Bug#1051401: general: PATH variable definition in debian 12

2023-09-14 Thread Timo Lindfors



On 9/7/23 14:59, robin hodges wrote:

Had a problem when I installed debian 12 onto my PC. As root the reboot and 
shutdown commands wouldnt work.
I have solved this on my PC by including the following into the root .bashrc 
file

export PATH=$PATH:/usr/local/sbin:/usr/sbin:/sbin


Did you login as root by typing "su" instead of "su --login"? If yes, 
this is normal behavior, check "man su":


"It is recommended to always use the --login option"

-Timo



Re: questionable massive auto-removal: buggy deps nvidia-graphics-drivers-tesla-470

2022-05-25 Thread Timo Lindfors



On 5/24/22 21:34, Paul Gevers wrote:
https://bugs.debian.org/1011268 (but apparently my first assumption 
was wrong and it's another bug, most likely Simon was right.


Thanks for the link. I was quite puzzled this morning when I saw several 
removals messages. I guess I should just wait and hope that the removals 
don't actually happen.


-Timo




Re: /usr/bin/ld.so as a symbolic link for the dynamic loader

2021-12-03 Thread Timo Lindfors

Hi,

On Fri, 3 Dec 2021, Florian Weimer wrote:

No objects to /usr/bin/ld.so then?


Just a random thought: If you have configured a restricted shell (e.g. 
rbash) that only lets you execute commands in PATH, will this make it 
possible to bypass the restriction with "ld.so /tmp/some-random-binary"? 
This is not necessary an argument to not do this, I'm just wondering what 
the implications could be.


-Timo



Re: Making Debian available

2021-01-18 Thread Timo Lindfors

On Mon, 18 Jan 2021, Marc Haber wrote:

My understanding is that the live images are primarily intended for
the typical "live system" use-case descended from things like Knoppix
(boot machine from USB stick or optical media; use machine with OS from
RAM, USB stick or optical media; shut down machine; no trace is left),
with permanent installation via Calamares being considered somewhat
experimental.


Ah indeed. My comment was a more abstract observation: if you can run the 
installer in a more familiar environment it is easier for users. I did not 
intend to say anything about the current state of any package or team.



I remember installing Debian once from a live image because the only
network I had available was using 802.1x. Must have been a rather
painless experience, since I don't remember much trouble about it.


I've also used it a couple of times. The only worry I have is that I end 
up with an installation that is different from d-i installations in some 
subtle way that will cause me to hit some weird bug in the future. I had 
similar problems with virt-install. I had to add all sorts of extra
hacks to produce installations that were as identical to manual 
installations as possible.


https://lindi.iki.fi/lindi/debian10/server has my notes on how to do this 
with a server and https://lindi.iki.fi/lindi/debian10/ is an attempt to do 
it for desktop as well. Basically I first installed debian manually. Then 
I automated it and tweaked preseed and a custom shell script until the 
filesystem difference was minimal.




Re: Making Debian available

2021-01-17 Thread Timo Lindfors


On Sun, 17 Jan 2021, Marc Haber wrote:

Absolutely. The Installation Experience is one of the first contacts
with the distribution for most people¹, and since we all know that the


Yep. I think using the live environment for installation could be more 
user-friendly as the user is already familiar with how to use e.g. USB drive or 
browser during the installation if they need to search for help or copy 
some additional firmware files. The number of people who know how to do 
this in the current d-i images is much lower since you'd need to use the 
command-line with a really restricted set of tools.

Re: isc-dhcp-client sends DHCPDISCOVER *before* wpa_supplicant authenticates/associates/connects.

2020-06-28 Thread Timo Lindfors

On Sun, 28 Jun 2020, Jaime wrote:

Given that this is guaranteed to fail, isn't it worth "not doing"? Is
there anyway that I can get dhclient to wait for a successful
connection *before* sending out any DHCPDISCOVERs? (In the wired
world, it doesn't make sense to issue a DHCPDISCOVER before plugging
the cable in.)


NetworkManager should be at least one such way. When the funciton 
supplicant_iface_state_cb is called with state 
NM_SUPPLICANT_INTERFACE_STATE_COMPLETED it will call 
nm_device_activate_schedule_stage3_ip_config_start that will end up 
starting dhclient only when wpa_supplicant has done its job.


At least last time I tried using ifupdown for wifi I hit all sorts of odd 
race conditions that probably still exist:


https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=587634




Re: Is an init required to obey policy-rc.d during boot ?

2020-04-23 Thread Timo Lindfors


On Thu, 23 Apr 2020, Thomas Goirand wrote:

doesn't exist but will proceed)
- apt install foo (shipping a foo.service)
→ foo.service will not be started


Good to know, thanks!


Is https://www.freedesktop.org/software/systemd/man/systemd.preset.html 
perhaps something that could help here?


-Timo


Re: default firewall utility changes for Debian 11 bullseye

2019-07-31 Thread Timo Lindfors

On Wed, 31 Jul 2019, Adam Borowski wrote:

A network firewall is useful.  But why would someone want a _host_ firewall
for on any sane operating system?  If a daemon is not supposed to listen on


Are libvirt and network-manager using firewalld to setup network sharing 
and virtual networks? Or do the still invoke iptables directly?