https://solariz.de/ Recent content on solariz.de - Tech & Thoughts Hugo en-us Sun, 22 Feb 2026 00:00:00 +0000 https://solariz.de/posts/26/02-kara-keep-tiles-addon/ Sun, 22 Feb 2026 00:00:00 +0000 https://solariz.de/posts/26/02-kara-keep-tiles-addon/ What is Karakeep? Well, I’ll write an extra blog article about it soon, so in short: It’s a self-hosted open-source bookmark manager with a cloud-hosted option available. I work on multiple devices and browsers, and the integrated browser bookmark management is not ideal for my workflow. Karakeep (formerly named Hoarder) solves this problem and adds lots of useful features, such as an actual page summary, allowing me to use it not only as a bookmark manager, but also as a replacement for my “read it later” list. What the Kara-Tiles Addon does? Well may but a video on later but for now a short Screenshot says more than 1000 words. To Explain a bit; It access your Karakeep Bookmarks via API, cache them in your browser local storage and display a nice “old style Tiles” overview with added features like: Offline Fuzzy Sarch, Tag and domain Filters, hotkeys, favorite Bar and most used. Availability Currently in the Chrome Extension store and soon (just submitted it for review) in Firefox Addon Store. Chrome Extension Firefox coming soon Options and How to use? After Installation you need to enter your Karakeep instance URL as well as an API Key; This Key you can get in your Karakeep user settings > API. What next I work with the extension everyday in all of my browsers, so be sure there will be updates to improve usability further. Nothing huge featurewise in my mind right now, I love the Principle of an Programm or extension fullfilling one purpose good instead of doing everything but mediocre. If there is anything bothering feel free to leave me a comment below this Article. I will use this Blog Post as “Extension Webpage” Link and may update it from time to time. Kara-Tiles Changelog I try to keep a small changelog here: 0.0.6 22 Feb ‘26 - First Firefox Build, grayscale images, better tag search 0.0.5 21 Feb ‘26 - Tiny bug fixes and UI Improvements 0.0.4 20 Feb ‘26 - First iteration published in chrome extension store https://solariz.de/posts/26/02_usb-travel-dongle/ Wed, 18 Feb 2026 00:00:00 +0000 https://solariz.de/posts/26/02_usb-travel-dongle/ Not an lenghy Article, just posting my findings for whoever may need it. The Hardware: i-tec USB-C 4K Metall Nano Docking Station Named by the bulky name i-tec USB-C 4K Metall Nano Docking Station 1x HDMI 1x GLAN 3x USB 3.0 1x SD/MicroSD 1x Audio Power Delivery 100W I found mine on Amazon. Port-wise, it has what I need. Mostly I use this kind of light dock if I work remotely or at events to attach one screen, network, mouse and have it charging my laptop. So I can quickly grab it when I need to go somewhere. dmesg On my Fedora 43 quick dmesg log when plugging in the dock. I had plugged in 2 USB devices, an HDMI screen, network, and a 90W USB-C PD charger. With some previous adapters, I had the issue that they were very picky about which port I used on my laptop, but this one didn’t care much. [13572.247251] usb 3-1: new high-speed USB device number 18 using xhci_hcd [13572.337179] usb 2-1: new SuperSpeed USB device number 9 using xhci_hcd [13572.358474] usb 2-1: New USB device found, idVendor=05e3, idProduct=0626, bcdDevice= 6.63 [13572.358485] usb 2-1: New USB device strings: Mfr=1, Product=2, SerialNumber=0 [13572.358488] usb 2-1: Product: USB3.1 Hub [13572.358490] usb 2-1: Manufacturer: GenesysLogic [13572.361782] hub 2-1:1.0: USB hub found [13572.362684] hub 2-1:1.0: 4 ports detected [13572.375448] usb 3-1: New USB device found, idVendor=05e3, idProduct=0610, bcdDevice= 6.63 [13572.375453] usb 3-1: New USB device strings: Mfr=1, Product=2, SerialNumber=0 [13572.375454] usb 3-1: Product: USB2.1 Hub [13572.375455] usb 3-1: Manufacturer: GenesysLogic [13572.376617] hub 3-1:1.0: USB hub found [13572.376853] hub 3-1:1.0: 4 ports detected [13572.661967] usb 2-1: USB disconnect, device number 9 [13572.678131] usb 3-1.1: new high-speed USB device number 19 using xhci_hcd [13572.777167] usb 3-1.1: New USB device found, idVendor=1a40, idProduct=0101, bcdDevice= 1.11 [13572.777175] usb 3-1.1: New USB device strings: Mfr=0, Product=1, SerialNumber=0 [13572.777177] usb 3-1.1: Product: USB 2.0 Hub [13572.778587] hub 3-1.1:1.0: USB hub found [13572.778612] hub 3-1.1:1.0: 4 ports detected [13572.877336] usb 3-1.2: new full-speed USB device number 20 using xhci_hcd [13572.979025] usb 3-1.2: New USB device found, idVendor=1e7d, idProduct=319c, bcdDevice= 1.00 [13572.979038] usb 3-1.2: New USB device strings: Mfr=1, Product=2, SerialNumber=0 [13572.979041] usb 