When I shut down the VPS, usually it is because I want to make a backup. Usually what I do is shut down the VPS internally and then also run shutdown from the control panel. The idea being that the control panel shutdown possibly is less likely to mess up the backup or even harm the VPS file system internally if the VPS already is shutdown internally and in the control panel when the backup starts. Maybe there really is no danger, so that I am taking unnecessary steps.
I haven't generally been using the control panel restart button. Just Shutdown and Boot.
Yesterday, there was, for the first time, a control panel shutdown failure (as usual, following an internal NetBSD shutdown). This failure manifested as stuck at 73% completion in the control panel log for a long time, about an hour. I did start a backup, after allowing what usually would be enough time (30 sec) for the control panel shutdown to complete, but without checking the shutdown log. The backup seemed to complete, and I received the email confirming the backup as completed. Later, when I checked the control panel log again, the control panel log reported the shutdown as FAILED.
Throughout the hour while the control panel shutdown seemed stuck at 73%, there seemed no way to reset the control panel. I tried reloading the page and logging out and logging in.
I imagined that installing Qemu-guest-agent might facilitate the control panel shutdown.
I never before saw the "power button pressed" message in this exact context. When I saw that message I assumed it was from Qemu-guest-agent. I didn't think about ACPI. There doesn't seem to be a "power button pressed" message in /var/log/messages:
netbsd# cd /var/log
netbsd# grep pressed messages
netbsd#
I am guessing that ACPI messages might not be included in /var/log/messages.
Maybe I should test the Shutdown from the control panel next? Any additional ideas for me, please?
Thanks for your question! I really, really appreciate your kind help!
netbsd# date
Sun Jul 5 22:02:07 UTC 2026
netbsd# # Trying control panel Shutdown
netbsd#
*** FINAL System shutdown message from [email protected] ***
System going down IMMEDIATELY
power button pressed # Don't recall seeing this message before installing Qemu-guest-agent
Connection to xxx.xxx.xxx.xxx closed by remote host.
Connection to xxx.xxx.xxx.xxx closed.
chronos@penguin:~/2026/netbsd$
@cmeerw said: I don't have qemu-guest-agent installed and I also get "power button pressed" when restarting from the control panel. I think that's using ACPI.
I will post more details later this evening or tomorrow about whether the compile completes.
Remember that this is an Offer thread? Thanks again to Linveo for sponsoring this VPS! If anyone is interested in sharing this VPS for free and helping me learn about NetBSD, that would be great!
NetBSD pkgsrc's `pkg_rolling-replace" usually just works. Yesterday, however, there seemed to be a stubborn pkgsrc dependency error:
netbsd# pkg_rolling-replace -rsuv
===> Building binary package for perl-5.44.0nb1
=> Creating binary package /usr/pkgsrc/packages/All/perl-5.44.0nb1.tgz
===> Replacing for perl-5.44.0nb1
===> Updating using binary package of perl-5.44.0nb1
/usr/bin/env /usr/sbin/pkg_add -K /usr/pkg/pkgdb -U -D /usr/pkgsrc/lang/perl5/wor
k/.packages/perl-5.44.0nb1.tgz
pkg_add: Can't open +CONTENTS of depending package p5-Net-Domain-TLD-1.75nb9
pkg_add: 1 package addition failed
*** Error code 1
Stop.
netbsd#
Here's why the dependency error was described above as "stubborn";
netbsd# date
Tue Aug 11 22:46:44 UTC 2026
netbsd# cd /usr/pkgsrc
netbsd# pkg_admin rebuild-tree
pkg_admin: Dependency p5-Net-Domain-TLD-[0-9]* of p5-Email-Valid-1.204nb3 unresolved
Done.
netbsd#
The dependency error eventually seemed to get fixed after some fun troubleshooting:
netbsd# date
Tue Aug 11 22:48:50 UTC 2026
netbsd# cd /usr/pkgsrc/net/p5-Net-Domain-TLD
netbsd# make install CLEANDEPENDS=yes
=> Bootstrap dependency digest>=20211023: found digest-20220214
===> Installing binary package of p5-Net-Domain-TLD-1.75nb10
netbsd# cd /usr/pkgsrc
netbsd# pkg_admin rebuild-tree
Done.
netbsd# cd /usr/pkgsrc/lang/perl5
pkg_add -K /usr/pkg/pkgdb -U work/.packages/perl-5.44.0nb1.tgz
netbsd# date
Tue Aug 11 22:54:35 UTC 2026
netbsd# pwd
/usr/pkgsrc/lang/perl5
netbsd# make clean
===> Cleaning for perl-5.44.0nb1
netbsd#
After perl had been installed and cleaned, pkg_rolling-replace finished successfully:
The clean finished, but I didn't post until now. Am I slower than a pkgsrc mega-clean?
netbsd# date
Wed Aug 12 01:00:28 UTC 2026
netbsd# pwd
/usr/pkgsrc
netbsd# time make clean CLEANDEPENDS=yes
[ . . . ]
===> Cleaning for zola-0.22.0
165193.93 real 114710.62 user 97052.25 sys
netbsd# echo $?
