rEFInd is _so_ much simpler: one efi entry, one text config file in the efi partition, nothing that needs to change when the kernel updates, and no massive pile of templating and moving parts to mysteriously break dumping you at an impenetrable grub “rescue” shell.
does refind support secure boot and measured boot? I loathe pretty much anything systemd but systemd-boot gives me this with zero effort, and it's legitimately useful
iSCSI is great and underated, I've been running iSCSI volumes for OrangePI boards which had only SD card support for years. Performance was much better on Gigabit Ethernet vs the SD card.
> UEFI fixes that to some extent, but it’s a pain to maintain the UEFI entries manually and change them every time the kernel updates.
… you don't have to update the UEFI entries every time the kernel updates. (I guess you might if you do like a kernel w/ CONFIG_EFI_STUB, and you place the new kernel under a different filename than what the UEFI boot entry point to then you might … but I was under the impression that that'd be kind of an unusual setup, and I thought most of us booting w/ EFI were doing so with Grub.)
I have 2 entries, /efi/current.efi and /efi/old.efi... when I upgrade, I copy current to old, and copy my new kernel out of /boot as current and reboot.
Or use UKI and throw the current kernel to /efi/boot/bootx64.efi; there's plenty of solutions to sane bootloader/kernel management if you're willing to invest 15 minutes into the topic and not act like it's scary and complicated (it really is the opposite).
Nice. I'm extra fond of ZFS backed network root filesystem, because it lets you put an OS on ZFS without needing to deal with ZFS support in that OS. (One of these days I want to try OpenBSD with its root on NFS on ZFS, either from Linux or FreeBSD.)
Well, iSCSI is a standard, so chances are better that it's supported in a non-Linux OS, e.g. MS Windows. Years ago I booted a Windows (7, iirc) client that way, but gave up on it (too much hassle and performance limited by the network) when SSDs became cheap.
I don’t have direct experience, but when I looked into it my takeaway that NBD was unable to reliably deal with network interruptions as well as iscsi.
I've done a lot of headless/diskless stuff. I haven't done much for years, because my NAS only has gigabit Ethernet ports. I can cascade them and get four Gbps downstream, but it's still painful.
I have recently upgraded my house to 10Gbps Ethernet, with only one room still stuck at gigabit, and unfortunately, it's my main office. I'm working on getting the drop there now (literally, just taking a break here).
Even once I'm done, accessing an iSCSI drive over 10GbE will be 4-8 times slower than a local NVMe drive, but it will sure be a lot better than it was!
Ideally, I could run VMs on the NAS and have great performance, but that's another hardware upgrade...
Using a proper NIC (Chelsio) with their iSCSI accelerator will boost your iSCSI performance significantly.
Another alternative is Mellanox with RDMA. You need CX4+ for optimal performance over TCP/IP, while the cheap CX3 is excellent with IPoIB.
If you have a lot of packet drops and retransmissions, another option for boosting iSCSI performance is getting a network switch with a lot of memory for packet buffering. This helps with incast congestion. There are special switches with gigabytes of memory built for this.
NVMe-oF is the best protocol with least overhead for network drives, with a proper setup you lose only 10-20% latency compared to local disk even with Intel Optane. Throughput should be almost similar.
Really I wonder how this turns out to be diskless while you're clearly accessing a disk/drive over the network. Shouldn't we refer to this as network boot?
This could be an interesting setup for booting off a NAS like Synology or QNAP. I haven't really used iSCSI, it's intimidating how much prep this takes...
iSCSI seems intentionally obscure. One of the improvements I made to NBD was invent a simple, standardized URI format so that you can specify servers easily, eg:
Requests and responses are pipelined so at least you're not serializing on round trips. However fundamentally if there is lots of latency, then you're going to be affected in some way. Usually we see problems where the OS accessing the remote drive times out which can sometimes be worked around by increasing timeouts, if you can work out how. (Latency is going to affect every block device protocol in about the same way)
You can actually see what happens quite easily if you've got an OS image handy. With a Fedora VM image:
("$uri" expands to the NBD URI of the nbdkit server which qemu can parse natively)
Even that 1 second delay is painful since it turns out that booting is quite serialized. Edit: I turned down the latency to 50ms in the example which is a bit more realistic. Still painful.
