@Crab said: Now more interesting question would be that how small you can actually make them both while still maintaining the features you really need. That will require quite a bit of trial and error with the kernel compilation parameters but could be a fun exercise.
This is what I done for the Toshiba Click Mini but is much simpler when you target just one platform. (I can never understand why Wi-Fi drivers/modules are included in server builds.)
Than=compare;then=sequence:brought=bring;bought=buy:staffs=pile of sticks:informations/infos=no plural. It wisnae me! A big boy done it and ran away. || NVMe2G for life! until death (the end is nigh).
@Crab said: Now more interesting question would be that how small you can actually make them both while still maintaining the features you really need. That will require quite a bit of trial and error with the kernel compilation parameters but could be a fun exercise.
This is what I done for the Toshiba Click Mini but is much simpler when you target just one platform. (I can never understand why Wi-Fi drivers/modules are included in server builds.)
They weren't always. You used to get a basic kernel with hardware support and had to create your own config and build from there. WiFi really messed things up because hardware now isn't expected to have ethernet (or a cheap USB NIC for installation). It's possible to pare down kernels pretty extreme. I built a very base QEMU VirtIO only kernel that ran on a 64MB VPS with nearly 48MB free with full networking, thttpd and dropbear sshd. It wasn't great for much, but it did give me a place to build a compatible script for Hetrix before Virmach's monitor went insane and kept shutting it down after I got sick of fighting it and literally overwrote the entire 4GB of storage with NOPs so it'd chew through all of that before it died. Still had to beg them to remove it because I got sick of the incessant messages.
ftp on NetBSD finally gets sane defaults, showing huge improvements in file transfer speeds:
322734471 bytes retrieved in 00:14 (21.49 MiB/s)
vs. 322734471 bytes retrieved in 03:28 (1.47 MiB/s)
see PR #59865 for all the details - basically, I was wondering why sysupgrade was always so slow downloading new sets, and it turned out that ftp disabled auto-sizing of the receive socket buffer (limiting the TCP receive window size).
Improvements rolling in as shown by today's CVS output of my Linveo NetBSD-current VPS :
[ . . . ]
P usr.bin/ftp/cmds.c
P usr.bin/ftp/fetch.c
P usr.bin/ftp/ftp.1
P usr.bin/ftp/ftp.c
P usr.bin/ftp/ftp_var.h
P usr.bin/ftp/main.c
P usr.bin/ftp/util.c
P usr.bin/ftp/version.h
[ . . . ]
Thanks for the template! I have added this to our available templates as NetBSD 11 RC1 Minimum. Please give it a try and let me know how well it works.
Slow Servers IPv6-native VPSs hosted on OpenBSD's VMM in Spokane, WA, USA. (I racked these.) SporeStack Resold Vultr VPS/baremetal, DigitalOcean, and a whitelabeled brand in Europe. KYC-free, simple to launch. (I didn't rack these.)
Neither are dirt cheap!
Thanks for the template! I have added this to our available templates as NetBSD 11 RC1 Minimum. Please give it a try and let me know how well it works.
I'm curious . . . may I please ask why you switched from Free to Open?
Best wishes!
Tom
Hi Tom,
Thank you!
I like FreeBSD a lot. I took FreeBSD kernel internals from McKusick, himself!
FreeBSD has some weird memory management issues with Firefox. It will tend to climb and and climb from one tab to another. It did that to me for several releases.
One of the last releases I used, my laptop wouldn't boot with 8GB of memory. I think I could override it down to 4GB (though if installing physical dims, would only boot with 2GB), but then Firefox was practically unusable. It would boot with 8GB if I disabled SMP, but disabling SMP introduced even more serious bugs. I never saw any progress on fixing these bugs and wasn't really sure where to begin, myself.
It just ended up not being possible to use it as a daily driver, so I tried OpenBSD and haven't looked back (too often.)
OpenBSD is definitely a bit weird in some ways, but I think it's a better overall package. For desktop use, far better. For server use, it has a nice collection of utilities and security models that mold together very nicely. OpenBSD handles Firefox much better as well.
OpenBSD seems to be developed by more people who use it as a daily driver than FreeBSD. I do think ZFS and bhyve are very impressive, though.