3-1.2: Product: ROCCAT Isku [13572.979043] usb 3-1.2: Manufacturer: ROCCAT [13573.884908] usb 2-1: new SuperSpeed USB device number 10 using xhci_hcd [13573.905813] usb 2-1: New USB device found, idVendor=05e3, idProduct=0626, bcdDevice= 6.63 [13573.905821] usb 2-1: New USB device strings: Mfr=1, Product=2, SerialNumber=0 [13573.905823] usb 2-1: Product: USB3.1 Hub [13573.905825] usb 2-1: Manufacturer: GenesysLogic [13573.908298] hub 2-1:1.0: USB hub found [13573.909336] hub 2-1:1.0: 4 ports detected [13574.192045] usb 2-1.1: new SuperSpeed USB device number 11 using xhci_hcd [13574.211934] usb 2-1.1: New USB device found, idVendor=0b95, idProduct=1790, bcdDevice= 2.00 [13574.211939] usb 2-1.1: New USB device strings: Mfr=1, Product=2, SerialNumber=3 [13574.211941] usb 2-1.1: Product: AX88179B [13574.211942] usb 2-1.1: Manufacturer: ASIX [13574.211943] usb 2-1.1: SerialNumber: 0034298FXXXXXX [13574.264676] cdc_ncm 2-1.1:2.0: MAC-Address: 98:fc:84:XX:XX:XX [13574.264687] cdc_ncm 2-1.1:2.0: setting rx_max = 16384 [13574.265125] cdc_ncm 2-1.1:2.0: setting tx_max = 16384 [13574.265667] cdc_ncm 2-1.1:2.0 eth1: register 'cdc_ncm' at usb-0000:00:0d.0-1.1, CDC NCM (NO ZLP), 98:fc:84:xx:xx:xx [13574.270671] usb 3-1.3: new full-speed USB device number 22 using xhci_hcd [13574.376775] usb 3-1.3: New USB device found, idVendor=0461, idProduct=4e9d, bcdDevice= 2.12 [13574.376787] usb 3-1.3: New USB device strings: Mfr=1, Product=2, SerialNumber=0 [13574.376789] usb 3-1.3: Product: Alienware 610M [13574.376792] usb 3-1.3: Manufacturer: DELL For the Resolution it can do: 1x HDMI up to 4K/60 Hz or 1x HDMI up to 1080p/120 Hz For my 2nd Screen this is fine for me, 4k 100hz would be … https://solariz.de/posts/26/01_signal-kde-encrypted-key-kwalletd/ Mon, 26 Jan 2026 00:00:00 +0000 https://solariz.de/posts/26/01_signal-kde-encrypted-key-kwalletd/ signal on Linux offer the way to store your encryption key in your system walled (encrypted) or local as plain-text unencrypted. Actually if you use a full disk encryption a clean text key isn’t a huge issue. But nevertheless to give it a bit more secure feeling I wanted to use KWalletD to store the signal key encrypted. Signal now even prompt you with a question if you want to do so or not. So easy I thought, quickly enabled it; But to verify I took a look at the logfile and found; It is actually trying to use kwallt6 but can’t. The problematic Log The Ugly is; if you dont look at the logs and config, you actually never know if it is encrypted or not. There is no Warning in the GUI that it couldnt save encrypted key or access kwallet :( @Signal Why not display an warning as banne or before login? Debug: Using password store: kwallet6 ... [2:0121/125647.169397:ERROR:dbus/object_proxy.cc:573] Failed to call method: org.kde.KWallet.isEnabled: object_path= /modules/kwalletd6: org.freedesktop.DBus.Error.ServiceUnknown: org.freedesktop.DBus.Error.ServiceUnknown [2:0121/125647.169447:ERROR:components/os_crypt/sync/kwallet_dbus.cc:113] Error contacting kwalletd6 (isEnabled) [2:0121/125647.170338:ERROR:dbus/object_proxy.cc:573] Failed to call method: org.kde.KLauncher.start_service_by_desktop_name: object_path= /KLauncher: org.freedesktop.DBus.Error.ServiceUnknown: org.freedesktop.DBus.Error.ServiceUnknown [2:0121/125647.170350:ERROR:components/os_crypt/sync/kwallet_dbus.cc:82] Error contacting klauncher to start kwalletd6 [2:0121/125647.171088:ERROR:dbus/object_proxy.cc:573] Failed to call method: org.kde.KWallet.close: object_path= /modules/kwalletd6: org.freedesktop.DBus.Error.ServiceUnknown: org.freedesktop.DBus.Error.ServiceUnknown [2:0121/125647.171103:ERROR:components/os_crypt/sync/kwallet_dbus.cc:408] Error contacting kwalletd6 (close) Check it… You can check yourself by opening your signal user dir, normal something like: /home/username/.var/app/org.signal.Signal/config/Signal There you should find a config.json, if inside you find "key": "12345678...." variable you know it’s not using the system wallet. A good config should have only "encryptedKey": "....." not key. I hope in the future Signal automatically remove the plaintext key if a valid encrypted one is found on startup. Reason and fix… The Problem is acutally caused by the default flatpak permissions of Signal. To communicate with KwalletD it need Session bus access but this permission is missing in the default flatpak settings. To enable it go to your Settings > Application Permissions > Flatpak > Signal and enable session bus access: Afterwards restart signal in the terminal and it should not show those errors from before. Finally close signal again and remove the plaintext key variable from the config file mentioned above. Now you should be