Made another update to the NetBSD 10.0 image to use more useful parameters during filesystem creation (only realised over the weekend that the scripts I based my image creation on used some rather questionable filesystem settings that result in an almost insane number of inodes after resizing).
As a bonus I then tried to also make an image for NetBSD 9.4. This required an additional reboot when resizing the partition/filesystem, but the end result should be very similar to the 10.0 image. (Note: 9.4 doesn't have any SSL certificates, so I am setting the default PKG_PATH to the http URL instead of https).
Hopefully, these are now stable enough and I can wait for NetBSD 10.1 for the next update.
BTW, for those interested in some of the technical details, I noticed that you can end up with a screwed up filesystem that might silently eat some of your data if you are not careful with all the resize stuff (but without ever seeing any warning from the tools) - that's PR #58723. A cosmetic bug in the df output I noticed has already been fixed on HEAD - PR #58718
@cmeerw said: some rather questionable filesystem settings that result in an almost insane number of inodes after resizing
With your two new images and with the NetBSD PRs that you mentioned, it looks like both you and NetBSD itself successfully have moved beyond the inodes issue.
Normally I would try to read up on the issue before asking a question, but, maybe, if you don't mind, you could please tell me whether you think the the inodes issue might exist on my install from your older image, and how to run a quick check.
@Not_Oles said: Normally I would try to read up on the issue before asking a question, but, maybe, if you don't mind, you could please tell me whether you think the the inodes issue might exist on my install from your older image, and how to run a quick check.
Yes, your image will be affected, a quick way to tell is to run df -mi /
In the iAvail column it will likely show something like 20 million inodes (instead of maybe 3 million), but in the 1M-blocks column you'll likely only see around 20000 blocks (instead of maybe 24300), e.g.
I still have to study up, but here is the output of the df -ml / command you suggested showing results as you predicted:
linveo# date
Mon Oct 7 23:27:04 UTC 2024
linveo# df -mi /
Filesystem 1M-blocks Used Avail %Cap iUsed iAvail %iCap Mounted on
/dev/dk2 20231 1920 17299 9% 307359 19440415 1% /
linveo#
I still need to try the rebuild from source. When I get a chance, I will do that and post about what happens. Trying the compile will be a lot of fun even if it doesn't work for any reason, including maybe the filesystem.
@cmeerw said:
Made another update to the NetBSD 10.0 image to use more useful parameters during filesystem creation (only realised over the weekend that the scripts I based my image creation on used some rather questionable filesystem settings that result in an almost insane number of inodes after resizing).
As a bonus I then tried to also make an image for NetBSD 9.4. This required an additional reboot when resizing the partition/filesystem, but the end result should be very similar to the 10.0 image. (Note: 9.4 doesn't have any SSL certificates, so I am setting the default PKG_PATH to the http URL instead of https).
Hopefully, these are now stable enough and I can wait for NetBSD 10.1 for the next update.
BTW, for those interested in some of the technical details, I noticed that you can end up with a screwed up filesystem that might silently eat some of your data if you are not careful with all the resize stuff (but without ever seeing any warning from the tools) - that's PR #58723. A cosmetic bug in the df output I noticed has already been fixed on HEAD - PR #58718
I have loaded up the latest NetBSD 10 template and added a new one for 9.4. Thanks again!
If you have any instructions I can follow to create qcow2 templates, I can try as well. I was given some documentation from VF to create my own, but not sure how well it will work.
@linveo said: I have loaded up the latest NetBSD 10 template and added a new one for 9.4. Thanks again!
Thanks so much. The NetBSD 10 image works as expected.
The NetBSD 9.4 image works, except that it doesn't see any configuration data, so no network or SSH keys are configured. I am not sure if this an issue with NetBSD 9.4 or if that config data is missing from the VirtFusion side (on my local tests, NetBSD 9.4 seemed to see the same drives as NetBSD 10).