I kind of expect the performance is worse, but one neat thing is that iscsi is a block device, so you could run e.g. disk crypto, volume management or whatever on it. Not to mention any FS. And you don't need to deal with NFS or RPC.
I used similar ipxe setup for robotic cluster - every robot booted from the same thing, then kubernetes managed the containe orchestration. it was fun.
NFS diskless was easier for me to setup when I was doing it.
THe caveat was, you needed readonly root, so that meant freezing the OS, anything that needed changing was either stored in a ram disk (that you need to setup) or a per host nfs area (kinda like overlayfs, but not)
You can download the rootfs, extract it to a ramdisk, and just run in memory. This is fast for everything. Unfortunately, memory just got super expensive. Fortunately, Linux requires ~no memory to do many useful things.
You might find it worth upgrading to 10gbps if you continue to go down this road. The Mikrotik CRS-309 has served me well, and a couple Intel X520-DA2s. I believe those NICs can do iSCSI natively, and pass the session to the operating system with iBFT.
SFP28 might be cheap enough now too, I'm not sure...
Pretty cool! You could also boot into an ephemeral minimal initrd that displays a selection menu instead of doing it in iPXE. That would grab the new kernel and initrd from the network and kexecs it without reboot.
No. PXE boots routinely load the kernel and initramfs directly into RAM with no local disk involved. The initramfs then mounts the real root FS over the network.
Then anaconda or whatover os installer picks up and installs the OS in a PXE install sequence when there is a local disk.
It is relatively easy to configure. Just install Linux after windows, and Linux will generally automatically setup a boot-selection screen for you. The installer should detect windows and even shrink the partitions for you.
You can install a prettier looking boot selection menu like rEFInd, but the default works just as well, and I think the mainstream distros all setup secure boot too. On my pc it was very easy, on my (8yr old) laptop I had to add some secure boot keys and the bios was very confusing, using terms that didn’t seem to match what they should have been.
My setup has worked almost entirely flawlessly and survived updates from both OSes. Only issue being “larger” windows feature updates putting windows back as the first OS in the list, but that happens maybe once or twice a year? And it’s a quick bios change to fix the order.
I've never had this experience dual-booting, neither with UEFI nor Grub. I've been using Linux for nearly 15 years, 13 of those dual booting. Dozens of systems from laptops to desktops. Windows would always purge the boot entries. I'd have to manually fix booting, constantly. This happened with Ubuntu, Arch Linux and recently NixOS and through all the Windows editions till 11. I had to install Windows for a lan party recently and lo and behold on the second day NixOS is gone from the boot list and unbootable. Nothing of value is lost when it gets purged, but it's a damn annoying tax to pay just to be able to play video games.
Luckily gaming now works well enough that the only reason to use Windows was gone. Well, apart from some online games played during lan parties.
The Debian installer is less than optimal for repartitioning.
The Linux NTFS resizing code also has a tendency to trigger data corruption. Not really Linux' fault, but it's a good reason to do partitioning from inside of Windows, which can be a pain already.
Another issue I've run into is Windows creating a very small (~300MiB) EFI partition that barely fits the Windows bootloader, let alone a Linux bootloader and kernel. You can resize and recreate the partition of course, but reconfiguring Windows to use a different boot partition is a special kind of hell I try to avoid.
For Debian and most other distros, secure boot isn't a problem. Installers are all using a signed, trusted-by-default bootloader.
There are some exceptions (some hardware from Microsoft doesn't trust the third party certificate used, for instance, and Red Hat Enterprise has their own root of trust if you opt into that), but they're very rarely ever an issue.
I know it was just a convenient pretext for a learning journey, but do not come away from this thinking llama.cpp needs to be compiled on Windows before use. The GitHib project has a cornucopia of pre-built artifacts to use.
I don't know if I'm understanding your question, but ZFS actively corrects data on disk when it finds a checksum error [0]. Those checksum errors can be found when accessing that data, or doing a 'scrub' action that scans the whole volume to check integrity.
ZFS supports self healing, you do not have scrub, it will be corrected during a bad read as long you have a copy. Metadata has 2 copies by default for additional safety for a single disk.
I like the concept, I don't like the presentation or the reasoning.
But this idea or ones much like it have been presented before...