0
netbsd#
Now, following a backup and a clean and a few days relaxing, this LinveoNetBSD sweetie VPS is yet again updating and re-compiling.
On Sunday I saw a locale issue which hasn't happened previously:
netbsd# date
Sun Aug 23 17:11:42 UTC 2026
netbsd# pwd
/usr/pkgsrc
netbsd# /root/update-pkgsrc.sh
[ . . . ]
===> Checking for vulnerabilities in py313-sphinx-9.1.0
===> Overriding tools for py313-sphinx-9.1.0
===> Extracting for py313-sphinx-9.1.0
tar: Pathname can't be converted from UTF-8 to current locale
tar: Error exit delayed from previous errors
*** Error code 1
Stop.
I kind of cheated:
netbsd# date
Sun Aug 23 17:24:37 UTC 2026
netbsd# export LANG=en_US.UTF-8
netbsd# nohup time pkg_rolling-replace -rsuv
[ . . . ]
RR> Re-checking for unsafe installed packages (unsafe_depends_strict=YES)
RR> Packages to rebuild:
rr> MISMATCH_TODO=[]
rr> REBUILD_TODO=[]
rr> UNSAFE_TODO=[]
RR> No more packages to replace; done.
+ py313-qemu.qmp
+ py313-sphinx
+ py313-sphinx-rtd-theme
+ py313-sphinxcontrib-jquery
+ py313-urwid_readline
+ qemu-guest-agent
337.61 real 222.69 user 100.30 sys
I think these might be the excellent defaults from @cmeerw's excellent install image which Linveo kindly uses:
netbsd$ date
Wed Aug 26 00:50:12 UTC 2026
netbsd$ echo $LANG
netbsd$ locale
LANG=""
LC_CTYPE="C"
LC_COLLATE="C"
LC_TIME="C"
LC_NUMERIC="C"
LC_MONETARY="C"
LC_MESSAGES="C"
LC_ALL=""
netbsd$
Maybe there still might be some advantages to the C locale, even now in the UTF-8 Era? Or maybe C.UTF-8?
Other the the above error, the August 23 rebuild went great! This VPS now yet again has newly updated and self-compiled NetBSD-current kernel, userland, and pkgsrc! This VPS has come in especially handy during the recent days when I have wanted to compare its default NetBSD builds with another not-quite-default NetBSD install.
I wondered which were the "77 classical algorithms." Google Search AI ssemed to think all 77 algorithms were used in each of the three language frameworks:
Before this paper, existing verification benchmarks only tested individual languages on completely separate tasks, making it impossible to directly compare them. The AlgoVeri team solved this by taking the exact same 77 core algorithmic problems and writing identical functional contracts (specifications) for each one across all three reasoning ecosystems.
and that I could use ls:
If you clone the repository locally on your computer and run an ls command within their benchmark task folders (usually labeled by language or problem categories under a benchmarks/ or tasks/ directory), you will see all 77 unique algorithmic filenames laid out directory-by-directory.
I don't think Rust Is installed on my NetBSD VPS yet, but it's probably available, though I doubt I will go that far. For right now, I just want to list the algorithms.
Of course, I could simply ask Google Search AI to list the algorithms. Alternatively, I could look on Github's HTML interface, Or maybe use a contemporary AI coding interface which integrates with Github. But the old command line way is still fun for me.
Am I going to be able to find all 77 algorithms?
