I have a Kindle and I kind of gave up on e-readers because I want to own my books. I've returned to physical books, but damn, they take too much space and it's hard to carry them around.
I would love to use this reader, but where can I buy books? Every place I searched is not as good as Amazon. THis is the thing blocking me from buying it.
I have a mini PC acting as my homeserver, but that poor thing does not have enough memory. I also would like to get into TrueNAS with ECC memory because I do care a lot about the data that I keep there. What would be the best budget build to get 64GB of ECC memory today?
CPU wise I don't need a lot, most of the services will be idle. I know that old servers may consume more memory, and I'm more or less ok with that. I'm not trying to compete with newer CPUs, but I would not want to draw 100+ at idle for example.
I'm looking for a new computer. I don't game, but I do use some heavy applications and I need tons of memory. Is this a good spec? I'll for sure not game.
I think the memory alone would be close to 700 euros, right?
>
Is 800 a good price for this computer in Europe? I think the memory alone could be something like 500, right? So it's basically 300 for everything else. Maybe this computer is worth 1.1k?
>Intel Core i7 12700 (Boxed with the original cooler)
Deepcool AK620
64GB DDR4 Kingston Fury Renegade 3600mhz
MSI MORTAR B660
1TB Western Digital Black NVMe SN850
Intel I350 T4
Intel WiFi
Corsair RM850x
NZXT H510
Im still trying to buy a homelab server. I saw a used PC with these settings:
>Intel Core i7 12700 (Boxed with the original cooler)
Deepcool AK620
64GB DDR4 Kingston Fury Renegade 3600mhz
MSI MORTAR B660
1TB Western Digital Black NVMe SN850
Intel I350 T4
Intel WiFi
Corsair RM850x
NZXT H510
Its been sold for 800 euros. Is this a good price? I think the memory alone is something like 600 euros now.
I was seeing a video from Wolfgang where he was explaining his homelab and how NixOS is used. I´m very inclined towards using it on my desktop computer, but I never really thought on using it at the server.
My homelab runs on a single node with Talos Linux, which is a Kubernetes distribution. Everything started quite simple, but now I have 10k lines of YAML. I use Kubernetes on my day job, but to some extend, it feels too much at my homelab. I dont want HA, scaling, and nothing like that. But I do appreciate how easy it is to drop things like gVisor and run containers with more safety, Cillium that allows me to control all the network, etc...
So now I m on this situation that I would like to have something simpler, and maybe NixOS is this solution for me since I already want to use it at my desktop, or stick with Kubernetes complexity.
If I get into NixOS my plan would be to run every service on its own user, although I think gVisor is still way more secure since it exposes just a few syscalls.
The big question for me is how to control the network how I do at Cillium where I allow only access to the internet, or a given service, or a given protocol, domain, etc... Maybe this will be a stretch.
I'm configuring the backups on my NAS to protect the photos I'm storing with Immich. The issue is, I would like to keep things from mobile phones out of the backup until I promote them to the permanent storage. This is to filter out all the crap that we receive from things like whatsapp, and at the same time to sync the good things I want and then selectively choose what I want to keep without paying the price of always having it on my backups.
I have a question about ECC. I know, I know, ZFS does not mandate ECC nor it requires ECC more than any other filesystem. But assuming that I'm on a system that does not have ECC and I was able to receive a file and properly store it at the drive. From this moment on there is any risk on data corruption?
My main question is that I have a system right now that I could install TrueNAS, but it does not have ECC, and I'm wondering what would be the impact of not having it.
At my homelab I have a Synology, a NanoPi, and a server running Kubernetes. I was looking on merging them all into a single appliance to make maintenance easier and save some space since I live in an apartment. My server has just 16GB of memory and it's struggling with my workloads, so I need more memory. ECC is a must for me because I'm saving my family photos and having a big server woulda allow to run the NAS part on ECC, that I already have with Synology, but also the server side giving me full ECC coverage.
I don't know much about the speed of this CPU, but I think it will be faster than my Synology and the server, N100, combined with a huge margin.
I think the only real worry I have is power consumption. How much would this draw? Idle and under load? I'm not sure how much I'm using right now, probably something like 50W.
Price wise it's ok, right? I can probably get the more or less the same by selling my current equip.
Dell 5810
- Intel E5 2696 v4 ( 22c/44t )
- 192Gb DDR4 ECC 6x32Gb
- Intel DC3600 1.6Tb NVME
- 2Tb Sata SSD
- 8Tb HDD
- Nvidia 1060 c/ 3Gb RAM
There are languages similar to Go, but there is something so unique about it that no one can replicate. It's even hard for me to understand as well. I've tried Elixir, Gleam, Rust, Zig, Odin, Java, Scala, and yet, every time I need to build something that I know needs to be good, fast, and well done Go is the language that I pick.
It could be familiarity, but it's really hard to beat the whole Go environment. I think most of the time engineers evaluate things into a single dimension axis either in terms of performance or type system expressiveness, but good code involves so may aspects and Go does surprisingly well in all of them.
I'm porting a few TB of photos and videos to my Immich instance. The issue that I'm having is that a few assets are failing to be processed and I can see them at the logs. But there is nothing at the UI and I can't filter at the UI to identify the asset.
There is anything that I can do? Also, any other suggestion when doing this port of TBs of data to an Immich instance? I just don't want to lose data.
Hey, me again making more Gleam questions. It's just that I'm very curious about the language and want to know more about it.
If I'm not mistaken right now Gleam is transpiling to Erlang. In case this is true, does the Gleam team has any plans to compile directly to the BEAM?