"Hassle-free diskless Virtual Machines with Xen and Alpine Linux"
https://jonnytyers.wordpress.com/2016/08/16/hassle-free-disk...
"How to make a Diskless Virtual Machine (KVM)"
https://github.com/lispydev/diskless-kvm
And if you just want a dead easy PXE server for bare metal...
https://www.iventoy.com/en/index.html
Friends don't let friends use grub.
rEFInd is _so_ much simpler: one efi entry, one text config file in the efi partition, nothing that needs to change when the kernel updates, and no massive pile of templating and moving parts to mysteriously break dumping you at an impenetrable grub “rescue” shell.
> rEFInd is _so_ much simpler
TBH I've only used it for Hackintoshes and BSDs. I don't know why you'd want it.
But I am finding this hard to post on account of this big hairy grey thing with a long prehensile nose that's in the way...
They only work on UEFI. I don't like UEFI. I have kit in near-daily use that doesn't have UEFI at all.
GRUB works on BIOS, UEFI, x86, Arm, whatever. It boots Linux, all the BSDs, Haiku, whatever. So GRUB wins.
I hate GRUB for its hostility and unfriendliness and impenetrability... But it does the job on more kit than, well, any of the alternatives.
does refind support secure boot and measured boot? I loathe pretty much anything systemd but systemd-boot gives me this with zero effort, and it's legitimately useful
Or just use systemd-boot which uses BLS thus there are no config files.
does it support encrypted /boot though?
Systemd-boot is my choice for any simple boot scenario. Love it. Agreed that grub is a mess
There's also systemd-boot, which seems to be getting more popular.
https://systemd.io/BOOT/
iSCSI is great and underated, I've been running iSCSI volumes for OrangePI boards which had only SD card support for years. Performance was much better on Gigabit Ethernet vs the SD card.
> UEFI fixes that to some extent, but it’s a pain to maintain the UEFI entries manually and change them every time the kernel updates.
… you don't have to update the UEFI entries every time the kernel updates. (I guess you might if you do like a kernel w/ CONFIG_EFI_STUB, and you place the new kernel under a different filename than what the UEFI boot entry point to then you might … but I was under the impression that that'd be kind of an unusual setup, and I thought most of us booting w/ EFI were doing so with Grub.)
Even if you do CONFIG_EFI_STUB, there should be a post-update hook to automatically call efibootmgr.
I have 2 entries, /efi/current.efi and /efi/old.efi... when I upgrade, I copy current to old, and copy my new kernel out of /boot as current and reboot.
or just copy the latest kernel to something like /vmlinux and /initramfs
Or use UKI and throw the current kernel to /efi/boot/bootx64.efi; there's plenty of solutions to sane bootloader/kernel management if you're willing to invest 15 minutes into the topic and not act like it's scary and complicated (it really is the opposite).
Grub2 is scary and complicated. Remove grub from the equation, and all the scary goes away.
grub is just a operating system. it is quite good when shit hits the fan
i never got it to work
Nice. I'm extra fond of ZFS backed network root filesystem, because it lets you put an OS on ZFS without needing to deal with ZFS support in that OS. (One of these days I want to try OpenBSD with its root on NFS on ZFS, either from Linux or FreeBSD.)
Does anyone have an opinion on iSCSI vs NBD?
Well, iSCSI is a standard, so chances are better that it's supported in a non-Linux OS, e.g. MS Windows. Years ago I booted a Windows (7, iirc) client that way, but gave up on it (too much hassle and performance limited by the network) when SSDs became cheap.
I don’t have direct experience, but when I looked into it my takeaway that NBD was unable to reliably deal with network interruptions as well as iscsi.
https://forums.gentoo.org/viewtopic.php?p=4895771&sid=f9b7ac...
https://github.com/NetworkBlockDevice/nbd/issues/93
Whether that’s the case with the latest version, I don’t know, but it’s something you might test if you choose to try it.
You might like https://smolbsd.org/
Well yes, I do like that:), but I don't see the connection to this thread?
I think you need NBD if you're going to use glustre, but I could also be thinking of ceph.
I've done a lot of headless/diskless stuff. I haven't done much for years, because my NAS only has gigabit Ethernet ports. I can cascade them and get four Gbps downstream, but it's still painful.
