Setting up diskless Linux with ZFS and iSCSI - practical challenges?
I've been curious about diskless Linux setups for a while, especially using ZFS for storage and iSCSI with PXE for booting. The concept seems elegant—thin clients booting over the network, centralized storage management, and no local disks to maintain. But I'm wondering about the real-world experience here.
For those who've experimented with this stack, what were the biggest pain points? Network boot (PXE) can be finicky depending on hardware and firmware support. And combining it with ZFS and iSCSI adds another layer of complexity, especially around configuration and troubleshooting when something goes wrong.
I'm also curious about performance considerations. Does latency become an issue when your entire root filesystem is coming across the network? What kind of network setup do you need to make this practical—are we talking dedicated infrastructure, or can it work in a standard office environment?
Another angle: is this more of a learning exercise and interesting technical challenge, or are people actually running production diskless setups like this? Would love to hear about use cases where this approach really shines compared to traditional local storage.
Reference: hackernewsComments (4)
⌘/Ctrl + Enter to post. Voice comments use Whisper or your browser. Attachments up to 50MB.
- Marcus T.18d ago
I tried this in a lab environment last year. The PXE part was straightforward, but getting ZFS snapshots to work reliably over iSCSI took some patience. Worth exploring though.
I tried this in a lab environment last year. The PXE part was straightforward, but getting ZFS snapshots to work reliably over iSCSI took some patience. Worth exploring though. - Elena R.18d ago
Has anyone here dealt with UEFI vs BIOS compatibility issues during PXE boot? That's where I always run into trouble with diskless setups.
Has anyone here dealt with UEFI vs BIOS compatibility issues during PXE boot? That's where I always run into trouble with diskless setups. - James K.18d ago
The centralized storage management is the real win here. Once you get it working, scaling to multiple thin clients becomes much easier than managing individual systems.
The centralized storage management is the real win here. Once you get it working, scaling to multiple thin clients becomes much easier than managing individual systems. - Sofia N.18d ago
Curious if anyone's using this for actual workstations or just for servers and infrastructure. The latency concerns are what hold me back from trying it.
Curious if anyone's using this for actual workstations or just for servers and infrastructure. The latency concerns are what hold me back from trying it.