creative writer. She uses Google docs and frequently writes on her phone. She (understandably) is becoming increasingly uncomfortable with Google having her work on their hardware and wants to move it to my nextcloud server. The only issue is that every word processor I have come across has a really awful android app experience when compared to Google docs.
She wants just some really basic features doesn't need crazy formatting or anything like that, just the following:
-Dark mode
-An app navigation interface that is more similar to a notes app that doesn't load the whole damn page and make you zoom in just to see what your writing
-a few font options
-the ability to toggle spell check off
-a nice to have would be an autocorrect feature
I have yet to find anything that really suits this use case that I can self host. Any ideas?
https://redd.it/1rrxdnx
@r_SelfHosted
Discord to Unifi Access Chatbot
I wanted a way to allow my friends access to my game servers without exposing the game server to the world. I created a chatbot that allows users with a specific role on my discord server to request access. It will then send them a unique link and allow them to add their public IP to the access rules on my firewall.
I'll have to update my docs to include this, but this app assumes you are running a Unifi firewall and that you are using some sort of domain services for a persistent hostname. (Static IP or DDNS)
Looking to make it more modular for other firewalls and game servers.
https://github.com/Copter64/chatbot\_access\_project
https://redd.it/1rrv006
@r_SelfHosted
Force generate missing hard/sim link for movie file
I have a fairly standard Arr-stack server running on a Synology NAS in docker. So far it has worked flawlessly at getting requests from Jellyseerr, finding the file, loading it into qBit, hard linking the file to media folder and populating Jellyfin automatically. Love it.
For some reason, one file I downloaded out of dozens didn’t hard link automatically, and the only way I can think to maintain the seed while populating jellyfin (without moving the file from downloads dir to media dir) was to actually copy and paste the 20 gig file effectively doubling its size on my hard drive.
Which part of the stack is actually responsible for generating the hardlink on the NAS and is there a way to force it to attempt to do it again without redownloading the whole file/interrupting the seed/ruining ratio?
https://redd.it/1rrra4k
@r_SelfHosted
Bosch Flexidome 8000i - Alarm triggered SD card recording locked while managed by VRM
I want to modify the settings of my Bosch Flexidome 8000i camera so that when an event or alarm occurs, it writes the footage to an SD card 5 seconds before and after the event. However, when I look at the web interface, it directs me to the "Bosch Configuration Manager" application for VCA and the "Bosch Configuration Client" application for recording. In both, the recording tab appears locked, and I cannot interact with most of the recording tools.
Is there any way to enable alarm-triggered SD card recording (Recording 2) while the camera is still managed by VRM? Or is the only option ANR?
https://redd.it/1rrq4bz
@r_SelfHosted
Could someone help me? I'm desperate
https://redd.it/1rr6ivg
@r_SelfHosted
Best way to configure SMB? (TempleOS)
My main server is now running TempleOS, but I'm not sure how to setup SMB so my other TempleOS devices can connect. Any tips or tutorials?
Thanks!
https://redd.it/1rr4n51
@r_SelfHosted
I turned my old Galaxy S10 into a self-hosted server running Ubuntu 24.04 LTS with Jellyfin, Samba, and Tailscale - no Docker, no chroot, no proot - fully integrated at the system level with pure init, auto-running the entire container at device boot if needed!
https://redd.it/1rqr8yq
@r_SelfHosted
Movie Roulette v5.2.0 released!
I just released a new version of Movie Roulette!
Just to get it out of the way: Yes, I used AI. It is not a secret, it is clearly stated on the GH page as well. Not AI Friday because first release was in 2024.
Github: [https://github.com/sahara101/Movie-Roulette](https://github.com/sahara101/Movie-Roulette)
\# What is Movie Roulette?
At its core it is a tool which chooses a random unwatched movie from your Plex/Jellyfin/Emby movie libraries. However it can do more!
Please check on github for complete info.
Introduced a new theme and also refreshed the original theme. Here some comparison screenshots between new and refreshed:
[https://imgur.com/a/JuF2AcT](https://imgur.com/a/JuF2AcT)
Here you will find screenshots of also old version:
[https://github.com/sahara101/Movie-Roulette/tree/main/.github/screenshots](https://github.com/sahara101/Movie-Roulette/tree/main/.github/screenshots)
New in v5.2.0 (kinda big :) )
# Movie Roulette Release Notes
# Major Feature: HeroUI Theme
* **Full Integration:** Modern glassmorphism and effects applied to all pages, including Settings, Collections, and Login.