I have recently upgraded my house to 10Gbps Ethernet, with only one room still stuck at gigabit, and unfortunately, it's my main office. I'm working on getting the drop there now (literally, just taking a break here).
Even once I'm done, accessing an iSCSI drive over 10GbE will be 4-8 times slower than a local NVMe drive, but it will sure be a lot better than it was!
Ideally, I could run VMs on the NAS and have great performance, but that's another hardware upgrade...
Using a proper NIC (Chelsio) with their iSCSI accelerator will boost your iSCSI performance significantly. Another alternative is Mellanox with RDMA. You need CX4+ for optimal performance over TCP/IP, while the cheap CX3 is excellent with IPoIB. If you have a lot of packet drops and retransmissions, another option for boosting iSCSI performance is getting a network switch with a lot of memory for packet buffering. This helps with incast congestion. There are special switches with gigabytes of memory built for this.
NVMe-oF is the best protocol with least overhead for network drives, with a proper setup you lose only 10-20% latency compared to local disk even with Intel Optane. Throughput should be almost similar.
Really I wonder how this turns out to be diskless while you're clearly accessing a disk/drive over the network. Shouldn't we refer to this as network boot?
It's diskless from the point of view of the device being booted.
This could be an interesting setup for booting off a NAS like Synology or QNAP. I haven't really used iSCSI, it's intimidating how much prep this takes...
iSCSI seems intentionally obscure. One of the improvements I made to NBD was invent a simple, standardized URI format so that you can specify servers easily, eg:
https://github.com/NetworkBlockDevice/nbd/blob/master/doc/ur...NBD looks pretty nice! I've been eyeing it from afar for a while.
How well does it work in environments with noticeable network latency?
Requests and responses are pipelined so at least you're not serializing on round trips. However fundamentally if there is lots of latency, then you're going to be affected in some way. Usually we see problems where the OS accessing the remote drive times out which can sometimes be worked around by increasing timeouts, if you can work out how. (Latency is going to affect every block device protocol in about the same way)
You can actually see what happens quite easily if you've got an OS image handy. With a Fedora VM image:
("$uri" expands to the NBD URI of the nbdkit server which qemu can parse natively)Even that 1 second delay is painful since it turns out that booting is quite serialized. Edit: I turned down the latency to 50ms in the example which is a bit more realistic. Still painful.
The 'target' moves slow so once you learn it, it all stays relevant forever.
... And it's very, very fun.
Does it offer performance advantages over NFS root?
Dunno about performance vs NFS, but I've stuffed an unaware OS onto ZVOL-over-iSCSI using a NIC with option ROM.
I kind of expect the performance is worse, but one neat thing is that iscsi is a block device, so you could run e.g. disk crypto, volume management or whatever on it. Not to mention any FS. And you don't need to deal with NFS or RPC.
something worth mentioning here is that iSCSI is quite unhappy on congested networks or packet loss caused by incast traffic.
to make this actually work well, consider modifying your switches QoS settings to carve out a priority VLAN for iSCSI traffic
or a north-south/east-west architecture, so there's an entirely separate network just for iSCSI. Control plane vs data plane.
I used similar ipxe setup for robotic cluster - every robot booted from the same thing, then kubernetes managed the containe orchestration. it was fun.
NFS diskless is the more common approach I've used but this is very cool.
NFS diskless was easier for me to setup when I was doing it.
THe caveat was, you needed readonly root, so that meant freezing the OS, anything that needed changing was either stored in a ram disk (that you need to setup) or a per host nfs area (kinda like overlayfs, but not)
Why would you need a read-only root? Do you mean to share across multiple machines?
Yeah it makes things a bit easier to debug. Originally my system was designed to run on multiple machines at once.
If you needed to update the root dir, you chrooted into it and did the (yum) update.
When I tried root-on-nfs I had a lot of issues. The Redhat and Arch package managers don't seem to like it (presumably a sqlite thing?).
You can download the rootfs, extract it to a ramdisk, and just run in memory. This is fast for everything. Unfortunately, memory just got super expensive. Fortunately, Linux requires ~no memory to do many useful things.
You might find it worth upgrading to 10gbps if you continue to go down this road. The Mikrotik CRS-309 has served me well, and a couple Intel X520-DA2s. I believe those NICs can do iSCSI natively, and pass the session to the operating system with iBFT.
SFP28 might be cheap enough now too, I'm not sure...