Thanks to Linveo for sponsoring the nice Ryzen VPS! If anyone wants to share it, please let me know.
checking whether vasprintf is declared... yes
--- configure-gas ---
checking for bison... /usr/src/../tools/bin/nbyacc
checking for flex... /usr/src/../tools/bin/nblex
checking lex output file root... lex.yy
--- configure-libiberty ---
checking whether snprintf is declared... yes
checking whether vsnprintf is declared... yes
checking whether calloc is declared... yes
checking whether getenv is declared... yes
checking whether getopt is declared... yes
checking whether malloc is declared... yes
checking whether realloc is declared... yes
--- configure-gas ---
checking lex library... -lfl
--- configure-libiberty ---
checking whether sbrk is declared... yes
checking whether strtol is declared... yes
checking whether strtoul is declared... yes
--- configure-gas ---
checking whether yytext is a pointer... yes
checking whether NLS is requested... no
--- configure-libiberty ---
checking whether strtoll is declared... yes
--- configure-gas ---
checking for msgfmt... /usr/bin/msgfmt
checking for gmsgfmt... /usr/bin/msgfmt
checking for xgettext... /usr/bin/xgettext
checking for msgmerge... /usr/bin/msgmerge
checking for ld used by GCC... ld
checking if the linker (ld) is GNU ld... yes
checking for shared library run path origin... done
--- configure-libiberty ---
checking whether strtoull is declared... yes
--- configure-gas ---
checking for CFPreferencesCopyAppValue... no
--- configure-libiberty ---
checking whether strverscmp is declared... no
--- configure-gas ---
checking for CFLocaleCopyPreferredLanguages... no
checking whether to use NLS... no
checking whether NLS is requested... no
checking whether to enable maintainer-specific portions of Makefiles... no
checking for memory.h... (cached) yes
--- configure-libiberty ---
checking whether strnlen is declared... yes
--- configure-gas ---
checking for sys/stat.h... (cached) yes
checking for sys/types.h... (cached) yes
checking for unistd.h... (cached) yes
--- configure-libiberty ---
checking whether canonicalize_file_name must be declared... yes
--- configure-gas ---
checking whether compiling a cross-assembler... yes
checking for strsignal... yes
checking for LC_MESSAGES... yes
checking for working assert macro... yes
checking for a known getopt prototype in unistd.h... yes
checking whether declaration is required for environ... yes
checking whether declaration is required for ffs... no
--- configure-libiberty ---
checking for SHA1 HW acceleration support... x86 SHA1
checking for stdlib.h... (cached) yes
checking for unistd.h... (cached) yes
checking for sys/param.h... (cached) yes
checking for getpagesize... (cached) yes
--- configure-gas ---
checking whether mempcpy is declared... yes
--- configure-libiberty ---
checking for working mmap... yes
--- configure-gas ---
checking whether stpcpy is declared... yes
checking for x86_64-unknown-netbsd11.99-pkg-config... no
checking for pkg-config... /usr/pkg/bin/pkg-config
checking pkg-config is at least version 0.9.0... yes
--- configure-libiberty ---
checking for working strncmp... yes
--- configure-gas ---
checking for struct stat.st_mtim.tv_sec in sys/stat.h... yes
--- configure-libiberty ---
configure: updating cache ./config.cache
configure: creating ./config.status
--- configure-gas ---
checking for struct stat.st_mtim.tv_nsec in sys/stat.h... yes
configure: updating cache ./config.cache
--- configure-libiberty ---
config.status: creating Makefile
--- configure-gas ---
checking that generated files are newer than configure... done
configure: creating ./config.status
--- configure-libiberty ---
config.status: creating config.h
config.status: executing default commands
--- configure-binutils ---
mkdir ./binutils
Configuring in ./binutils
configure: creating cache ./config.cache
Happy Sunday! Yet again rebuilding NetBSD-current on a kindly donated, fast Linveo Ryzen VPS!
Does anyone want to join, get a shell account on this excellent VPS?
Interesting day today! NetBSD pkgsrc builds seemed to break while pkg_rolling-replace was compiling Lynx.
There was an interesting entry in /var/log/messages:
Oct 4 21:50:19 netbsd /netbsd: [ 5.7789064] ffs_wapbl_replay_finish: /: unable to cleanup inode 3306969
A reboot into single user mode, and single user fsck seemed to fix the issue.
Normal reboot succeeded following the single user fsck. Shut down again for backup. Reboot after backup looked good:
Last login: Mon Oct 5 00:31:24 2026 from xxx.xxx.xxx.xxx
NetBSD 11.99.9 (GENERIC) #0: Sun Oct 4 19:42:02 UTC 2026
netbsd$ date; uptime
Mon Oct 5 00:57:29 UTC 2026
12:57AM up 1 min, 1 user, load averages: 0.00, 0.00, 0.00
netbsd$ ls -l /bin | head -n 5
total 3916
-r-xr-xr-x 2 root wheel 22048 Oct 4 19:28 [
-r-xr-xr-x 1 root wheel 22016 Oct 4 19:28 cat
lrwxr-xr-x 1 root wheel 11 Oct 4 19:28 chgrp -> /sbin/chown
-r-xr-xr-x 1 root wheel 22232 Oct 4 19:28 chio
netbsd$
For whatever it's worth, the context here involves the entire VPS file system recently having been restored from a backup as a way of checking the VPS' ability to backup and restore and as a way of reverting a recent extra project.