If you have any instructions I can follow to create qcow2 templates, I can try as well. I was given some documentation from VF to create my own, but not sure how well it will work.
I have made my scripts available here - this script just needs to be run as root on a NetBSD 10 host (it should work without any additional dependencies on a minimal NetBSD 10 installation)
./build.sh -e -v 10.0
will create a netbsd-10.0.raw file.
That can then be converted with qemu-img -c -O qcow2 netbsd-10.0.raw netbsd-10.0.qcow2 to a qcow2 image (qemu-img is not installed on the minimal NetBSD 10 image, but you could do that step on Linux).
(for NetBSD 9.4 I use ./build.sh -v 9.4 without the -e (EFI boot) option as that didn't seem to work, so the 9.4 image is BIOS boot only - which is what VirtFusion is using anyway).
I followed the above linked steps in the Guide, Now I have my unprivileged user, tom, who successfully can connect via ssh and use su.
I hope that my posting reference links and steps I followed might encourage others who haven't yet tried building their entire NetBSD system to actually try it.
Next up will be making the build directories and downloading and updating the sources.
Thanks to @linveo for providing the fun test VM for free and to the BSD guys here for watching my back and catching my mistakes.
Switching to tom, checking the environment variables, then switching to tom with su and double checking the environment variables. Skipping the X Window System sources for now.
linveo# su tom
linveo$ whoami
tom
linveo$ echo $CVSROOT
[email protected]:/cvsroot
linveo$ echo $CVS_RSH
ssh
linveo$ whoami
tom
linveo$
Haha, since it wasn't running inside tmux, I borked the above checkout by shutting my Chromebook when I wanted to go to sleep. Probably all that was necessary was to rerun the checkout command, which I did, and the results seemed okay.
For the sake of completeness, I moved src to src-old, made a new src directory owned by tom, and reran the checkout from the beginning, inside tmux. A few lines of terminal output from new checkout are shown below. It looks like the full checkout took around 3.5 hours.
I think that the final checkout "Updating" lines which mention X are not a problem. If I remember right, NetBSD together with pkgsrc offer X.org and XFree86.org versions of X. There also seem to be some X libraries in the main NetBSD sources.
Next up might be to see whether the newly checked out sources build successfully. In the newly checked out README.md file, it says:
Building
You can cross-build NetBSD from most UNIX-like operating systems.
To build for amd64 (x86_64), in the src directory:
./build.sh -U -u -j4 -m amd64 -O ~/obj release
I think I'd have to use -j 2 on my two vCore VPS. I'm not sure if ~/obj needs to exist. I should skim build.sh.
Wondering about whether the VPS has enough memory. Wondering about the effect of the inode issue mentioned above. Wondering how long the build will take, if it completes.
Decided to try a compile, because, why not? Following nia's procedure in the above linked page, I extracted the tar file into /home/tom and made a /home/tom/obj directory.
Ran this build.sh command inside tmux: time ./build.sh -U -u -j 2 -m amd64 -O ~/obj release
Since starting the build, it's been maybe half an hour while I have been writing this post, and the build seems to be still running.
@linveo Linveo control panel shows CPU utilization ranging from about 88% to about 102.3%.
