Hello!

According to the given error message:

$ stap-prep
Configuring for kernel release 6.1.0-16-amd64
Debuginfo automatic downloading is not configured via $DEBUGINFOD_URLS
/usr/bin/stap-prep: line 80: rpm: command not found

Now, I found time to look into the script.
The error is obviously easy to see, the root cause is in line 182:

(180) DISTRO="$(lsb_release --id --short 2> /dev/null)"
(181) case "$DISTRO" in
(182) Debian|Ubuntu) <===
(183)        prep_deb_based "$@"
(184)        ;;
(185) *)
(186)        prep_rpm_based "$@"
(187)        ;;
(188) esac

Just insert "|Devuan" after "Ubuntu" in the case expression and it works.
Sad, but usual for the SystemTap installations all the time, the script
contains - ALWAYS(??) - such errors.
Additionally, at the end of the result (of running 'stap-prep'), you get this:

......
Processing triggers for runit (2.1.2-54) ...
Processing triggers for libc-bin (2.36-9+deb12u3) ...
You need package linux-image-6.1.0-16-amd64-dbgsym but it does not seem to be available

No such package exist, but systemtap is running now (say, at least my old script).
Another pointless waste of time.

Regards,
Manfred

This looks for me like an individual package install, which I wished to escape.
In debian, the newer version is in the testing repository, so no single package install.
In devuan, so far I looked on the lists, there is no testing, but experimental - dont know the difference.
This might have led to your package links. A devuan repo named testing is not listed on the page:
https://www.devuan.org/os/packages

In debian it is listed:

Due to my preferences-file it is listed with "-10".
But in devuan (experimental in sources):

After your reply, I tried a deeper search and found 'testing' mentioned, without explicit link.
Entered it in the source list and updated - seems to be present!
But does NOT list xtables-addons-dkms, the list above stays the same which lead me
to "individual package install".
BTW, while trying this, the 'experimental' disappeard!

BTW, the mentioned bug, according to the mailinglist debate,
has been fixed in 2020(!!), but never came into the repositories.
Continues the pain with the old patch-o-matic (10 years ago).
Has nothing to do with "willing", I have not the competition
to build a security relevant package.

Though I am trying, I'll make an exception and download the
package - can you give a hint? The daedalus or the ceres version?
I looked to the packages and it needs "xtables-addons-common (>= 3.23)",
which is also not in the repository (has 3.13) .....

Regards,
Manfred

Yes, I stumpled over this on my naive first try and desktop users can even not save all of their,
settings, eben though it "principially" worked.
For samba, there is probably hope via ksmb in the future, I asked them on github. No time now to
ask on cifs mailing-list currently.
For kvm, where the most of my work happens, there is the same for virtiofs(d) and a request is open
to pass a full featured host filesystem - would apply for me too (using zfs).
I need to integrate additional software and will do this using chroot and/or containers.
The best bet for the moment, it looks, is to pass a blockdev to the guest and install zfs in the guest system.
I am on the road to test this and it basicall works. This would allow me to make snapshots and even replicate
them over the network. So I would be, at least, able to have history on a users homedir.
I'll look really to check your example, because it would allow me, to get rid of systemd.
With systemd, if you have configured, say, 10 users, you'll see their mounts, even they are not logged in .....
This is improper state management for users, in my eyes.

Hi!

Thank you for your investigation and a big sorry for my late reply.
I am having too much trouble at the moment with something like
this.
I've heart about pam-mount, but, due to the fact, that this moment
I stay on debian, I looked there(!), but missed it.
I just (this moment) checked for that package in the VM and it is really there.
But I must delay further, because at the moment, it's not clear if/how
I can use the whole users homedir on a share - there are hints to this,
but they do not work. I asked in the samba groub, but they directed
mit to "cifs" mailinglist and I have first to make that clear and understandable
completely. If this does not become possible, I'll install ZFS in the VM.
The whole thing is about snapshots (the samba share can be snapshotted
on the server).

Regards,
Manfred

Hi!

Comes to my mind, but must learn about it first.
Probably I need it on that debian-vm.

Manfred

Hi,

thank you.

I've already read and tried this (using the "exclude root" option) - but with different success, so the question
came up for me. The remaining processes are something like gfvs.*

You implicitely said, under systemV/etc this is not the case?

I have a lot of computers, am a linux-noob (nearly), countless VMs (kvm+lxc)
and would not be able to migrate that all fast, but were probably unable to work
with different systems.

Hallo!
I have the same problem.
The answer does'nt fit.
I have ipv6 disabled on all computers (acp-cacher-ng, proxy, chimera-host).
Just to be sure, I applied the same settings like in your answer, but no changes.
Additionally, on the chimera-box I have a lxc-container (running debian buster), which updates fine.
I found a similar post regarding ubuntu and then, I made a try to load one of the failing urls
using curl over acng:.

curl -p http://192.168.12.132:4455 http://pkgmaster.devuan.org/merged/pool/DEBIAN-SECURITY/updates/main/x/xorg-server/xvfb_1.20.11-1%2bdeb11u2_amd64.deb
curl: (56) Received HTTP code 403 from proxy after CONNECT

Went to the acng-box and I see the response from the final squid-proxy and I get:

<p>Some aspect of the requested URL is incorrect.</p>

<p>Some possible problems are:</p>
<ul>
<li><p>Missing or incorrect access protocol (should be <q>http://</q> or similar)</p></li>
<li><p>Missing hostname</p></li>
<li><p>Illegal double-escape in the URL-Path</p></li>
<li><p>Illegal character in hostname; underscores are not allowed.</p></li>
</ul>

Changing apt-settings from acng to squid, doing update, changing back to acng following an upgrade
works.
There were wrong urls/hostnames in the apt-sources (which were probably very old, because
it is an experiment (which I use to see devuan instead of debian/systemd).

Hope this helps someone, wo felt over this too.

Regards,
Manfred