I'm just wondering if gleam_otp will eventually support all features from OTP. I was reading the project readme and saw this https://github.com/gleam-lang/otp#limitations-and-known-issues
I'm not sure what the framework is missing or not, nor I know OTP well enough, but it's one of the things that brought me to Erlang/BEAM.
I have way too many cloud providers and I want to centralize everything at my homelab. I'm almost there, I already have 2TB at my server, missing just another TB then I'll review to make sure everything is there and there is no corruption.
What are the tool you used to do this migration? I'm using immich-go. What is worrying me is this output:
2026-06-05 14:17:04 INF Asset Tracking Report:
2026-06-05 14:17:04 INF =====================
2026-06-05 14:17:04 INF Total Assets: 144233 (1.5 TB)
2026-06-05 14:17:04 INF Processed: 125421 (1.3 TB)
2026-06-05 14:17:04 INF Discarded: 18703 (221.1 GB)
2026-06-05 14:17:04 INF Errors: 8 (20.4 MB)
2026-06-05 14:17:04 INF Pending: 101 (378.2 MB)
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF ⚠️ WARNING: 101 assets did not reach a final state!
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF Event Report:
2026-06-05 14:17:04 INF =============
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF Discovery (Assets):
2026-06-05 14:17:04 INF discovered image : 124341 (905.0 GB)
2026-06-05 14:17:04 INF discovered video : 19892 (646.4 GB)
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF Discovery (Non-Assets):
2026-06-05 14:17:04 INF discovered sidecar : 130003 (88.8 MB)
2026-06-05 14:17:04 INF discovered unknown file : 3 (-)
2026-06-05 14:17:04 INF discovered unsupported file : 5 (2.1 KB)
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF Asset Lifecycle (PROCESSED):
2026-06-05 14:17:04 INF uploaded successfully : 143 (1.4 GB)
2026-06-05 14:17:04 INF server asset upgraded : 5 (12.8 MB)
2026-06-05 14:17:04 INF metadata updated : 125534 (1.3 TB)
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF Asset Lifecycle (DISCARDED):
2026-06-05 14:17:04 INF server has duplicate : 125271 (1.3 TB)
2026-06-05 14:17:04 INF discarded local duplicate : 18587 (220.8 GB)
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF Asset Lifecycle (ERROR):
2026-06-05 14:17:04 INF server error : 8
2026-06-05 14:17:04 INF
2026-06-05 14:17:04 INF Processing Events:
2026-06-05 14:17:04 INF associated metadata : 144132
2026-06-05 14:17:04 INF missing metadata : 101
2026-06-05 14:17:04 INF stacked : 2961
2026-06-05 14:17:04 INF added to album : 606
2026-06-05 14:17:04 INF tagged : 363
2026-06-05 14:17:04 INF
2026-06-05 14:56:07 ERR Some errors have occurred. Look at the log file for details
The second question that I have is what to do with failed jobs? At the `Job Queues` page I see a handful of warnings about failed jobs. What to do with those? There is very little information at the UI.
So many files discarded or with errors. How have you deal with this situation? A part from that, I think everything is ok. I'll configure the backup once I have everything verified.
Been writing Rust for a while, recently started messing with Odin. I get the basic pitch (C alternative, no borrow checker, manual memory, the context system), but I'm trying to understand what the language is actually like to live with day to day.
A few things from people who've used both:
context, how much of that safety do you actually miss? Do the allocator patterns cover most of it, or do you just hit more runtime bugs?Result and ?, how does Odin's multiple returns hold up in a big codebase? Stays disciplined, or do errors start getting ignored?Like both languages, just trying to build an accurate mental model. Real "I hit this wall" stories welcome.
I have a Immich instance running on a low power N100 CPU and I'm just importing terabytes of data into it. The poor machine is screaming and it will likely take weeks to finish the process. Can I run something on my desktop to speed up the process? Like maybe run the same docker containers, pointing to the same places?
At the minipc Immich is running inside Kubernetes.
Ah and one last question, I'm seeing failed jobs, what am I suppose to do what that? There are no retry buttons.
There is a ton of hate in the internet towards Go, and yet, Go is heavily used to build API servers. I wonder if someday Odin could be good enough where using it instead of Go is actually a viable choice.
Odin is a general purpose language, so yeah, it can be used, so can C, and we don't build lots and lots of APIs using C. Not trying to dunk on Odin or anything like that.
I just added a server to my homelab using Talos Linux. It is amazingly easy to configure it and put pods there. So far I'm doing the same I usually do on my day job, and everything is working alright. The question that I have now is the following, how to make it secure?
What are the steps that I need to take to execute containers, that I don't trust, on my cluster? I have Cilium installed and I already have network policies to reach the containers I need, on this part I think I'm good. But what about execution?
I know that one big step is to remove root execution, but then some images/helm are based on root. For that I was thinking on running something like gVisor or Kata.
I have Immich running on my homelab and I have the mobile app on my phone, and on my wife's. This solved a huge problem for us because now we can share our photos, so she can see the photos I'm taking from our sons when I'm alone with them, and the other way around. So no more "you did not sent me this photo".
The issue is, our sons have this tendency to pick our phones from time to time to see the photos. He loves it, and I think that's nice because he can come back in time and revive some memories like "daddy, who was on the beach with us this day?", but I fear that they could be deleting some photos by accident. I know, I can recover from the trash, but still, not the best.
There is any permission that I could set for a user to be kind of upload and read only? Or maybe a "destructive" mode that a user could get in once adding again it's login.
This is the only thing I'm missing from Immich to say it's the perfect photo management for me.