linveo# su - tom
linveo$ pwd
/home/tom
linveo$ whoami
tom
linveo$ mkdir obj
linveo$ tar xf /usr/src-new-checkout-20241014.tar
linveo$ ls -l
total 3
drwxr-xr-x 2 tom users 512 Oct 14 21:11 obj
-rw------- 1 tom users 14 Oct 11 02:24 password
drwxr-xr-x 25 tom users 512 Oct 14 09:15 src
linveo$ cd src
linveo$ ls
BUILDING UPDATING crypto external regress tests
CVS bin dist games rescue tools
Makefile build.sh distrib include sbin usr.bin
Makefile.inc common doc lib share usr.sbin
README.md compat etc libexec sys
linveo$ ls -l
total 373
-rw-r--r-- 1 tom users 42791 Apr 26 17:38 BUILDING
drwxr-xr-x 2 tom users 512 Oct 14 09:15 CVS
-rw-r--r-- 1 tom users 16367 Sep 8 2023 Makefile
-rw-r--r-- 1 tom users 355 May 2 2018 Makefile.inc
-rw-r--r-- 1 tom users 1748 Sep 5 2021 README.md
-rw-r--r-- 1 tom users 20004 Sep 26 20:08 UPDATING
drwxr-xr-x 37 tom users 1024 Oct 14 09:15 bin
-rwxr-xr-x 1 tom users 73189 Jul 23 20:46 build.sh
drwxr-xr-x 6 tom users 512 Oct 14 05:45 common
drwxr-xr-x 10 tom users 512 Oct 14 09:15 compat
drwxr-xr-x 5 tom users 512 Oct 14 05:47 crypto
drwxr-xr-x 4 tom users 512 Oct 14 09:14 dist
drwxr-xr-x 61 tom users 1536 Oct 14 09:14 distrib
drwxr-xr-x 4 tom users 512 Oct 14 06:05 doc
drwxr-xr-x 77 tom users 3072 Oct 14 09:13 etc
drwxr-xr-x 23 tom users 512 Oct 14 09:02 external
drwxr-xr-x 54 tom users 1024 Oct 14 09:01 games
drwxr-xr-x 8 tom users 2048 Oct 14 08:59 include
drwxr-xr-x 67 tom users 1536 Oct 14 08:59 lib
drwxr-xr-x 29 tom users 1024 Oct 14 08:59 libexec
drwxr-xr-x 6 tom users 512 Oct 14 08:59 regress
drwxr-xr-x 3 tom users 512 Oct 14 08:14 rescue
drwxr-xr-x 116 tom users 2560 Oct 14 08:59 sbin
drwxr-xr-x 20 tom users 512 Oct 14 08:59 share
drwxr-xr-x 37 tom users 1024 Oct 14 08:59 sys
drwxr-xr-x 22 tom users 512 Oct 14 08:57 tests
drwxr-xr-x 120 tom users 2560 Oct 14 08:57 tools
drwxr-xr-x 260 tom users 4608 Oct 14 08:57 usr.bin
drwxr-xr-x 171 tom users 3072 Oct 14 08:57 usr.sbin
linveo$ date
Tue Oct 15 00:32:09 UTC 2024
linveo$ time ./build.sh -U -u -j 2 -m amd64 -O ~/obj release
===> build.sh command: ./build.sh -U -u -j 2 -m amd64 -O /home/tom/obj release
===> build.sh started: Tue Oct 15 00:32:27 UTC 2024
===> NetBSD version: 10.99.12
===> MACHINE: amd64
===> MACHINE_ARCH: x86_64
===> Build platform: NetBSD 10.99.12 amd64
===> HOST_SH: /bin/sh
===> No $TOOLDIR/bin/nbmake, needs building.
===> Bootstrapping nbmake
[ . . . ]
I just grabbed the following output lines as they flashed by:
make distribution started at: Tue Oct 15 00:32:39 UTC 2024
make distribution finished at: Tue Oct 15 04:45:35 UTC 2024
[ . . . ]
/home/tom/obj/tooldir.NetBSD-10.99.12-amd64/bin/nbmake -C /home/tom/obj/sys/arch/amd64/compile/GENERIC depend && /home/tom/obj/tooldir.NetBSD-10.99.12-amd64/bin/nbmake -C /home/tom/obj/sys/arch/amd64/compile/GENERIC && /home/tom/obj/tooldir.NetBSD-10.99.12-amd64/bin/nbmake -C /home/tom/obj/sys/arch/amd64/compile/GENERIC debuginstall
Time for sleep soon.
I wonder if we are going to run out if disk space.
linveo# df -h .
