[SOLVED](Error) message "bash: append_path: command not found"

Luckylinux · 2020-09-28 22:33:05

seth wrote:

Fyi, .pacnew files *rarely* cause issues - the original mistake here is to edit /etc/profile (likely based on some aged interwebz advise): use an /etc/profile.d droplet and leave /etc/profile alone.
All other "conflicts" in #12 are expected to be edited on your site and iff a package introduces breaking changes to such a file that will require manual intervention, you can typically expect to get a heads up on the frontpage (or by mail, if you sign up for such messages)

Oh, thank you very much for supplementing the information.

Scimmia · 2020-09-28 22:49:19

seth wrote:

Fyi, .pacnew files *rarely* cause issues

I really don't like this statement, that may make people think they aren't important. When they do cause issues, it can be severe; recently, a bunch of people ignored PAM related .pacnew files and ended up with no way to log into their machines. Do NOT ignore them, ever. Sometimes they're just comments, sure, but if you ignore them, you should *expect* the related software to stop working.

Last edited by Scimmia (2020-09-28 23:03:56)

Luckylinux · 2020-09-28 23:07:44

Scimmia wrote:

seth wrote:
Fyi, .pacnew files *rarely* cause issues
I really don't like this statement, that may make people think they aren't important. When they do cause issues, it can be severe; recently, a bunch of people ignored PAM related .pacnew files and ended up with no way to log into their machines. Do NOT ignore them, ever. Sometimes they're just comments, sure, but if you ignore them, you should *expect* the related software to stop working.

Thank you for the observation. Rest assured that I will be attentive, yes. Take care \ o /

seth · 2020-09-28 23:12:47

Scimmia wrote:

When they do cause issues, it can be severe

Yes - and I hold the opinion that the tally removal would have mandated a warning because pambase is a downstream package and provides files that are expected to be locally adapted by the upstream.
Or alternatively make the files in pambase more modular/ready for client extensions w/o messing w/ the package files, https://bbs.archlinux.org/viewtopic.php … 4#p1925144

Likewise I'd expect a heads-up (or post install script) if an update would fundamentally change the syntax of /etc/passwd, but otoh feel entirely responsible for changes I made to /etc/ImageMagick-6/*.xml as a hotfix.

Scimmia · 2020-09-28 23:15:02

Both did give a heads-up, that's exactly what a .pacnew file is.

Trilby · 2020-09-28 23:22:17

I agree with Scimmia that they can / should never be ignored. There is no simple rule of "always delete" or "always overwrite" that would result in good outcomes. If there was, pacman could do this itself and not bother with the files. Pacnew files exist because there is no trivial algorithmic means to determine what should be done as it depends on countless factors around each user's system. That said, if one insisted on a simple heuristic, I'm pretty sure the strategy of "always delete the pacnews" would be - on average - significantly less bad than "always overwrite the old file".

The reason for this is two-fold: 1) in a substantial portion of cases, simply deleting the pacnew is appropriate, and 2) in the cases where it is the wrong approach, doing so will not destroy any data that cannot be recovered (as all that's deleted are the defaults which are still available in the package cache or in the package on the mirrors). If one overwrites their existing files with pacnew files, they will often destroy data that cannot be recovered (aside from going through the steps again that created it: e.g., recreating users, etc).

EDIT: and none of this should imply that these two extremes are the only options. One should merge content from pacnew files into their existing files.

Last edited by Trilby (2020-09-28 23:25:58)

seth · 2020-09-28 23:25:04

It's not really a "heads up" if one gets that w/ every update for a bunch of files that users are expected to alter locally.

A pacnew is certainly sufficient to figure "hmmm, this doesn't work anymore and there's a unresolved config update - maybe I should look at that" but not if a reboot kicks your out of the system.
If I shall server as such, there neeeds to be less noise.

Or I just don't know what "heads up" actually means *shrug*

Scimmia · 2020-09-28 23:30:56

How in the world do you get .pacnew files that often, seth? It only happens if you have edited the file AND it's different than in the previous package. They're fairly rare.