good. Remember; Kwallet also does not protect you from people having access to your computer. If someone have full access to your device no key / password is save. So always lock your Screen ! https://solariz.de/posts/26/01_tuxedo_laptop_battery_replaced/ Thu, 22 Jan 2026 00:00:00 +0000 https://solariz.de/posts/26/01_tuxedo_laptop_battery_replaced/ Currently I use as my productivity working device a Tuxedo Laptop, the Infinity book pro 14". It is about 1.5yrs old and started acting up in december when it comes to battery. The Battery sometimes was overnight discharging to 0% even the laptop was powered off. Also it showed a reduced maximal capacity. In Linux you can check your battery stats quickly with upower -b the old battery looked like this: native-path: BAT0 vendor: OEM model: standard power supply: yes updated: Wed 21 Jan 2026 02:01:39 PM CET (7 seconds ago) has history: yes has statistics: yes battery present: yes rechargeable: yes state: fully-charged warning-level: none energy: 64.68 Wh energy-empty: 0 Wh energy-full: 64.68 Wh energy-full-design: 80.08 Wh voltage-min-design: 15.4 V capacity-level: Full energy-rate: 0 W voltage: 16.795 V charge-cycles: 69 percentage: 100% capacity: 80.7692% technology: lithium-ion The important thing here is energy-full: 64.68 Wh which is way too low; it should be ~80 Wh. So a loss of 20% capacity in only 69 charge cycles. Well, I am honest-I mostly use the laptop docked on constant power. But when I go to events, I need to rely on it. The biggest issue, besides the smaller capacity, was this weird random discharge, so after checking everything, even reinstalling another OS for testing purposes, it still didn’t fix that. So my rough guess was the battery. Luckily the people at Tuxedo Computers did not hesitate long and also did not require me to send in the laptop (which is a no-go for me because daily need) - they sent me an advanced replacement battery. So I exchanged the old one with the new one. This I love about the Tuxedo laptops, they’re so easy to access. 6 screws, and the full back pops open. Unfortunately, I needed to remove the M2 Drive to acces one of the two more screws for the battery, but that’s it - old one out, new one in. After changing to the new battery the upower -b looked much better again: native-path: BAT0 vendor: OEM model: standard power supply: yes updated: Thu 22 Jan 2026 08:39:56 AM CET (22 seconds ago) has history: yes has statistics: yes battery present: yes rechargeable: yes state: fully-charged warning-level: none energy: 80.08 Wh energy-empty: 0 Wh energy-full: 80.08 Wh energy-full-design: 80.08 Wh voltage-min-design: 15.4 V capacity-level: Full energy-rate: 0 W voltage: 16.654 V charge-cycles: N/A percentage: 100% capacity: 100% technology: lithium-ion And surprise, after night the battery did NOT discharge itself into nirvana. I left it with 100%, powered it off, and when I powered it on this morning it was still 100% as it should be. Yay, happy battery again. https://solariz.de/posts/26/01_photos-hugo-blog-from-pixelfed/ Fri, 09 Jan 2026 00:00:00 +0000 https://solariz.de/posts/26/01_photos-hugo-blog-from-pixelfed/ Recently we had the Digital Independence Day what better Reason could one have to rethink how I use published photos. I actually wanted to integrate them better into my blog already since long. But Digital Independence Day was a good kick in the ass to finally quit Instagram for real. I’m not here to start a big moral talk… let’s focus on the Tech Aspects of what I did. As you may noted we now have the new Menu entry /photos which offers a small Gallery. Instead of integrating another workflow which is doomed to fail because of my lazyness, I need something easy usable from my Android phone. So I had my pixelfed account laying there, mostly collecting dust. Why not start using it again and integrate it into the blog. This post is about how I wired this up. What I wanted (requirements) I only post pictures to pixelfed, no markdown, no ftp not manual sync or upload process Pull new Pixelfed posts automatically (incremental, not re-downloading everything) Only public posts, only images (no videos, no private stuff) Hugo-friendly structure: each post becomes a page bundle Gallery view with: thumbnails grid lightbox modal responsive images (srcset) some metadata and a link back to Pixelfed During build, generate: square thumbnails watermarked bigger sizes Integrate into my deployment process The Hugo photos page Let us start with the simple stuff, creating /photos. I added a dedicated /photos section and a custom list template. A 3-column grid layout with Yearly horizontal dividers would be best and easy to get responsive. The Structure looks like this: ├── assets/ │ ├── css/ │ │ └── extended/ │ │ └── photos.css Gallery & lightbox styles │ └── watermark.png Used for image watermarking │ ├── layouts/ │ └── photos/ │ └── list.html Gallery template with lightbox | └── content/ └── photos/ ├── _index.md Gallery page frontmatter └── YYYYMMDD-HHMMSS/ photo directories, one per post ├── photo.jpg Actual Photo └── index.md Photo metadata frontmatter ok, as you can see this is actually not that much. The /content/photos/YYMMDD-HHMMSS is later auto-generated by a python script which pulls from pixelfed. All other things are static, add once and forget :) _index.md --- title: "Photos" layout: "photos" summary: "Photo gallery" --- Lorem Ipsum photos.css To keep the post a bit more clean I posted my css file in a gist here to not spam the long CSS here. list.html This is where the actual “magic” happens. In my post about how I create my OG Images automatically, I learned a lot about hugos capabilities in image manipulation. So why not use it to do my gallery thumbnails and final images? {{- define "main" }} <header class="page-header"> <h1>{{ .Title }}</h1> {{- if .Content }} <div class="post-description"> {{ .Content }} </div> {{- end }} </header> <div class="photo-gallery" itemscope itemtype="https://schema.org/ImageGallery"> {{- range (.Pages.GroupByDate "2006") }} <div class="year-section"> <h2 class="year-header">{{ .Key }}</h2> <div class="year-photos"> {{- range .Pages.ByDate.Reverse }} {{- $image := .Resources.GetMatch (.Params.image) }} {{- if $image }} {{/* Generate square thumbnail (no watermark) */}} {{- $thumb := $image.Fill "320x320 Center MitchellNetravali webp q72" }} {{/* Get watermark PNG */}} {{- $watermark := resources.Get "watermark.png" }} {{/* Generate large version with watermark overlay */}} {{- $largeTmp := $image.Fit "1400x1400 MitchellNetravali" }} {{- $large := $largeTmp.Filter (images.Overlay $watermark (sub $largeTmp.Width 151) (sub $largeTmp.Height 43)) }} {{- $large = $large.Process "webp q88" }} {{/* Generate small version with watermark overlay */}} {{- $smallTmp := $image.Fit "740x740 MitchellNetravali" }} {{- $small := $smallTmp.Filter (images.Overlay $watermark (sub $smallTmp.Width 151) (sub $smallTmp.Height 43)) }} {{- $small = $small.Process "webp q84" }} {{- … https://solariz.de/posts/25/windows-git-ssh-agent/ Wed, 31 Dec 2025 00:00:00 +0000 https://solariz.de/posts/25/windows-git-ssh-agent/ Fixing Git Clone SSH Agent on Windows Last days I run Git on Windows for a quick upload of something, tried cloning the private repo over SSH, and it bombed out with a permission denied error. Even though a plain SSH test to GitHub worked fine. As said, normally I work on my Linux Laptop but during Holidays I –had– to build some csharp stuff in windows. So old pains came back :) fatal: Could not read from remote repository. Here’s what I saw: git clone git@github.com:solariz/xxxxxxx.git test Cloning into 'test'... git@github.com: Permission denied (publickey). fatal: Could not read from remote repository. Please make sure you have the correct access rights But verifying SSH alone: ssh -T git@github.com Hi solariz! You've successfully authenticated, but GitHub does not provide shell access. That checks out, so why the clone failure? If I check ssh-add -L it correctly showed me all of my keys in the agent. Turns out Git on Windows was using some wrong SSH client by default. Fixed it by pointing Git to the built-in OpenSSH. git config --global core.sshCommand C:/Windows/System32/OpenSSH/ssh.exe Retried the clone: git clone git@github.com:solariz/xxxxxxx.git test Cloning into 'test'... remote: Enumerating objects: 3, done. remote: Counting objects: 100% (3/3), done. remote: Compressing objects: 100% (2/2), done. remote: Total 3 (delta 0), reused 0 (delta 0), pack-reused 0 (from 0) Receiving objects: 100% (3/3), done. It works now. git windows commit Editor While at it, I usually skip vim in Windows shells since I don’t have it installed. Wanted to set Sublime Text as my Git editor instead. Tried this: git config --global core.editor "C:/Program Files/Sublime Text/subl.exe -w" But spaces in the path messed it up. Escaping didn’t help either. Quick fix: Add the Sublime folder to your Windows PATH, then: git config --global core.editor "subl.exe -w" Now Git opens Sublime for commits or whatever, no hassle. Small tweaks like these save time when you’re jumping between systems. https://solariz.de/posts/25/starcitizen-4-5-vulkan-crash-startup/ Thu, 18 Dec 2025 00:00:00 +0000 https://solariz.de/posts/25/starcitizen-4-5-vulkan-crash-startup/ With today’s released Patch 4.5, Star Citizen switched to using the Vulkan renderer by default. This is a big step forward. But unfortunately, for