Very neat. I've never seen the Debian installer using iscsi. No other installer either for that matter.
Looks like ZFS is only used to store the image on the server, though. I was expecting this to be more interesting because of that.
Pretty cool! You could also boot into an ephemeral minimal initrd that displays a selection menu instead of doing it in iPXE. That would grab the new kernel and initrd from the network and kexecs it without reboot.
> You could also boot into an ephemeral minimal initrd
Wouldn't that need a local disk?
No. PXE boots routinely load the kernel and initramfs directly into RAM with no local disk involved. The initramfs then mounts the real root FS over the network.
Then anaconda or whatover os installer picks up and installs the OS in a PXE install sequence when there is a local disk.
"I didn’t want to get into the hassle of repartitioning everything that the boot loader works with both Linux & Windows."
Hmmh? I haven't done so in years, but configuring multi-boot used to be considerably easier than disk-less operation.
It is relatively easy to configure. Just install Linux after windows, and Linux will generally automatically setup a boot-selection screen for you. The installer should detect windows and even shrink the partitions for you.
You can install a prettier looking boot selection menu like rEFInd, but the default works just as well, and I think the mainstream distros all setup secure boot too. On my pc it was very easy, on my (8yr old) laptop I had to add some secure boot keys and the bios was very confusing, using terms that didn’t seem to match what they should have been.
My setup has worked almost entirely flawlessly and survived updates from both OSes. Only issue being “larger” windows feature updates putting windows back as the first OS in the list, but that happens maybe once or twice a year? And it’s a quick bios change to fix the order.
I've never had this experience dual-booting, neither with UEFI nor Grub. I've been using Linux for nearly 15 years, 13 of those dual booting. Dozens of systems from laptops to desktops. Windows would always purge the boot entries. I'd have to manually fix booting, constantly. This happened with Ubuntu, Arch Linux and recently NixOS and through all the Windows editions till 11. I had to install Windows for a lan party recently and lo and behold on the second day NixOS is gone from the boot list and unbootable. Nothing of value is lost when it gets purged, but it's a damn annoying tax to pay just to be able to play video games.
Luckily gaming now works well enough that the only reason to use Windows was gone. Well, apart from some online games played during lan parties.
The Debian installer is less than optimal for repartitioning.
The Linux NTFS resizing code also has a tendency to trigger data corruption. Not really Linux' fault, but it's a good reason to do partitioning from inside of Windows, which can be a pain already.
Another issue I've run into is Windows creating a very small (~300MiB) EFI partition that barely fits the Windows bootloader, let alone a Linux bootloader and kernel. You can resize and recreate the partition of course, but reconfiguring Windows to use a different boot partition is a special kind of hell I try to avoid.
>Not really Linux' fault
If Linux corrupts someone files, it is 100% Linux's fault and is absolutely unacceptable.
SecureBoot is a PITA.
For Debian and most other distros, secure boot isn't a problem. Installers are all using a signed, trusted-by-default bootloader.
There are some exceptions (some hardware from Microsoft doesn't trust the third party certificate used, for instance, and Red Hat Enterprise has their own root of trust if you opt into that), but they're very rarely ever an issue.
This man is diskless
I know it was just a convenient pretext for a learning journey, but do not come away from this thinking llama.cpp needs to be compiled on Windows before use. The GitHib project has a cornucopia of pre-built artifacts to use.
https://github.com/ggml-org/llama.cpp/releases
what i want to play with is rdma and having a bcache block device with the remote as a backing and a small local nvme as a write-through cache
Does zfs support error correcting instead of just finding (already) broken files?
I have been waiting for such a feature for like 15 years now. Without it, zfs is just a fad and useless filesystem (all that complexity for NOTHING).
ext2 for the win! still
I don't know if I'm understanding your question, but ZFS actively corrects data on disk when it finds a checksum error [0]. Those checksum errors can be found when accessing that data, or doing a 'scrub' action that scans the whole volume to check integrity.
--
ZFS supports self healing, you do not have scrub, it will be corrected during a bad read as long you have a copy. Metadata has 2 copies by default for additional safety for a single disk.
I have to admit, I misread "Diskless Linux" initially ...
I would probably recommend to look into NVMe over TCP over iSCSI, especially for fast NVMe drives.