The crazy fun compile output shown above seems to have resulted from a restored file system fault injecting binary input into the terminal and the terminal attempting to render the binary input as text. Or, at least, that's a guess.
Does anybody have another idea about how the crazy terminal input might have happened?
Does anyone have an idea about how the filesystem issues mentioned above might have happened?
As you can see above, the filesystem issues seem to have been fixed by fsck, and the VPS is again running great! So,
does anybody want to join me on this fast, fun, free Ryzen self-compiled NetBSD-current VPS?
Thanks to Linveo for sponsoring!
There is nothing constantly running on this VPS. Its main job is following NetBSD-current and rebuilding everything once or twice per week to check that it can. Occasionally there might be an extra, short term project, a few of which have been posted above, but, except for the 3 or 4 hour compiles and subsequent reboots once or twice a week, the VPS is free almost all of the time.
As it says in the OP, I'm looking for a friend who wants to help me learn! Thanks everyone!
Comments
Couldn't you do that before already without qemu-guest-agent?
I don't have qemu-guest-agent installed and I also get "power button pressed" when restarting from the control panel. I think that's using ACPI.
@cmeerw
When I shut down the VPS, usually it is because I want to make a backup. Usually what I do is shut down the VPS internally and then also run shutdown from the control panel. The idea being that the control panel shutdown possibly is less likely to mess up the backup or even harm the VPS file system internally if the VPS already is shutdown internally and in the control panel when the backup starts. Maybe there really is no danger, so that I am taking unnecessary steps.
I haven't generally been using the control panel restart button. Just Shutdown and Boot.
Yesterday, there was, for the first time, a control panel shutdown failure (as usual, following an internal NetBSD shutdown). This failure manifested as stuck at 73% completion in the control panel log for a long time, about an hour. I did start a backup, after allowing what usually would be enough time (30 sec) for the control panel shutdown to complete, but without checking the shutdown log. The backup seemed to complete, and I received the email confirming the backup as completed. Later, when I checked the control panel log again, the control panel log reported the shutdown as FAILED.
Throughout the hour while the control panel shutdown seemed stuck at 73%, there seemed no way to reset the control panel. I tried reloading the page and logging out and logging in.
I imagined that installing Qemu-guest-agent might facilitate the control panel shutdown.
I never before saw the "power button pressed" message in this exact context. When I saw that message I assumed it was from Qemu-guest-agent. I didn't think about ACPI. There doesn't seem to be a "power button pressed" message in /var/log/messages:
I am guessing that ACPI messages might not be included in /var/log/messages.
Maybe I should test the Shutdown from the control panel next? Any additional ideas for me, please?
Thanks for your question! I really, really appreciate your kind help!
From control panel log:
Task Requested Duration Progress
Shutdown Sunday, July 5, 2026 3:02 PM 17 sec 100% Complete
Worked okay to Boot from control panel:
As usual it seems like @cmeerw is right!
The single log entry below seems consistent with the use of ACPI when I pushed the Restart button on the Control Panel.
Qemu-guest-agent doesn't seem to be doing much. Tagging @linveo in case he might be interested.
Since NetBSD 11 is out I decided to recompile my NetBSD Ryzen VPS at Linveo.
Today is August 3, longer than usual since the last compile:
The VPS is cooking quite well, as expected:
I will post more details later this evening or tomorrow about whether the compile completes.
Remember that this is an Offer thread? Thanks again to Linveo for sponsoring this VPS!
If anyone is interested in sharing this VPS for free and helping me learn about NetBSD, that would be great! 
Okay, compile of userland and kernel finished successfully! Yaaay! Thanks NetBSD!
New kernel and userland installed! Up to date again!
Thanks @linveo!
Nice VPS available to share for someone who wants to teach me more about NetBSD. 