some players, the game simply does not start anymore. You run the launcher, and then the game immediately crashes. I ran into this myself, and eventually, after trying everything, I found I got it working. Also a recommendation after you got Vulkan running I wrote about best Performance Settings for Star Citizen Vulkan in my citizen-history blog. NVIDIA User on AMD CPU I have an RTX 5080 on a Ryzen 9800X3D; when the game starts, it immediately crashes. In the LIVE Directory, you find a Game.log file showing clearly that the error is the initialization of the Vulkan Renderer: Failed to create Vulkan logical device GPUDeviceVulkan.cpp:825 vkCreateDevice: Failed to create device chain STATUS_CRYENGINE_FATAL_ERROR (0x2BADF00D) Game.log <2025-12-18T16:56:32.471Z> VulkanMetric - Device (NVIDIA GeForce RTX 5080) [GPU (Discrete)] ID:11266 <2025-12-18T16:56:32.471Z> VulkanMetric - Driver 591.59.0.0 <2025-12-18T16:56:32.471Z> VulkanMetric - VulkanAPI 1.4.325 <2025-12-18T16:56:32.471Z> VulkanMetric - Supports 6 Queue Families <2025-12-18T16:56:32.471Z> VulkanMetric - Device (NVIDIA GeForce RTX 5080) valid GPU <2025-12-18T16:56:32.471Z> VulkanMetric - Device (AMD Radeon(TM) Graphics) [GPU (Integrated)] ID:5056 <2025-12-18T16:56:32.471Z> VulkanMetric - Driver 2.0.353 <2025-12-18T16:56:32.471Z> VulkanMetric - VulkanAPI 1.4.315 <2025-12-18T16:56:32.471Z> VulkanMetric - Supports 5 Queue Families <2025-12-18T16:56:32.471Z> [VK] Chosen Vulkan GPU Device (NVIDIA GeForce RTX 5080) Driver Version (591.59.0.0) Vulkan API (1.4.325) <2025-12-18T16:56:32.471Z> [VK] Current Device: (NVIDIA GeForce RTX 5080) VendorID: (4318) <2025-12-18T16:56:32.471Z> [VK] Current Device: (NVIDIA GeForce RTX 5080) VendorID: (4318) <2025-12-18T16:56:32.471Z> [VK_INFO] vkCreateDevice layer callstack setup to: <2025-12-18T16:56:32.471Z> [VK_INFO] <Application> <2025-12-18T16:56:32.471Z> [VK_INFO] || <2025-12-18T16:56:32.471Z> [VK_INFO] <Loader> <2025-12-18T16:56:32.471Z> [VK_INFO] || <2025-12-18T16:56:32.471Z> [VK_INFO] <Device> <2025-12-18T16:56:32.471Z> [VK_INFO] Using "NVIDIA GeForce RTX 5080" with driver: "C:\WINDOWS\System32\DriverStore\FileRepository\nv_dispi.inf_amd64_4646b47d9477048e\.\nvoglv64.dll" <2025-12-18T16:56:32.506Z> GPU benchmark: 14.52 ms (Adapter index: 0) <2025-12-18T16:56:32.622Z> [VK_ERROR] terminator_CreateDevice: Failed in ICD C:\WINDOWS\System32\DriverStore\FileRepository\nv_dispi.inf_amd64_4646b47d9477048e\.\nvoglv64.dll vkCreateDevice call <2025-12-18T16:56:32.622Z> [VK_ERROR] vkCreateDevice: Failed to create device chain. As you see in the above Log I even tried to revert the NVIDIA Driver to an older Version 591.59 which did not helped. The Ugly Crash: Now I run all the common things you do: driver uninstall, clean reinstall, game repair, config delete, default settings, Vulkan SDK, … nothing helped for me. What really helped for me To skip the blabla, it was something absolutely not obvious to me. As said, everything I tried didn’t work, so you get desperate and you try all the stuff you hear. Ensure that ASLR IS TURNED ON: For me it was a combination of Exploit Protection ASLR and Custom Application Profile. Open your Windows Settings and go to App & browser control at the bottom, open the Exploit protection settings First, ensure that force randomization for images (Mandatory ASLR) is set to “On by default” Also, the next Option Randomize memory allocations to “On by default” This was the case for me already before, but above you find a Tab “Program settings” which has overrides for some programs. Open the Program Settings Tab and check carefully if Star Citizen is in there somewhere; if so… remove the entry. Beside that before I did a DDU + NV Clean install of … https://solariz.de/posts/25/2025-ram-price-spikes/ Mon, 15 Dec 2025 00:00:00 +0000 https://solariz.de/posts/25/2025-ram-price-spikes/ Uhhm sometimes you realize how lucky you been buying your PC Upgrade at the right time. Shortly after Gamescom 2025 I upgraded my home desktop to new CPU and Motherboard also decided to make the jump from DDR4 to DDR5… I read about the recent explosion in RAM prices but little did I realize how huge it actually is on some specific RAM Models. An eye opener screenshot, the Store page where I bought my 64 GB DDR5 MT6000 Kit. Same Kit in September vs exactly 3 Month later in December. 