Trilby · 2020-09-28 23:50:17

Rare is subjective. I've average 2 per month on my current machine this year, and my server has averaged 2.5 per month over the past 6 years.

seth · 2020-09-29 06:04:29

Actually, because I went with my gut feeling (apparently a shit adviser - just sayin' … in case somebody wants to apply it on sth. like a global pandemic or so…)

Looking at the actual data, it seems largely tilted by the rather constant mirrorlist conflicts (file now went NoExtract) and the recent bulk from the filesystem package.
Otherwise, there's just "always something" but actually not that bad (leading offenders are systemd/{journald,system}.conf (I'll have a look whether I can move the local changes) and locale.gen (NoExtract as well)

And I still think that "hey, we're breaking PAM" would have mandated a PSA (and no: wasn't personally affected - just from the constellation)

GaKu999 · 2020-09-29 07:10:10

seth wrote:

And I still think that "hey, we're breaking PAM" would have mandated a PSA (and no: wasn't personally affected - just from the constellation)

That would depend on one's perspective, unsupervised update caused half of the issues with that update (which is not quite sane to do), mainly because the locker activated and locked out the user.
Or worse, the user manually locked during the update...

Users must watch the entire update process and later merge their pacnews if any before shutdown, logout, or lock.
Since they aren't surprises, because they are expected and properly logged, they don't have room in the announcements.

Maybe a pacman hook doing 'pacdiff -o' at the very end of every update would be a "heads up"?
If it's at the end there's absolutely no chance that someone will miss it by accident. But that is yet again dependent on one's perspective.

nl6720 · 2020-09-29 07:51:28

seth wrote:

leading offenders are systemd/{journald,system}.conf (I'll have a look whether I can move the local changes)

Just place your changes in /etc/systemd/{journald,system}.conf.d/*.conf.

kyzxr · 2020-09-30 02:44:32

I found this error with a machine that had the same issue with an /etc/profile.pacnew that I hadn't noticed. Setting aside my own error of not paying close attention to updates to fix issues such as this, I wanted to point out to the support team here that the top "2020" google hit for the search term "/etc/profile.d/perlbin.sh:6: command not found: append_path 2020" leads to bug 67999, which is closed without any leads to a fix. This extremely useful thread here unfortunately doesn't show up anywhere in the first page of google search results.

If you continue to get activity at the bug ticket because of Google searches, I would suggest leaving a comment at the bottom of that ticket pointing to this thread which has great advice and information on how to fix the issue.

Last edited by kyzxr (2020-09-30 02:45:11)

Trilby · 2020-09-30 02:50:07

The solution is in the message on that bug report.

svfedorov · 2020-10-12 19:47:55

I have the same issue, but the /etc/profile.pacnew file is missing (I never have edited it manually).
I use bash and tmux.
The actual output after opening a new pane:

bash: append_path: command not found
bash: append_path: command not found
bash: append_path: command not found
#fedorov@fedorov:~:[0]$

Output of pacman -Qikk filesystem

Name            : filesystem
Version         : 2020.09.03-1
Description     : Base Arch Linux files
Architecture    : x86_64
URL             : https://www.archlinux.org
Licenses        : GPL
Groups          : None
Provides        : None
Depends On      : iana-etc
Optional Deps   : None
Required By     : base  fakeroot  glibc  mkinitcpio
Optional For    : None
Conflicts With  : None
Replaces        : None
Installed Size  : 31.64 KiB
Packager        : S
Build Date      : Thu Sep 3 02:30:21 2020
Install Date    : Mon Oct 12 23:21:01 2020
Install Reason  : Explicitly installed
Install Script  : No
Validated By    : Signature

