Networking

In my previous post, I mentioned that have a remote portal machine in my friend’s house where I use RustDesk to access into. However, in some use case, like access LAN resources and managing online services across multiple devices, the best practice is not using a remote desktop tool but having a VPN tunnel. Old Stories In the very early days, I used to use Hamachi to play LAN games like Counter Strike, Warcraft 3 and Minecraft with my internet friends. After I started doing homelab, I have used ZeroTier for quite a while before Tailscale came out. ...

On the Hardware This post is originally wrote earlier this year, back then I bought an used HP EliteDesk 705 G4 Mini as a cheap NUC alternative. For its price and availability, I choose this model over Intel NUC but it turned out NUC may be the better option. It comes with a 4 core 8 thread CPU, Ryzen 5 2400GE with 35W TDP. If set it up as a Proxmox server with enough RAM, it can run about 2-3 VMs to run some light services. But now, my plan is to configure it as a remote portal with RustDesk client. ...

Two month after flashing the Flint 2 GL-MT6000, I got its little brother GL-MT3000 in my hands as a powerful mini travel router runs stable build of OpenWrt. There are a few comparable routers in the same market, such as its older model GL-MT1300 v1 which is not very appealing price-wise; Cudy TR3000 v1 which is a direct competitor by its naming but has confusing variations that is not well supported by OpenWrt; and yet another older model Cudy TR1200 which is not available in the US. ...

After almost two years later, the hardware of my Openmediavault NAS is going to be upgraded from Vostro to Precision. Which is a big leap, that I don’t have to do any hardware modification to use Xeon CPUs. Thanks for my friend’s donation. This junkstation comes with E3-1245, 16GB and 4 SATA HDD port. What I added is a ASMedia 106x SATA PCIE card for 2 more SATA HDD capacity (6 HDD in total), and a PCIe to NVMe adapter for accommodating a OS drive (128GB SSD). ...

Background Last month, Cisco researchers detected over 1,000 Ollama instances within the first 10 minutes using Shodan scanning on 11434 port. Other services such as vLLM/llama.cpp/LangChain on 8000, LM Studio on 1234, GPT4All on 4891, are also identified. Later, Censys found 10.6K Ollama instances publicly available online, and 1.5K of these instances respond to prompts. That poses not only a great security risk of RCE, injection and poisoning, but also possible to expose private chat memory via unauthorized prompting. ...