Filesystem Size Used Avail %Cap Mounted on
/dev/dk2 20G 15G 4.0G 78% /
linveo#
[ 874397.7878684] /: write failed, file system is full
[ 874397.7878684] /: write failed, file system is full
[ 874397.7878684] /: write failed, file system is full
[ 874397.7878684] /: write failed, file system is full
[ 874397.7878684] /: write failed, file system is full
[ 874397.7878684] /: write failed, file system is full
[ 874397.7878684] /: write failed, file system is full
[ 874397.7878684] /: write failed, file system is full
you have mail
linveo$ df -h .
Filesystem Size Used Avail %Cap Mounted on
/dev/dk2 20G 19G 540K 99% /
linveo$
@Not_Oles said: Anybody have a genius idea? Or two or three?
Looks like you didn't start with the new NetBSD template, which would have given you maybe 4 GB more space on the file system (not sure if that would have been enough space).
Just out of interest, how many inodes are you using? df -hi /
Filesystem Size Used Avail %Cap iUsed iAvail %iCap Mounted on
/dev/dk2 24G 1.3G 21G 5% 16853 3100457 0% /
One of the reasons why I went ahead with the old image was that I wondered whether the large number of inodes would cause any kind of a problem. Looks like, until we ran out of space, everything went fine despite the large number of inodes.
When things started looking like space was going to get tight, I deleted pkgsrc and also my previously mentioned initial src checkout which had been moved to src-old.
If @linveo wants to bump my VM's disk size, I am happy to rebuild NetBSD, this time beginning by reinstalling with @cmeerw's latest image. @linveo Rather than increasing disk size, if it's easier to wipe the existing VM and provision a replacement, that's fine. If increasing my VM's disk size causes any issues, it's also fine to leave it as is. I really like this VM! It has a fast processor, and it has only 23 ms ping from my current location in Sonora.
Google ""NetBSD" inode max 3100457" says "Your search did not match any documents".
@cmeerw If you do not mind, may I please ask, what is the best way to find out where, when, and why the inode max was increased? Why was 3100457 selected?
Google ""NetBSD" inode max 3100457" says "Your search did not match any documents".
@cmeerw If you do not mind, may I please ask, what is the best way to find out where, when, and why the inode max was increased? Why was 3100457 selected?
Sorry... that's just the (max) number of inodes you will get when re-installing NetBSD from my template on that particular VM.
-i bytes-per-inode
This specifies the density of inodes in the file system. If
fewer inodes are desired, a larger number should be used; to
create more inodes a smaller number should be given. The
default is to create an inode for every (4 * frag-size) bytes
of data space:
Note that when creating the NetBSD image I am creating a 512 MB root filesystem, but as I am expecting the file system to be resized during first boot, I am setting "-b 16384 -f 2048" (so we get more reasonable defaults for the final filesystem size).
So the initial 512 MB file system will get approximately 65 thousand inodes, and when the filesystem gets resized to 24.5 GB, the inode density stays the same and you'll end up with roughly 3 million inodes.
BTW, the actual maximum number of inodes is 3117310 (3100457 was the number of available inodes on my system), but I think df only shows that number when using df -G / (and the values are currently swapped around on 10.0).
@Not_Oles said:
If @linveo wants to bump my VM's disk size, I am happy to rebuild NetBSD, this time beginning by reinstalling with @cmeerw's latest image. @linveo Rather than increasing disk size, if it's easier to wipe the existing VM and provision a replacement, that's fine. If increasing my VM's disk size causes any issues, it's also fine to leave it as is. I really like this VM! It has a fast processor, and it has only 23 ms ping from my current location in Sonora.
You got it! I upped your VM to 50GB, just need a reboot. You might need to manually increase the FS too if cloud-init is not able to interact.
@linveo said: You got it! I upped your VM to 50GB, just need a reboot.
@linveo The pending changes show up on the Control Panel at my end. Before I do anything, though, I just want to double check with @cmeerw because he's been working on the NetBSD images. If there is anything that @cmeerw might want to do or check or ask me to do or check, I want to make sure that whatever @cmeerw wants is done.
My inclination might be to reinstall from @cmeerw's latest NetBSD image. I might just do Debian for a moment first, simply to exercise the system and to double check that everything works with an OS that VF might more likely be expecting. But I could proceed differently if @cmeerw or anybody else wants. Thanks!
Comments
Looks like the info on how to build NetBSD from source is in Part VI of the NetBSD Guide.
I hope everyone gets the servers they want!
OpenBSD ftw!
Made another update to the NetBSD 10.0 image to use more useful parameters during filesystem creation (only realised over the weekend that the scripts I based my image creation on used some rather questionable filesystem settings that result in an almost insane number of inodes after resizing).
As a bonus I then tried to also make an image for NetBSD 9.4. This required an additional reboot when resizing the partition/filesystem, but the end result should be very similar to the 10.0 image. (Note: 9.4 doesn't have any SSL certificates, so I am setting the default
PKG_PATHto thehttpURL instead ofhttps).Hopefully, these are now stable enough and I can wait for NetBSD 10.1 for the next update.
@linveo could you update the NetBSD 10.0 image and add the NetBSD 9.4 image please?
BTW, for those interested in some of the technical details, I noticed that you can end up with a screwed up filesystem that might silently eat some of your data if you are not careful with all the resize stuff (but without ever seeing any warning from the tools) - that's PR #58723. A cosmetic bug in the
dfoutput I noticed has already been fixed on HEAD - PR #58718With your two new images and with the NetBSD PRs that you mentioned, it looks like both you and NetBSD itself successfully have moved beyond the inodes issue.
Normally I would try to read up on the issue before asking a question, but, maybe, if you don't mind, you could please tell me whether you think the the inodes issue might exist on my install from your older image, and how to run a quick check.
Thanks @cmeerw!
I hope everyone gets the servers they want!
Yes, your image will be affected, a quick way to tell is to run
df -mi /In the
iAvailcolumn it will likely show something like 20 million inodes (instead of maybe 3 million), but in the1M-blockscolumn you'll likely only see around 20000 blocks (instead of maybe 24300), e.g.@cmeerw
I still have to study up, but here is the output of the
df -ml /command you suggested showing results as you predicted:I still need to try the rebuild from source. When I get a chance, I will do that and post about what happens. Trying the compile will be a lot of fun even if it doesn't work for any reason, including maybe the filesystem.
As always, thanks for your kind help!
I hope everyone gets the servers they want!
OpenBSD 7.6 has been released.
I wish they release drivers for Raspberry Pi Zero 2 W - or atleast FreeBSD.
I have loaded up the latest NetBSD 10 template and added a new one for 9.4. Thanks again!
If you have any instructions I can follow to create qcow2 templates, I can try as well. I was given some documentation from VF to create my own, but not sure how well it will work.
linveo.com | Shared Hosting | KVM VPS | Dedicated Servers
This is what you get with a new NetBSD 10 install:
Thanks so much. The NetBSD 10 image works as expected.
The NetBSD 9.4 image works, except that it doesn't see any configuration data, so no network or SSH keys are configured. I am not sure if this an issue with NetBSD 9.4 or if that config data is missing from the VirtFusion side (on my local tests, NetBSD 9.4 seemed to see the same drives as NetBSD 10).
I have made my scripts available here - this script just needs to be run as root on a NetBSD 10 host (it should work without any additional dependencies on a minimal NetBSD 10 installation)
will create a
netbsd-10.0.rawfile.That can then be converted with
qemu-img -c -O qcow2 netbsd-10.0.raw netbsd-10.0.qcow2to a qcow2 image (qemu-imgis not installed on the minimal NetBSD 10 image, but you could do that step on Linux).(for NetBSD 9.4 I use
./build.sh -v 9.4without the-e(EFI boot) option as that didn't seem to work, so the 9.4 image is BIOS boot only - which is what VirtFusion is using anyway).Finally getting around to trying this.
The above linked guide says to make an unprivileged user account to use for building.
Documentation for how to make a user account is in Part III, Section 5.6 of the NetBSD Guide.
I followed the above linked steps in the Guide, Now I have my unprivileged user, tom, who successfully can connect via
sshand usesu.I hope that my posting reference links and steps I followed might encourage others who haven't yet tried building their entire NetBSD system to actually try it.
Next up will be making the build directories and downloading and updating the sources.
Thanks to @linveo for providing the fun test VM for free and to the BSD guys here for watching my back and catching my mistakes.
I hope everyone gets the servers they want!
Now we have the directory for the sources. Reference: https://www.netbsd.org/docs/guide/en/chap-fetch.html#chap-fetch-dirs
linveo# cd /usr
linveo# pwd
/usr
linveo# ls /home
tom
linveo# mkdir /usr/src
linveo# chown tom /usr/src
linveo# ls -l | grep src
drwxrwxr-x 53 600 125 1024 Sep 28 01:01 pkgsrc
-rw-r--r-- 1 root wheel 84995861 Sep 28 01:15 pkgsrc.tar.gz
-rw-r--r-- 1 root wheel 64 Sep 28 01:15 pkgsrc.tar.gz.SHA1
drwxr-xr-x 2 tom wheel 512 Oct 14 00:11 src
linveo#
I hope everyone gets the servers they want!
Setting the CVS environment variables as instructed in https://www.netbsd.org/docs/guide/en/chap-fetch.html#chap-fetch-cvs
linveo# pwd
/usr
linveo# export CVSROOT="[email protected]:/cvsroot"
linveo# echo $CVSROOT
[email protected]:/cvsroot
linveo# export CVS_RSH="ssh"
linveo# echo $CVS_RSH
ssh
linveo#
I hope everyone gets the servers they want!
We have options to download tarballs of the sources, but, even though it is slower, let's try CVS as specified at https://www.netbsd.org/docs/guide/en/chap-fetch.html#chap-fetch-cvs-netbsd-current
Switching to
tom, checking the environment variables, then switching totomwithsuand double checking the environment variables. Skipping the X Window System sources for now.Now checking out the sources with CVS as per https://www.netbsd.org/docs/guide/en/chap-fetch.html#chap-fetch-cvs-netbsd-current
I hope everyone gets the servers they want!
Haha, since it wasn't running inside
tmux, I borked the above checkout by shutting my Chromebook when I wanted to go to sleep. Probably all that was necessary was to rerun the checkout command, which I did, and the results seemed okay.For the sake of completeness, I moved
srctosrc-old, made a newsrcdirectory owned bytom, and reran the checkout from the beginning, insidetmux. A few lines of terminal output from new checkout are shown below. It looks like the full checkout took around 3.5 hours.I think that the final checkout "Updating" lines which mention X are not a problem. If I remember right, NetBSD together with pkgsrc offer X.org and XFree86.org versions of X. There also seem to be some X libraries in the main NetBSD sources.
Next up might be to see whether the newly checked out sources build successfully. In the newly checked out README.md file, it says:
./build.sh -U -u -j4 -m amd64 -O ~/obj releaseI think I'd have to use
-j 2on my two vCore VPS. I'm not sure if~/objneeds to exist. I should skimbuild.sh.I hope everyone gets the servers they want!
I made a tar archive of the newly checked out
/srcso it's easy to revert if / when the build fails or for any other reason.I hope everyone gets the servers they want!
Building NetBSD-current on a Linveo 2 vCore 2 GB RAM VPS
Reference: https://www.netbsd.org/docs/guide/en/chap-build.html
Wondering about whether the VPS has enough memory. Wondering about the effect of the inode issue mentioned above. Wondering how long the build will take, if it completes.
Decided to try a compile, because, why not? Following nia's procedure in the above linked page, I extracted the tar file into /home/tom and made a /home/tom/obj directory.
Ran this build.sh command inside tmux:
time ./build.sh -U -u -j 2 -m amd64 -O ~/obj releaseSince starting the build, it's been maybe half an hour while I have been writing this post, and the build seems to be still running.
@linveo Linveo control panel shows CPU utilization ranging from about 88% to about 102.3%.
I hope everyone gets the servers they want!
Here's the current output of
topin case anybody might be interested.I hope everyone gets the servers they want!
The build still seems like it's going strongly.
I just grabbed the following output lines as they flashed by:
Time for sleep soon.
I wonder if we are going to run out if disk space.
I hope everyone gets the servers they want!
Not out of disk space yet, but it's getting close.
I hope everyone gets the servers they want!
Uh . . . File system is full.
Sleep now for me. . . .
Anybody have a genius idea? Or two or three?
I hope everyone gets the servers they want!
Looks like you didn't start with the new NetBSD template, which would have given you maybe 4 GB more space on the file system (not sure if that would have been enough space).
Just out of interest, how many inodes are you using?
df -hi /Right. I went with what I already had, just for fun, to see if it would work. Now I have had some fun, and learned a little. So, all good!
Yes.
16853 vs 795510
One of the reasons why I went ahead with the old image was that I wondered whether the large number of inodes would cause any kind of a problem. Looks like, until we ran out of space, everything went fine despite the large number of inodes.
When things started looking like space was going to get tight, I deleted pkgsrc and also my previously mentioned initial src checkout which had been moved to src-old.
If @linveo wants to bump my VM's disk size, I am happy to rebuild NetBSD, this time beginning by reinstalling with @cmeerw's latest image. @linveo Rather than increasing disk size, if it's easier to wipe the existing VM and provision a replacement, that's fine. If increasing my VM's disk size causes any issues, it's also fine to leave it as is. I really like this VM! It has a fast processor, and it has only 23 ms ping from my current location in Sonora.
I hope everyone gets the servers they want!
Was actually wondering how many inodes you had used compared to the new max of 3100457 - your 795510 will still comfortably fit into that max.
Google ""NetBSD" inode max 3100457" says "Your search did not match any documents".
@cmeerw If you do not mind, may I please ask, what is the best way to find out where, when, and why the inode max was increased? Why was 3100457 selected?
I hope everyone gets the servers they want!
Sorry... that's just the (max) number of inodes you will get when re-installing NetBSD from my template on that particular VM.
See the newfs man page:
Note that when creating the NetBSD image I am creating a 512 MB root filesystem, but as I am expecting the file system to be resized during first boot, I am setting "-b 16384 -f 2048" (so we get more reasonable defaults for the final filesystem size).
So the initial 512 MB file system will get approximately 65 thousand inodes, and when the filesystem gets resized to 24.5 GB, the inode density stays the same and you'll end up with roughly 3 million inodes.
BTW, the actual maximum number of inodes is
3117310(3100457was the number of available inodes on my system), but I thinkdfonly shows that number when usingdf -G /(and the values are currently swapped around on 10.0).You got it! I upped your VM to 50GB, just need a reboot. You might need to manually increase the FS too if cloud-init is not able to interact.
linveo.com | Shared Hosting | KVM VPS | Dedicated Servers
@linveo The pending changes show up on the Control Panel at my end. Before I do anything, though, I just want to double check with @cmeerw because he's been working on the NetBSD images. If there is anything that @cmeerw might want to do or check or ask me to do or check, I want to make sure that whatever @cmeerw wants is done.
My inclination might be to reinstall from @cmeerw's latest NetBSD image. I might just do Debian for a moment first, simply to exercise the system and to double check that everything works with an OS that VF might more likely be expecting. But I could proceed differently if @cmeerw or anybody else wants. Thanks!
I hope everyone gets the servers they want!