backup file: filesystem: /etc/fstab (Modification time mismatch)
backup file: filesystem: /etc/fstab (Size mismatch)
backup file: filesystem: /etc/group (Modification time mismatch)
backup file: filesystem: /etc/group (Size mismatch)
backup file: filesystem: /etc/gshadow (Modification time mismatch)
backup file: filesystem: /etc/gshadow (Size mismatch)
backup file: filesystem: /etc/hosts (Modification time mismatch)
backup file: filesystem: /etc/hosts (Size mismatch)
backup file: filesystem: /etc/passwd (Modification time mismatch)
backup file: filesystem: /etc/passwd (Size mismatch)
backup file: filesystem: /etc/resolv.conf (Modification time mismatch)
backup file: filesystem: /etc/resolv.conf (Size mismatch)
backup file: filesystem: /etc/shadow (Modification time mismatch)
backup file: filesystem: /etc/shadow (Size mismatch)
backup file: filesystem: /etc/shells (Modification time mismatch)
backup file: filesystem: /etc/shells (Size mismatch)
filesystem: 118 total files, 0 altered files

Trilby · 2020-10-12 20:14:12

svfedorov, do you have a long-running tmux session you are reattaching to? Or even medium-running as it looks like you just got the filesystem update ... have you logged out and back in again?

That would be part of an explanation to your systems - of course not a full explanation. Nothing outside of /etc/profile or the files it sources in /etc/profile.d should attempt to use that function. Clearly something in one of your startup files is trying to use that function without actually sourcing /etc/profile itself. Please post your shellrc file(s).

svfedorov · 2020-10-12 20:50:23

svfedorov, do you have a long-running tmux session you are reattaching to?

No, I can reproduce the issue by just starting the tmux session. The new konsole tab opens without any errors.

EDITED: Found the sourced bash script that referenced /etc/profile.d/*.sh files directly. Sorry for the inconvenience.
PS: This script worked well for around 7 years. Didn't expect it to fail.

Last edited by svfedorov (2020-10-12 21:22:49)

2ManyDogs · 2020-10-12 22:14:08

Closing this solved topic.

Arch Linux

#26 2020-09-28 22:33:05

Re: [SOLVED](Error) message "bash: append_path: command not found"

#27 2020-09-28 22:49:19

Re: [SOLVED](Error) message "bash: append_path: command not found"

#28 2020-09-28 23:07:44

Re: [SOLVED](Error) message "bash: append_path: command not found"

#29 2020-09-28 23:12:47

Re: [SOLVED](Error) message "bash: append_path: command not found"

#30 2020-09-28 23:15:02

Re: [SOLVED](Error) message "bash: append_path: command not found"

#31 2020-09-28 23:22:17

Re: [SOLVED](Error) message "bash: append_path: command not found"

#32 2020-09-28 23:25:04

Re: [SOLVED](Error) message "bash: append_path: command not found"

#33 2020-09-28 23:30:56

Re: [SOLVED](Error) message "bash: append_path: command not found"

#34 2020-09-28 23:50:17

Re: [SOLVED](Error) message "bash: append_path: command not found"

#35 2020-09-29 06:04:29

Re: [SOLVED](Error) message "bash: append_path: command not found"

#36 2020-09-29 07:10:10

Re: [SOLVED](Error) message "bash: append_path: command not found"

#37 2020-09-29 07:51:28

Re: [SOLVED](Error) message "bash: append_path: command not found"

#38 2020-09-30 02:44:32

Re: [SOLVED](Error) message "bash: append_path: command not found"

#39 2020-09-30 02:50:07

Re: [SOLVED](Error) message "bash: append_path: command not found"

#40 2020-10-12 19:47:55

Re: [SOLVED](Error) message "bash: append_path: command not found"

#41 2020-10-12 20:14:12

Re: [SOLVED](Error) message "bash: append_path: command not found"

#42 2020-10-12 20:50:23

Re: [SOLVED](Error) message "bash: append_path: command not found"

#43 2020-10-12 22:14:08

Re: [SOLVED](Error) message "bash: append_path: command not found"

Board footer