I wish it was developed with git, but overall I'm glad I switched.
Slow Servers IPv6-native VPSs hosted on OpenBSD's VMM in Spokane, WA, USA. (I racked these.) SporeStack Resold Vultr VPS/baremetal, DigitalOcean, and a whitelabeled brand in Europe. KYC-free, simple to launch. (I didn't rack these.)
Neither are dirt cheap!
Next I got the NetBSD-current userland and kernel source code with CVS, patched /usr/src/sys/arch/x86/x86/cpu_topology.c as per @cmeerw's PR58693, built everything, installed the kernel, and rebooted.
The VPS came back up again, seemingly fine! Next up is to install the userland, check for any post install issues, and run etcupdate. We will see what happens next. . . .
The tech-kern email conversation linked in PR58693 is interesting! Haha, I have a lot more studying to do! But, apparently it might not be fully clear about the fundamental cause of the boot issue or which of two possible fixes should be applied.
Thanks to @cmeerw for incredible work isolating the issue and finding a fix! Thanks to @linveo for making @cmeerw's images available and for a great Intel VPS at only $2.85/month.
@cmeerw said:
Not sure where exactly it's stuck there, but it's likely something in VirtFusion still. Maybe you could have just retried?
Just to make sure I did a quick test with my VPS, and it's working as expected. I don't think it's the Intel issue at this point, as that would only affect the NetBSD boot after installation has finished.
Something is not quite right yet. I have installed (twice now) from what shows as "NetBSD 11 RC2 Minimum", but that still gives me 11.0 RC1. I have double-checked that the RC2 image was downloaded from my server, and that that image does indeed contain 11.0 RC2. Could you maybe check again that the correct image is used in VirtFusion?
NetBSD 11.0_RC1 (GENERIC) #0: Fri Feb 6 08:24:25 UTC 2026
Welcome to NetBSD!
This is a release candidate for NetBSD.
Bug reports: https://www.NetBSD.org/support/send-pr.html
Donations to the NetBSD Foundation: https://www.NetBSD.org/donations/
We recommend that you create a non-root account and use su(1) for root access.
x86# date
Sun Mar 8 20:14:08 UTC 2026
x86# uname -a
NetBSD x86.metalvps.com 11.0_RC1 NetBSD 11.0_RC1 (GENERIC) #0: Fri Feb 6 08:24:25 UTC 2026 [email protected]:/usr/src/sys/arch/amd64/compile/GENERIC amd64
x86#
The following are the https site, directory, user, and password that will be used. If
"user" is "ftp", then the password is not needed.
a: Host cdn.NetBSD.org
>b: Base directory pub/NetBSD/NetBSD-11.0_RC2
c: Binary set directory /amd64/binary/sets
d: Source set directory /source/sets
e: File extension .tar.xz
f: User ftp
g: Password
h: Proxy
i: Transfer directory /usr/INSTALL
j: Delete after install No
k: Configure network
l: Exit
x: Get Distribution
It seemed to work.
Last login: Sun Mar 8 20:13:50 2026 from xxx.xxx.xxx.xxx
NetBSD 11.0_RC2 (GENERIC) #0: Wed Mar 4 21:02:00 UTC 2026
Welcome to NetBSD!
This is a release candidate for NetBSD.
Bug reports: https://www.NetBSD.org/support/send-pr.html
Donations to the NetBSD Foundation: https://www.NetBSD.org/donations/
We recommend that you create a non-root account and use su(1) for root access.
x86# uname -a
NetBSD x86.metalvps.com 11.0_RC2 NetBSD 11.0_RC2 (GENERIC) #0: Wed Mar 4 21:02:00 UTC 2026 [email protected]:/usr/src/sys/arch/amd64/compile/GENERIC amd64
x86#
Something is not quite right yet. I have installed (twice now) from what shows as "NetBSD 11 RC2 Minimum", but that still gives me 11.0 RC1. I have double-checked that the RC2 image was downloaded from my server, and that that image does indeed contain 11.0 RC2. Could you maybe check again that the correct image is used in VirtFusion?