264 € vs 1.748 € Uhmmm… But why? RAM prices keep climbing because the big memory makers are chasing higher profits from AI and enterprise contracts. Hyperscalers building AI data centers are locking down huge long-term deals for HBM, DDR5, and LPDDR5X, leaving almost no wafer capacity for the consumer-level DDR4 and DDR5 sticks everyone else needs. At the same time, Samsung, SK Hynix, and Micron are shutting down older production lines and focusing on pricier components, which squeezes the PC and mobile market even more. Micron already walked away from retail, and there aren’t any new fabs coming online soon, so the market is more consolidated than ever. That gives the remaining suppliers even more control over pricing. Since mid-2025, consumer RAM costs have basically tripled. DIY builders are waiting out upgrades, and even big PC brands can’t get enough inventory. Unless demand from AI projects cools off or new production capacity shows up, DRAM isn’t a cheap commodity anymore—it’s a pressure point across the entire electronics supply chain. PC-Part Picker have a good general average graph which is showing the Range of the Impact: Back to Normal? Not what I hoped but many Experts expect the DRAM crunch to drag on for years, with only a small hint of balance showing up around late 2027 or 2028. Through most of 2026, AI projects will keep hogging wafer allocations, so DDR5 and LPDDR5X stay pricey while DDR4 pops in and out of stock. Even when SK hynix lights up M15X or Micron ramps its Japan fab, most of that fresh output goes straight into HBM for datacenter GPUs, not the consumer kits we actually buy. Memory makers seem happier protecting margins than cranking out extra volume, which keeps oversupply risk tiny but locks in scarcity pricing. Analysts figure DRAM will sit about 50-100% above pre-2025 price levels until at least 2027, easing only when new global capacity and a cooler AI buildout finally line up. Honestly, cheap RAM might be gone for good. AI turned DRAM into a strategic resource, not the impulse add-on at checkout. If you’re planning a build, take a careful look, the time you shop for Brand Name or optics in a Gaming Rig are sadly over :( https://solariz.de/posts/25/winscp-1password-ssh-agent/ Wed, 10 Dec 2025 00:00:00 +0000 https://solariz.de/posts/25/winscp-1password-ssh-agent/ My Setup and the Problem I use Linux on my main work laptop, where remote access is effortless: I can SSH with Kitty’s remote edit kitten or mount remotes via KDE Dolphin. But at home, my desktop runs Windows 11 for gaming (thanks to stupid anti-cheat restrictions). I normally just game on the desktop; but sometimes I have to do some work from the Desktop, already at the desk and need to quickly edit stuff remote. Most time I just used scp. But remembering in the past as win user, I used WinSCP for remote file editing. Found out quicky it natively refuses to connect using my SSH key from the already running agent in 1password. Solve it quickly Ensure 1password SSH Agent is running (Settins > Developer) Disable the Windows built in OpenSSH Agent Service (Services.msc > find OpenSSH Auth… and set it to Disabled) Get winssh-pageant Install it and ensure it is run in Autostart adding the Param --no-pageant-pipe In WinSCP under Authentication of a Session chosse “Allow Agent forwarding” Thats it, your winscp should work as intended using the SSH Keys in your 1password. https://solariz.de/posts/25/pockethernet-v2-test/ Mon, 08 Dec 2025 00:00:00 +0000 https://solariz.de/posts/25/pockethernet-v2-test/ Pockethernet 2: Hands-On Test of the Revived Network Tester Some years ago, at a event, I could try the original Pockethernet V1. At this time I owned a bigger Tester with display, mostly not with me because to heavy for every day carry. The Pocketethernet looked like a interesting0 small tester for cables and simple network checks, but after playing a bit with it I didn’t like it much. The App was mediocre, I had issues with the connection but now, fast forward: The V2 came out a couple months ago, after the company was gone for years. I decided to buy one at about 249€. It’s comperitative cheap for a tester with TDR. Now in my everyday backpack more than half a Year, I’ve used it on several jobs and events and it holds up so far. No regrets, but it’s still not without issues. Pockethernet V2 Hardware Pockethernet V2 (9 x 6.5 x 2.5 cm) fits nice in a pocket or bag. The case have more solid gray ends instead of clear plastic that seemed to broke easy on V1. In my use, it feels strong enough. Battery is bigger, 900mAh from 750mAh, also they state that the circuits are more efficient for longer battery life. Which is now less of a problem because charging is USB-C, better than old Micro-USB. Bad part: Not so good display of LEDs for quick checks. There’s a “quick test” with a button for basics like wiremap, PoE, link, and DHCPv4, shown on four LEDs. But it’s not very helpful if you dont learn their color and blinken codes by mind. I Wish they would have LEDs like the Link Sprinter which light up as a symbole, one for hardware