NetBSD pkgsrc's `pkg_rolling-replace" usually just works. Yesterday, however, there seemed to be a stubborn pkgsrc dependency error:
Here's why the dependency error was described above as "stubborn";
The dependency error eventually seemed to get fixed after some fun troubleshooting:
After perl had been installed and cleaned, pkg_rolling-replace finished successfully:
Shutdown, backup, and reboot:
I decided to run
make cleaninside/usr/pkgsrc:Apparently there are about 22,000 packages? The clean inside /use/pkgsrc started yesterday and seems still running today:
As of just now, the clean is still running!
The clean finished, but I didn't post until now. Am I slower than a pkgsrc mega-clean?
Now, following a backup and a clean and a few days relaxing, this Linveo NetBSD sweetie VPS is yet again updating and re-compiling.
From the OP:
Thanks @linveo for sponsoring this excellent VPS!
On Sunday I saw a locale issue which hasn't happened previously:
I kind of cheated:
I think these might be the excellent defaults from @cmeerw's excellent install image which Linveo kindly uses:
Maybe there still might be some advantages to the C locale, even now in the UTF-8 Era? Or maybe C.UTF-8?
Other the the above error, the August 23 rebuild went great! This VPS now yet again has newly updated and self-compiled NetBSD-current kernel, userland, and pkgsrc! This VPS has come in especially handy during the recent days when I have wanted to compare its default NetBSD builds with another not-quite-default NetBSD install.
This free offer remains open!
Thanks Linveo! <3 Thanks @cmeerw! Thanks everyone! 
Following up on
I took a quick look at Mike's personal website, from which I followed the link to a paper he co-authored, AlgoVeri: An Aligned Benchmark for Verified Code Generation on Classical Algorithms.
For context, the AlgoVeri paper seems to use three verification language frameworks:
The AlgoVeri paper's Abstract says that AlgoVeri uses "77 classical algorithms in Dafny, Verus, and Lean."
Also, each of Dafny, Varus, and Lean "covers very different tasks, so the performance numbers are not directly comparable."
The purpose of AlgoVeri perhaps could be described as adjusting and adapting the verification frameworks to each other and to use by LLMs.
The final sentence of the Abstract linked to source code and data at https://github.com/haoyuzhao123/algoveri.
I wondered which were the "77 classical algorithms." Google Search AI ssemed to think all 77 algorithms were used in each of the three language frameworks:
and that I could use
ls:I decided to take a quick look:
I don't think Rust Is installed on my NetBSD VPS yet, but it's probably available, though I doubt I will go that far. For right now, I just want to list the algorithms.
Of course, I could simply ask Google Search AI to list the algorithms. Alternatively, I could look on Github's HTML interface, Or maybe use a contemporary AI coding interface which integrates with Github. But the old command line way is still fun for me.
Am I going to be able to find all 77 algorithms?
Thanks to Linveo for sponsoring the nice Ryzen VPS!
If anyone wants to share it, please let me know.
Happy Sunday! Yet again rebuilding NetBSD-current on a kindly donated, fast Linveo Ryzen VPS!
Does anyone want to join, get a shell account on this excellent VPS?
Best wishes!
Tom
Thanks! Best wishes!
Interesting day today! NetBSD pkgsrc builds seemed to break while pkg_rolling-replace was compiling Lynx.
There was an interesting entry in /var/log/messages:
Oct 4 21:50:19 netbsd /netbsd: [ 5.7789064] ffs_wapbl_replay_finish: /: unable to cleanup inode 3306969A reboot into single user mode, and single user fsck seemed to fix the issue.
Normal reboot succeeded following the single user fsck. Shut down again for backup. Reboot after backup looked good:
For whatever it's worth, the context here involves the entire VPS file system recently having been restored from a backup as a way of checking the VPS' ability to backup and restore and as a way of reverting a recent extra project.
The crazy fun compile output shown above seems to have resulted from a restored file system fault injecting binary input into the terminal and the terminal attempting to render the binary input as text. Or, at least, that's a guess.
Does anybody have another idea about how the crazy terminal input might have happened?
Does anyone have an idea about how the filesystem issues mentioned above might have happened?
As you can see above, the filesystem issues seem to have been fixed by fsck, and the VPS is again running great! So,
Thanks to Linveo for sponsoring!
There is nothing constantly running on this VPS. Its main job is following NetBSD-current and rebuilding everything once or twice per week to check that it can. Occasionally there might be an extra, short term project, a few of which have been posted above, but, except for the 3 or 4 hour compiles and subsequent reboots once or twice a week, the VPS is free almost all of the time.
As it says in the OP, I'm looking for a friend who wants to help me learn! Thanks everyone!