Sorry, I left the filename for the cache image as RC1, so VF did not try to redownload it for the node since it was cached. I have removed the cached qcow2 from the nodes and also changed the filename, so it will force fetch a new copy.
@Crab said: @cmeerw Excellent job! Do you know whether the Intel bug you fixed is finally mainlined into v11?
I am afraid there are no updates here - probably no one else has seen that.
Hold on, what's happening over here??
I wanted to try out your latest image without any high hopes after seeing these comments of yours, but lo and behold 11.0 RC2 booted right up! I haven't seen that happening on this awesome @linveo VPS before.
Just to verify this further I did install "NetBSD 10.1 Minimal for Intel (PR58693)" which worked fine as expected, but "NetBSD 10.1 Minimum" also worked! I started to scratch my head and I tried the all the other NetBSD images and they all worked even v9.4!
For a moment I was thinking that perhaps the hardware platform has changed, but the control panel is still showing the old Intel E5 and not Ryzen. Something in VirtFusion settings must've changed to enable this, but what it could've been... regardless it is amazing to see this finally working!
Tried installing OpenBSD 7.8 amd64 both by dd of miniroot.img to disk in rescue system and by using iso. Both borked -- no login prompt. Not on my fantastic free Linveo Ryzen 9 9950X which still is running NetBSD-current. Instead on my fantastic, newly bought Intel 6140.
However, using the i386 iso worked. Didn't try dd method on 6140.
I thought that OpenBSD amd64 might now work on the 6140, but apparently not. Anyone know why not? Thanks!
openbsd# date
Tue Mar 10 06:21:11 UTC 2026
openbsd# uname -am
OpenBSD openbsd.metalvps.com 7.8 GENERIC.MP#115 i386
openbsd# dmesg | grep Gold
cpu0: Intel(R) Xeon(R) Gold 6140 CPU @ 2.30GHz ("GenuineIntel" 686-class) 136 MHz, 06-55-04
# Why the 136 MHz speed? Apparently harmlless "CPU scaling?"
# See https://tech.openbsd.narkive.com/Mz75c34H/top-systat-and-hw-cpuspeed
cpu1: Intel(R) Xeon(R) Gold 6140 CPU @ 2.30GHz ("GenuineIntel" 686-class) 2.30 GHz, 06-55-04
openbsd# uptime
6:21AM up 32 mins, 1 user, load averages: 0.23, 0.10, 0.03
openbsd#
This line from dmesg on the i386 that works is missing from the above amd64 partial boot dmesg on the Intel 6140 which does not seem to work with OpenBSD amd64:
root on sd0a [ . . . ] swap on sd0b dump on sd0b
Maybe the problem could be that the 6140 with OpenBSD amd64 somehow fails to mount the root and swap?
If yes, maybe the root cause is something in the Virtfusion Qemu configuration?
Comments
This is what I done for the Toshiba Click Mini but is much simpler when you target just one platform. (I can never understand why Wi-Fi drivers/modules are included in server builds.)
Than=compare;then=sequence:brought=bring;bought=buy:staffs=pile of sticks:informations/infos=no plural.
It wisnae me! A big boy done it and ran away. || NVMe2G for life! until death (the end is nigh).
They weren't always. You used to get a basic kernel with hardware support and had to create your own config and build from there. WiFi really messed things up because hardware now isn't expected to have ethernet (or a cheap USB NIC for installation). It's possible to pare down kernels pretty extreme. I built a very base QEMU VirtIO only kernel that ran on a 64MB VPS with nearly 48MB free with full networking, thttpd and dropbear sshd. It wasn't great for much, but it did give me a place to build a compatible script for Hetrix before Virmach's monitor went insane and kept shutting it down after I got sick of fighting it and literally overwrote the entire 4GB of storage with NOPs so it'd chew through all of that before it died. Still had to beg them to remove it because I got sick of the incessant messages.
My pronouns are like/subscribe.
ftpon NetBSD finally gets sane defaults, showing huge improvements in file transfer speeds:322734471 bytes retrieved in 00:14 (21.49 MiB/s)vs.