connectivity with color for Connection speed, one with DHCP info received, one with GW reachable and one with internet test success. Maybe another color for PoE. As said they do have some quick tests but the way of signaling it could be improved. I still love the LinkSprinter and its symbolic LEDs for speed, PoE, and such are clear and fast without a phone. Good side: Bluetooth LE to the app works great, way better than LinkSprinter’s WiFi crap, which drops on my Google Pixel constantly complains about no internet even you tell it not to worry. The Pockethernet Bluetooth connects easy, even at some meters of distance withing few seconds. Due it is BT LE you dont need to pair it with the phone, just click the connect button in the app and nearly instant it tells connected. Tested on my Google Pixel and a Samsung Galaxy S22 no issues in months. From their Webpage, compare V1 vs V2: Core Features The main reason I got it, is cheap TDR for cable tests and it does it very well. Cable Testing Wiremap: Checks pins for TIA 568A/B, shields, shorts, opens. Use the remote wiremap adapter at the other end. It worked well. Extra remotes for finding multiple cables are coming, says the Webpage but currently I wouldn’t bet on that to much. They told that since about a year now and there has been no Update. TDR: Measures length and faults like shorts or splits. Resolution is 0.4m now (from 0.8m on V1), with graphs for crosstalk and issues. Some users on Reddit reported wrong lengths measures but for me +/-1m is totally fine; I don’t expect super exact measurements on cheap every day carry tool. If we do certification we get the big tools in which are proper calibrated but not for quick every day testing. Extras: Toner with tones for probes, and port blinker to flash LEDs for tracing. Network Checks Link Speed: Detects Portspeed Announcements from 10M to 10G (includes 2.5G/5G). (Link only up to 1G) PoE: Tests voltage, type (AF/AT/BT), class. Has load test for real checks, even on passive. Discovery and VLAN: Gets CDP/LLDP for switch and VLAN info. Detects VLANs Tags IP: IPv4 with DHCP or static, plus IPv6. Pings targets, shows external IP WiFi Bridge: Beta mode for Ethernet to WiFi. I tried it, can be handy if you forgot your laptops RJ45 dongle. The App: Workflow could be improved easy The App, It’s way better than V1, useful graphs for TDR and wiremap, PDF reports with photos for sharing Sample PDF Report. But problems: On my Pixel, things … https://solariz.de/posts/25/unifi-ubiquiti-10g-rj45-adapter/ Mon, 17 Nov 2025 00:00:00 +0000 https://solariz.de/posts/25/unifi-ubiquiti-10g-rj45-adapter/ Got one Ubiquiti Unifi 10G Ethernet Adapter, model UACC-Adapter-RJ45-USBC-10GE. In the past, I’ve mostly got StarTech adapters for 10G Copper, but right now they’re either sold out or going for double the price. The Unifi 10G is 150€ plus tax, it fits what I need. First off, it plugged right in on my Linux laptop running kernel 6.14 and a MacBook Air. No fuss with drivers or anything. Here’s what the kernel spit out on Linux: thunderbolt 0-1: new device found, vendor=0xa5 device=0x27 thunderbolt 0-1: Ubiquiti Inc. UACC-Adapter-RJ45-USBC-10GE thunderbolt 0-0:1.1: new retimer found, vendor=0x8087 device=0xd9c pcieport 0000:00:07.0: pciehp: Slot(12): Card present pcieport 0000:00:07.0: pciehp: Slot(12): Link Up atlantic 0000:03:00.0 eth0: atlantic: link change old 0 new 10000 TCP: eth0: Driver has suspect GRO implementation, TCP performance may be compromised. Speeds It showed up as expected, and speeds looked good from the start. One thing to Note: this adapter only works on Thunderbolt 3 or 4. Forget about regular USB-C it won’t work. Some blogs go negative about Adapters PCIe lane requirements, like how your host needs enough lanes to hit full speed. But in my case, that wasn’t a problem. I pulled 9.37 Gbit with iperf3, and OpenSpeedTest clocked around 9.6 Gbit. That’s basically as good as it gets for 10G, considering overhead. I tested it hooked to a Unifi XG 10Gbit switch over a 10m CAT 7 cable with a default MTU of 1500. ### iperf3 -P 4 [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 2.47 GBytes 2.12 Gbits/sec 764 sender [ 5] 0.00-10.00 sec 2.47 GBytes 2.12 Gbits/sec receiver [ 7] 0.00-10.00 sec 3.03 GBytes 2.60 Gbits/sec 657 sender [ 7] 0.00-10.00 sec 3.03 GBytes 2.60 Gbits/sec receiver [ 9] 0.00-10.00 sec 3.02 GBytes 2.59 Gbits/sec 752 sender [ 9] 0.00-10.00 sec 3.02 GBytes 2.59 Gbits/sec receiver [ 11] 0.00-10.00 sec 2.39 GBytes 2.05 Gbits/sec 712 sender [ 11] 0.00-10.00 sec 2.38 GBytes 2.05 Gbits/sec receiver [SUM] 0.00-10.00 sec 10.9 GBytes 9.37 Gbits/sec 2885 sender [SUM] 0.00-10.00 sec 10.9 GBytes 9.36 Gbits/sec receiver ### iperf3 -P 4 -R [ ID] Interval Transfer Bitrate Retr [ 5] 0.00-10.00 sec 2.74 GBytes 2.35 Gbits/sec 3820 sender [ 5] 0.00-10.00 sec 2.73 GBytes 2.35 Gbits/sec receiver [ 7] 0.00-10.00 sec 2.43 GBytes 2.09 Gbits/sec 3908 sender [ 7] 0.00-10.00 sec 2.42 GBytes 2.08 Gbits/sec receiver [ 9] 0.00-10.00 sec 2.76 GBytes 2.37 Gbits/sec 2279 sender [ 9] 0.00-10.00 sec 2.76 GBytes 2.37 Gbits/sec receiver [ 11] 0.00-10.00 sec 2.99 GBytes 2.57 Gbits/sec 4103 sender [ 11] 0.00-10.00 sec 2.99 GBytes 2.57 Gbits/sec receiver [SUM] 0.00-10.00 sec 10.9 GBytes 9.38 Gbits/sec 14110 sender [SUM] 0.00-10.00 sec 10.9 GBytes 9.37 Gbits/sec receiver Features It’s built on an AQC113 chip from Marvell, which handles the 10GBASE-T side. The chip is capable of doing Multigig NBASE-T at 10G/5G/2.5G/1G/100M/10M. Feature wise some to note: 8 QoS Classes WoL LSO, RSS, DCA PTPv2 Multigen PCI Express: 4>3>2 Build It is build all metal, fanless, so it stays quiet. I’ve read other reviews where it hits 44-45°C under load, around 40°C idle, which matches what I noticed, gets warm but nothing crazy. It’s smaller than the StarTech TB 10G adapters I own a few of, so that’s a win for portability. But tbh. I normally not taking those on travel anyway. All in all, a solid buy. If you’re eyeing a portable 10G upgrade and have the right ports, it’s worth considering. https://solariz.de/posts/25/hugo-dynamic-ogimage/ Thu, 13 Nov 2025 00:00:00 +0000 https://solariz.de/posts/25/hugo-dynamic-ogimage/ To be honest, the switch from a classic CMS to Hugo had a steep learning curve at first, but Hugo is incredible. Even though we have static pages, you can do a lot of “dynamic” stuff. The Problem From the start when I began using Hugo, I had an issue with the open graph images, per default you can specify one image as default in hugo and use post header mages as og:image. Well - I try to keep my blog posts free of images which are not beneficiary to the content. Screenshots are fine but general pseudo matching placeholder images just that you have something for the header 🤷‍♂️. So the problem that came up was all posts ended up with the same image when you linked to the article. The Solution I found an Article from aarol who had similar issue some years back. Thanks for writing about it! Thats the reason I’m doing this post, it might help someone else avoid the same. The Idea is simple: Instead of having a dull placeholder we can use hugo image manipulation functions to automatically create an fitting og image for each post. You specify an base background and it automatically take the post title and write an og.jpg to the filesystem and insert the meta accordingly. How to ? I Link to the Original Article from aarol here as base. I changed and updated it for my implementation a bit. Partial override for the opengraph template. Always take the latest version from here before you start and put it into: layouts/partials/opengraph.html I removed the original static / header based og:image part from it completely because I dont want that. Afterwards I added this modified code: {{/* Generate opengraph image */}} {{- if .IsPage -}} {{ $img := resources.Get "og_base.jpg" }} {{ $Font := resources.Get "/Lexend.ttf"}} {{ $img = $img.Filter (images.Text .Page.Title (dict "color" "#ffffff" "size" 58 "linespacing" 2 "x" 42 "y" 272 "font" $Font ))}} {{ $img = resources.Copy (path.Join .Page.RelPermalink "og.jpg") $img }} <meta property="og:image" content="{{$img.Permalink}}"> <meta property="og:image:width" content="{{$img.Width}}" /> <meta property="og:image:height" content="{{$img.Height}}" /> <!-- Twitter metadata (used by other websites as well) --> <meta name="twitter:card" content="summary_large_image" /> <meta name="twitter:title" content="{{ .Title }}" /> <meta name="twitter:description" content="{{ with .Description }}{{ . }}{{ else }}{{if .IsPage}}{{ .Summary }}{{ else }}{{ with .Site.Params.description }}{{ . }}{{ end }}{{ end }}{{ end -}}"/> <meta name="twitter:image" content="{{$img.Permalink}}" /> {{ end }} Different to Aarols implementation I simplified it and only use one generic font and block for the Title only. I just want an eye Catcher where display of the page.Title is enough. But beside that same same. Re-build you page and you should see all the og.jpg images created. As a tip, make sure your base image isn’t to large or all generations will be even larger. If you just use a plain background use png, otherwise if you use a picture like in my example use a compressed jpg. The Result The Actual generated result for this Article will look like this: And Everytime a page link to one of my articles is shared somewhere the opengraph protocol is used, it will look like this Example from a Paste to Discord: