Hi, all I'm surprising, the pkg_info -D package_name not run correctly. Example, on my 9.2 fresh install, after installed nano: $ pkg_info -D nano pkg_info: cant find package 'nano' if(window.hljsLoader && !document.currentScript.parentNode.hasAttribute('data-s9e-livepreview-onupdate')) { window.hljsLoader.highlightBlocks(document.currentScript.parentNode); } On the NetBSD matrix chanel, someone said me to view this manual instructions https://pkgsrc.org/pkgdb-change/ But, I'm on fresh install, with pkgin installed during the install process. I'm very surprised that it is necessary to do in addition! (too, I've not install pkgsrc) I'm going to test-it, but I remain skeptical! 😅

OK, I do changes segun manual instructions. Now, the sh returns: $ pkg_info -D pkgin pkg_info: Entry pkg in pkgdb is not a valid package name Information for pkgin-20.12.1nb1: $ pkg_info -D nano pkg_info: Entry pkg in pkgdb is not a valid package name pkg_info: can't find package 'nano' if(window.hljsLoader && !document.currentScript.parentNode.hasAttribute('data-s9e-livepreview-onupdate')) { window.hljsLoader.highlightBlocks(document.currentScript.parentNode); } Ok, I've not seen this line In every file in the refcount directory, change the old path to the new path. (See the script section above.) But on /usr/pkg/pkgdb.refcount/files/usr/pkg/etc/pkgin/repositories.conf/pkgin-20.12.1bn1, it"s clearly wrote: /usr/pkg/pkgdb/pkgin-20.12.1nb1 It's not case for nano, or others packages installed before migrating. Why it's so complicated?

EDIT : script updated to work on _STABLE too. CiotBSD Why it's so complicated? It need not be. Since you're running NetBSD in a VM, just redo the install, but this time skip the pkgsrc and pkgin steps. After install finishes, login to the installed system as root and run the script below first thing. Then, you can use pkgin to install packages. #!/bin/sh set -eux D="PKG_DBDIR=/usr/pkg/pkgdb" M=$(uname -m) # arch. R=$(uname -r) # release R=${R%%_*} # strip `_whatever' # Ensure install is pristine. test \! -r /etc/mk.conf test \! -r /etc/pkg_install.conf test \! -d /usr/pkg/etc # Ensure the new `/usr/pkg/pkgdb' dir. is used by everything. echo "$D" >> /etc/mk.conf echo "$D" >> /etc/pkg_install.conf mkdir -p /usr/pkg/etc echo "$D" >> /usr/pkg/etc/pkg_install.conf # Install pkgin(1) first. PKG_PATH="https://cdn.NetBSD.org/pub/pkgsrc/packages/NetBSD/$M/$R/All" export PKG_PATH pkg_add pkgin # Update pkgin(1) database. pkgin update # Remove obsolete dir. This *should* succeed if everything went OK. rmdir /var/db/pkg || { echo "$0: procedure failed."; exit 1; } if(window.hljsLoader && !document.currentScript.parentNode.hasAttribute('data-s9e-livepreview-onupdate')) { window.hljsLoader.highlightBlocks(document.currentScript.parentNode); } And, use the latest -STABLE images nstead of -RELEASE : they're quite stable, and have all the important fixes. I just tried this with 9.2-RELEASE & 9.2_STABLE on real HW and had no issues.

pin The issue is that the pkgdb path needs to be correct in the system building the binaries, because the default path to pkg_admin is hardcoded at build time Yeah, it's not a happy state of affairs, I'm afraid (esp. for new users). From my tests this morning (with no PKG_DBDIR= overrides anywhere): The /usr/sbin/pkg_* binaries on 9.2-RELEASE default to using /usr/pkg/pkgb for the DB path. So a pkg_add pkgin will stash the pkgin package's metadata there. But, the pkgin package also brings in similarly named binaries which are installed as /usr/pkg/sbin/pkg_* (This is pkgsrc stuff being kept separate from the OS stuff, I think.) However, these binaries seem to prefer /var/db/pkg in the absence of any config. Because doing pkgin blah writes metadata into that location. I thought this had been fixed--last December...oh well, at least the work-around is simple.

CiotBSD # test \! -r /etc/mk.conf test: -r: unexpected operator # test \! -r /etc/pkg_install.conf test: -r: unexpected operator # test \! -d /usr/pkg/etc test: -d: unexpected operator if(window.hljsLoader && !document.currentScript.parentNode.hasAttribute('data-s9e-livepreview-onupdate')) { window.hljsLoader.highlightBlocks(document.currentScript.parentNode); } This shouldn't have happened (was there a space between \ and ! ?) Which shell did you use? (Just run the script instead of typing everything in, mate, otherwise what're those tests I put in for?) pin It doesn't matter, you can have all three overrides and it will still use /var/db/pkg if the binaries were build on a system with that default. Any more good news, father?

pkg_info -D binary: cant find package

CiotBSD

OK, I will go to reinstall on my Laptop, not into VM (was my first try, abandonned because VM suspended at time to install XFCE; I tested tree times)
with your information @rvp

And Yes, I downloaded the ultimate 9.2 stable on the netbsd website.

Egual, I stop this: https://www.unitedbsd.com/d/567-sh-package-name-not-found/21
(enough!)

rvp

CiotBSD not into VM (was my first try, abandonned because VM suspended at time to install XFCE; I tested tree times) with your information

Yeah. I thought it might be some sort of VM misconfiguration. Barring HW issues (I saw something very much like this on Android some years ago), there's no way any OS is going to let its files disappear into a black hole like that.

rvp

pin The issue is that the pkgdb path needs to be correct in the system building the binaries, because the default path to pkg_admin is hardcoded at build time

Yeah, it's not a happy state of affairs, I'm afraid (esp. for new users). From my tests this morning (with no PKG_DBDIR= overrides anywhere):

The /usr/sbin/pkg_* binaries on 9.2-RELEASE default to using /usr/pkg/pkgb for the DB path.
So a pkg_add pkgin will stash the pkgin package's metadata there.
But, the pkginpackage also brings in similarly named binaries which are installed as /usr/pkg/sbin/pkg_* (This is pkgsrc stuff being kept separate from the OS stuff, I think.)
However, these binaries seem to prefer /var/db/pkg in the absence of any config.
Because doing pkgin blah writes metadata into that location.

I thought this had been fixed--last December...oh well, at least the work-around is simple.

rvp Yeah, it's not a happy state of affairs, I'm afraid (esp. for new users). From my tests this morning (with no PKG_DBDIR= overrides anywhere)

It doesn't matter, you can have all three overrides and it will still use /var/db/pkg if the binaries were build on a system with that default.

CiotBSD

OK, New fresh install of 9.2 RELEASE, on my laptop - an oldier Dell XPS M1330:

sha512: OK
Install: OK
I followed the script process without writing the script, into the shell:

# D="PKG_DBDIR=/usr/pkg/pkgdb"
# M=$(uname -m)
# R=$(uname -r)
# R=${R%%_*}	

# test \! -r /etc/mk.conf
test: -r: unexpected operator
# test \! -r /etc/pkg_install.conf
test: -r: unexpected operator
# test \! -d /usr/pkg/etc
test: -d: unexpected operator

# echo "$D" >> /etc/mk.conf
# echo "$D" >> /etc/pkg_install.conf
# mkdir -p /usr/pkg/etc
# echo "$D" >> /usr/pkg/etc/pkg_install.conf

# PKG_PATH="https://cdn.NetBSD.org/pub/pkgsrc/packages/NetBSD/$M/$R/All"
# export PKG_PATH
# echo $PKG_PATH
https://cdn.NetBSD.org/pub/pkgsrc/packages/NetBSD/amd64/9.2/All

# pkg_add pkgin
pkg_add: Warning: package 'pkgin-20.12.1nb1' was built for a platform:
pkg_add: NetBSD/x86_64 9.0 (pkg) vs. NetBSD/x86_64 9.2 (this host)
pkg_add: Warning: package 'pkg_install-20210410' was built for a platform:
pkg_add: NetBSD/x86_64 9.0 (pkg) vs. NetBSD/x86_64 9.2 (this host)

Stored 17 files and 0 explicit directories from 1 package in /usr/pkg/pkgdb
/usr/pkg/pkgdb/pkgdb.byfile.db.
Done.
/usr/pkg/pkgdb: + *
                                 .* *
                                      .., *: not found

# pkgin update
processing remote summary (http://cdn.NetBSD.org/pub/pkgsrc/packages/NetBSD/amd64/9.2/All)...
pkg_summary.bz2                                                                                                                                        100% 3617KB 226.0KB/s   00:16

# rmdir /var/db/pkg || { echo "$0: procedure failed."; exit 1; }
# pkg_info -D pkgin
Information for pkgin-20.12.1nb1:

#

It's normal pkgin was built for 9.0 and used on 9.2?

bsduck

CiotBSD It's normal pkgin was built for 9.0 and used on 9.2?

Yes, there is binary compatibility between point releases.

rvp

CiotBSD

# test \! -r /etc/mk.conf
test: -r: unexpected operator
# test \! -r /etc/pkg_install.conf
test: -r: unexpected operator
# test \! -d /usr/pkg/etc
test: -d: unexpected operator

This shouldn't have happened (was there a space between \ and !?) Which shell did you use?
(Just run the script instead of typing everything in, mate, otherwise what're those tests I put in for?)

pin It doesn't matter, you can have all three overrides and it will still use /var/db/pkg if the binaries were build on a system with that default.

Any more good news, father?

rvp Any more good news, father?

I haven't used official repositories for more then 2 years but, I hope those are fixed. pkgsrc-current builds for 9.x_STABLE provided by mef are good and builds from jperkin are also good.

Honestly, right now I don't know if others are ok. Maybe I could try in a Qemu-vm but, right now I simply don't have the time for it.

CiotBSD

rvp there a space between \ and !?)

rvp Which shell did you use?

the shell, by default: /bin/sh

rvp Just run the script instead of typing everything in

how ? when I have not a functional USB pen to copy from another station ? Either, I do to write with vi… or into the shell! I choose to type into shell, because I dont like use vi.

rvp

CiotBSD the shell, by default: /bin/sh

Puzzling then...

CiotBSD how ? when I have not a functional USB pen to copy from another station ?

Since this step is post-install, you should've networking. So ftp/http/ssh/netcat the blessed file in.

CiotBSD I dont like use vi.

You like ed(1)?

CiotBSD

rvp You like ed(1)?

nano, damned! :p

rvp So ftp/http/ssh/netcat the blessed file in.

ohh, yes, with termbin. 😃
(sorry)

JuvenalUrbino

rvp You like ed(1)?

Ed, man! !man ed

rvp

CiotBSD nano, damned! :p

nano is not in base. In a pinch, you could use the shell itself.

JuvenalUrbino

CiotBSD ohh, yes, with termbin. 😃
(sorry)

cat $file | nc termbin.com 9999
then you can retrieve it from the url provided using ftp(1)

However, as suggested, if you enable ftpd or sshd you can use ftp and scp respectively. If you gain remote access to your netbsd box (on a local network you just use telnet, enable it on /etc/inetd.conf) you can easily fix stuff by working from you daily driver. Think about having 2 vertical tiled windows, with half screen occupied by your browser at UnitedBSD and half by a terminal remotely connected to your netbsd client, copy pasting stuff across windows using the built-in features of your terminal emulator of choice. How does it sound? Damn how do you think people manage servers at home? Mine have always run headless, and only exceptionally would I have been forced to attach a serial console.

CiotBSD

rvp nano is not in base.

yes, I known; even, it's not on others *BSD, by default.

rvp you could use the shell itself.

ohhh, really interesting! Ty! 😃

I think I love this community. I learn many things. Really, thank you.

CiotBSD

JuvenalUrbino However, as suggested, if you enable ftpd or sshd you can use ftp and scp respectively. If you gain remote access to your netbsd box (on a local network you just use telnet, enable it on /etc/inetd.conf) you can easily fix stuff by working from you daily driver. Think about having 2 vertical tiled windows, with half screen occupied by your browser at UnitedBSD and half by a terminal remotely connected to your netbsd client, copy pasting stuff across windows using the built-in features of your terminal emulator of choice. How does it sound? Damn how do you think people manage servers at home? Mine have always run headless, and only exceptionally would I have been forced to attach a serial console.

Eh, oh, keep calm. I manage too OpenBSD or Linux servers, with SSH.
excuse me for forgetting this, because I discover netbsd, and that I focus on the problem.
And, you must admit that since the beginning when I try, I still encounter a lot of problems!

Your attitude does not help solve anything!
🙁

JuvenalUrbino

CiotBSD sorry, I must have been misunderstood. I owe you an apology. It was not my intention to make you feel criticized or judged
My tone (playful, positive, benevolent) got lost across the written communication. My sole intention was to provide you with some nice/convenient suggestions, based on the solutions I adopted while dealing, as an amateur, with self-hosting

rvp

JuvenalUrbino Yeah. I was gonna post that exact link had OP also complained about using ed(1), but he didn't.

oui

a
This thread is wholesome.
.
q
?
q
^D

rvp

oui "The mighty ed(1) has spoken"