322734471 bytes retrieved in 03:28 (1.47 MiB/s)see PR #59865 for all the details - basically, I was wondering why
sysupgradewas always so slow downloading new sets, and it turned out thatftpdisabled auto-sizing of the receive socket buffer (limiting the TCP receive window size).@cmeerw Another congrats!
I hope everyone gets the servers they want!
And talking about NetBSD, we now have an RC1 for NetBSD 11
@linveo I have now built an image for NetBSD 11.0 RC1
@cmeerw
My Linveo NetBSD-current VPS is now rebuilding with patches fixing the ftp bug you found!
From https://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=59865 :
Improvements rolling in as shown by today's CVS output of my Linveo NetBSD-current VPS :
Thanks @linveo!
Thanks @cmeerw! 
I hope everyone gets the servers they want!
@cmeerw Excellent job! Do you know whether the Intel bug you fixed is finally mainlined into v11?
I am afraid there are no updates here - probably no one else has seen that.
Thanks for the template! I have added this to our available templates as NetBSD 11 RC1 Minimum. Please give it a try and let me know how well it works.
linveo.com | Shared Hosting | KVM VPS | Dedicated Servers
My daily driver is OpenBSD. I was using FreeBSD for a number of years before that.
I host VPSs under OpenBSD's VMM. Currently only offering OpenBSD 7.8 and Debian 13.
I haven't tested FreeBSD or NetBSD yet to see how easy it would be to support them.
Slow Servers IPv6-native VPSs hosted on OpenBSD's VMM in Spokane, WA, USA. (I racked these.)
SporeStack Resold Vultr VPS/baremetal, DigitalOcean, and a whitelabeled brand in Europe. KYC-free, simple to launch. (I didn't rack these.)
Neither are dirt cheap!
Hi @slowservers!
Welcome to LES!
I'm curious . . . may I please ask why you switched from Free to Open?
Best wishes!
Tom
I hope everyone gets the servers they want!
Thanks, worked absolutely fine.
Hi Tom,
Thank you!
I like FreeBSD a lot. I took FreeBSD kernel internals from McKusick, himself!
FreeBSD has some weird memory management issues with Firefox. It will tend to climb and and climb from one tab to another. It did that to me for several releases.
One of the last releases I used, my laptop wouldn't boot with 8GB of memory. I think I could override it down to 4GB (though if installing physical dims, would only boot with 2GB), but then Firefox was practically unusable. It would boot with 8GB if I disabled SMP, but disabling SMP introduced even more serious bugs. I never saw any progress on fixing these bugs and wasn't really sure where to begin, myself.
It just ended up not being possible to use it as a daily driver, so I tried OpenBSD and haven't looked back (too often.)
OpenBSD is definitely a bit weird in some ways, but I think it's a better overall package. For desktop use, far better. For server use, it has a nice collection of utilities and security models that mold together very nicely. OpenBSD handles Firefox much better as well.
OpenBSD seems to be developed by more people who use it as a daily driver than FreeBSD. I do think ZFS and bhyve are very impressive, though.
I wish it was developed with git, but overall I'm glad I switched.
-Slow Servers
Slow Servers IPv6-native VPSs hosted on OpenBSD's VMM in Spokane, WA, USA. (I racked these.)
SporeStack Resold Vultr VPS/baremetal, DigitalOcean, and a whitelabeled brand in Europe. KYC-free, simple to launch. (I didn't rack these.)
Neither are dirt cheap!
@slowservers Thanks for the detailed reply! Now I know why you switched! 🤠
I hope everyone gets the servers they want!
Hello!
I bought a nifty Linveo Intel Xeon Gold VPS to compare with my nifty Linveo Ryzen VPS!
I tried reinstalling from Debian 13 to the NetBSD 11 RC1 Minimum image. The install process stuck for 30 minutes.
Happily, @cmeerw's NetBSD 10.1 Minimal for Intel (PR58693) booted just fine.
I grabbed the compiler via ftp by following @cmeerw's previous instructions.
Next I got the NetBSD-current userland and kernel source code with CVS, patched /usr/src/sys/arch/x86/x86/cpu_topology.c as per @cmeerw's PR58693, built everything, installed the kernel, and rebooted.
The VPS came back up again, seemingly fine! Next up is to install the userland, check for any post install issues, and run etcupdate. We will see what happens next. . . .
The tech-kern email conversation linked in PR58693 is interesting! Haha, I have a lot more studying to do! But, apparently it might not be fully clear about the fundamental cause of the boot issue or which of two possible fixes should be applied.
Thanks to @cmeerw for incredible work isolating the issue and finding a fix!
Thanks to @linveo for making @cmeerw's images available and for a great Intel VPS at only $2.85/month. 
I hope everyone gets the servers they want!
Might have managed to get through postinstall checks and etcupdate plus another reboot.
I hope everyone gets the servers they want!
Not sure where exactly it's stuck there, but it's likely something in VirtFusion still. Maybe you could have just retried?
Just to make sure I did a quick test with my VPS, and it's working as expected. I don't think it's the Intel issue at this point, as that would only affect the NetBSD boot after installation has finished.
NetBSD 11.0 RC2 has appeared, so there is now a NetBSD 11.0 RC2 image @linveo
Jails for NetBSD looks like a really interesting project - something I'll likely have a closer look at soon.
Thank you, I have updated the NetBSD 11 template.
linveo.com | Shared Hosting | KVM VPS | Dedicated Servers
Something is not quite right yet. I have installed (twice now) from what shows as "NetBSD 11 RC2 Minimum", but that still gives me 11.0 RC1. I have double-checked that the RC2 image was downloaded from my server, and that that image does indeed contain 11.0 RC2. Could you maybe check again that the correct image is used in VirtFusion?
Of course I should have mentioned that I did try twice.
Perhaps now it might be best to wait briefly for the 11.0 RC2 image and try again. That's fine. All good!
I appreciate @cmeerw making the images and @linveo making the images available!
I hope everyone gets the servers they want!
Hello!
Tried the NetBSD 11.0 RC2 image from the Linveo control panel using my Intel VPS.
The image installed and booted quickly, but, as @cmeerw mentioned above, seemed to describe itself as RC1 instead of RC2.
Terminal output from the boot and screenshot from the install are posted below.
Thanks @linveo!
Thanks @cmeerw! 
Tom
I hope everyone gets the servers they want!
Tried using sysinst to upgrade to RC2 minimal.
It seemed to work.
I hope everyone gets the servers they want!
Sorry, I left the filename for the cache image as RC1, so VF did not try to redownload it for the node since it was cached. I have removed the cached qcow2 from the nodes and also changed the filename, so it will force fetch a new copy.
linveo.com | Shared Hosting | KVM VPS | Dedicated Servers
Hold on, what's happening over here??
I wanted to try out your latest image without any high hopes after seeing these comments of yours, but lo and behold 11.0 RC2 booted right up! I haven't seen that happening on this awesome @linveo VPS before.
Just to verify this further I did install "NetBSD 10.1 Minimal for Intel (PR58693)" which worked fine as expected, but "NetBSD 10.1 Minimum" also worked! I started to scratch my head and I tried the all the other NetBSD images and they all worked even v9.4!
For a moment I was thinking that perhaps the hardware platform has changed, but the control panel is still showing the old Intel E5 and not Ryzen. Something in VirtFusion settings must've changed to enable this, but what it could've been... regardless it is amazing to see this finally working!
Tried installing OpenBSD 7.8 amd64 both by dd of miniroot.img to disk in rescue system and by using iso. Both borked -- no login prompt. Not on my fantastic free Linveo Ryzen 9 9950X
which still is running NetBSD-current. Instead on my fantastic, newly bought Intel 6140. 
However, using the i386 iso worked. Didn't try dd method on 6140.
I thought that OpenBSD amd64 might now work on the 6140, but apparently not. Anyone know why not? Thanks!
I hope everyone gets the servers they want!
This line from dmesg on the i386 that works is missing from the above amd64 partial boot dmesg on the Intel 6140 which does not seem to work with OpenBSD amd64:
root on sd0a [ . . . ] swap on sd0b dump on sd0bMaybe the problem could be that the 6140 with OpenBSD amd64 somehow fails to mount the root and swap?
If yes, maybe the root cause is something in the Virtfusion Qemu configuration?
Thanks!
I hope everyone gets the servers they want!