* **Default Active:** The theme is now enabled by default via the `USE_HEROUI_THEME` variable.
# New Features and UI Improvements
* **Now Watching Card:** Real-time playback status on the main page with progress tracking and PNG sharing.
* **Grid View Overhaul:** New card layout featuring hover-to-play overlays and a current-set shuffle mode.
* **Seerr Service Migration:** Merged Overseerr and Jellyseerr into a single unified "Seerr" request service.
* **Integrated Cache Management:** Moved service and user cache tools into the main Settings sidebar for admins.
* **In-App Media Details:** Collections movie titles now open internal overlays instead of external TMDb links.
* **Unified Navigation:** Combined desktop and mobile menus to ensure full page access on small screens.
* **Mobile Button Fix:** Restored Grid View and Collections buttons previously hidden in the legacy mobile theme.
* **iOS Tap-to-Top:** Status bar taps now smoothly scroll active modals and filmography back to the top.
* **Cast Display (Issue #58):** Limited display to 4 actors to prevent layout wrapping on posters and screensavers.
* **Markdown Release Notes:** The update notification popup now renders formatted markdown for better readability.
* **Other UI Enhancements:** Added service-specific SVG logos, improved user role badges, and added total movie counts to the collections search.
# Bug Fixes
* **Collections Playback:** Fixed failed playback caused by sending TMDb IDs instead of library IDs.
* **iOS Search Zoom:** Set 16px font minimums to prevent browser auto-zoom on search inputs.
* **Session Purging:** Resolved a bug where expired sessions were never deleted from the database file.
* **Grid Mismatches:** Fixed an issue where movie cards occasionally opened details for the wrong film.
* **Jellyfin Metadata:** Fixed "Unknown" video and audio formats in poster and screensaver modes.
* **Playback Tracking:** Resolved poster hijacking and start-time drift during stream resumes.
* **Trakt Sync:** Fixed token refresh failures and resolved incorrect unwatched warnings.
* **Asset Handling:** Replaced missing actor photos with SVG placeholders to stop 404 network errors.
* **Filter UI:** Implemented immediate count updates when switching between media services.
# Security and Technical Changes
* **Runtime Upgrade:** Upgraded to Python 3.12 and Debian Bookworm for the latest security patches.
* **API Hardening:** Enforced authentication requirements on 38 previously exposed endpoints.
* **Password Security:** Migrated to PBKDF2-HMAC-SHA256 hashing and enforced an 8-character minimum.
* **Brute-Force Lockout:** Accounts now lock for 15 minutes after 5 failed login attempts.
* **Credential Masking:** API keys and tokens are now stripped from settings responses.
* **Trakt PKCE:** Migrated OAuth flow to PKCE for more secure token exchanges.
* **Security Headers:** Added XSS, CORS, and Referrer-Policy protection to
Why does a simple, free, self hosted file storage platform not exist?
I've tried everything from Nextcloud, ownCloud, OpenCloud, and Pydio Cells. But I still can't seem to find exactly what I'm looking for, and I'm wondering why it doesn't already exist. File storage is (in my opinion) one of the most helpful use cases for a self-hosting setup, but I don't understand why there isn't a self hosted cloud storage platform that:
- is cross-platform
- has relatively low resource usage
- uses a flat file structure, not S3-style blobs
- handles thumbnailing for more file types than just images
- has virtual filesystems OR selective sync for common operating systems
- has decent sharing or multi-user tools
- has good upload and download speeds
Essentially, I don't understand why a fully self-hostable and user-friendly Google Drive alternative doesn't exist. I'm a developer and I understand that it would obviously be a large undertaking to build, but it's a type of software that's very common for self-hosters and I don't see why a better option doesn't exist than the established players. NextCloud is too heavy/is trying to do too much, ownCloud is too corporate and a pain to maintain (plus the interface is crap), Pydio is good but the client apps (aside from the web app) are horrendous, Seafile is limited to blobs and is slightly proprietary, FileRun is paid, etc. Just seems to me like a major gap in the space. Anyone have any insight on why something like this doesn't exist?
https://redd.it/1rp2vup
@r_SelfHosted
Self hosting authoritative DNS servers
https://devblog.yvn.no/posts/self-hosted-dns/
https://redd.it/1rox1qz
@r_SelfHosted
