Re: Buster without systemd? [with backtrack]

[mick crane]

On 2020-03-27 03:12, David Wright wrote:


I'm still quite happy to run with their choices.


I'm also very happy to use this free software that works.
for example I used to use fetchmail and procmail now I use getmail and 
dovecot-deliver.
I have no idea how dovecot does the mail transport but I'm sure I could 
find out, same way I could probably find out about systemd even though 
this old fart moans about things getting more complicated.


mick
--
Key ID4BFEBB31

Re: Buster without systemd? [with backtrack]

[David Wright]

On Thu 26 Mar 2020 at 08:03:38 (+0100), deloptes wrote:
> David Wright wrote:
> 
> Well I do not know o use systemd-networkd, so your assumption is not
> correct. It was meant and in the context of some kind of GUI be it CLI or X

OK, so I completely misinterpreted what the "it" referred to in
your response to Andrei ("I find it amazing how *it* works").

We'd been talking about the increasing "desires" of systemd,
including "[to] manage network interfaces", ie systemd-networkd,
which was specifically mentioned by Andrei. NetworkManager had
only cropped up in a list of examples of other software, so
I wasn't aware we were changing subject.

That's cleared that up.

> > I don't want a desktop. In fact, wicd doesn't even need X, as it can
> > run quite happily on a VC to configure a new AP.
> > 
> > When I return to somewhere I have been before, wicd (the daemon)
> > usually connects before I have typed my passphrase to unlock /home.
> > (That assumes I'm logging in.)
> 
> This is good that you have something useful and you can work with it. Please
> do not impose to others. Everybody is free to use whatever suits the needs
> the best.

Where do you get the idea that I'm imposing anything on anyone?
I just wrote what I don't use (system-networkd) and what I do
(wicd), and gave my reason, something that people often overlook
when they suggest software. (Look at the thread "how to keep 2 PCs
partially in sync" for example.)

> May be 10y ago I've used wicd several time. In my world all GTK is ... well
> crap. I do not know why people want to write something like this in C, but
> on the other hand if it works - it works.

I've never used wicd-gtk so, as well as not talking about NetworkManager,
I'm not talking about GTK, nor about what is or isn't written in C.
I use wicd-curses, a TUI that runs, as I wrote, on VCs as well as X.
I have no idea what language(s) wicd and curses are written in, and
that played no part in my decision of what to use.

> Anyway this is my personal opinion. If you ask Torvald about C++ he has the
> same opinion as me about C.
> It is good that there are different opinions

Sure, but I wonder why you interpret my posts as complaints, or as
impositions on other people. Especially in a thread about systemd,
which always seems to bring out the worst in some people. I'm neutral
on systemd, and use it because Debian does. Having used Debian since
its first release, I'm still quite happy to run with their choices.

Cheers,
David.

Re: Buster without systemd? [with backtrack]

[rhkramer]

I can't help you (don't use WiFi / Network Manager / etc. very often), but I 
applaud you for putting the effort into trying to clarify the discussion!

(No new content below this line.)

On Wednesday, March 25, 2020 10:02:17 PM David Wright wrote:
> On Wed 25 Mar 2020 at 20:18:29 (+), Liam O'Toole wrote:
> > For what it's worth, Network Manager doesn't need a GUI either. Tools
> > such as nmcli and nmtui allow you to configure and control network
> > connections from the command line.
> 
> There seems to be some confusion in this subthread. Going back a little:
> 
> On Tue 24 Mar 2020 at 12:14:57 (+0200), Andrei POPESCU wrote:
> > systemd-networkd is not meant to do very complicated configurations.
> 
> On Tue 24 Mar 2020 at 14:49:03 (+0100), deloptes wrote:
> > well - for static IP you don't need systemd, but it is a relief for
> > the dynamic stuff - i.e. wireless and cabled networks that change. I
> > find it amazing how it works  not that I say I am starting to
> > love systemd :) […]
> 
> On Tue 24 Mar 2020 at 10:44:54 (-0500), David Wright wrote:
> > I looked at what documentation I could find, but carried on using
> > 
> > wicd-curses, and the arch wiki seems to agree with what I found:
> >"systemd-networkd does not have a proper interactive management
> >
> > interface neither via command-line nor graphical".
> > 
> > The interface is what matters when you're travelling with a laptop.
> 
> So this is a conversation about systemd-networkd, not NetworkManager,
> in a thread that's about systemd, not Gnome or any other desktop/DE.
> 
> I think the next message in the subthread led to others' confusion,
> so I'm going to add two annotations to the quote (which should clarify
> what I understand it to mean), and then repeat the reply I gave before:
> 
> On Tue 24 Mar 2020 at 20:34:24 (+0100), deloptes wrote:
> > David Wright wrote:
> > > I looked at what documentation I could find, but carried on using
> > > wicd-curses, and the arch wiki seems to agree with what I found:
> > > 
> > > "systemd-networkd does not have a proper interactive management
> > > interface neither via command-line nor graphical".
> > > 
> > > The interface is what matters when you're travelling with a laptop.
> > 
> > But the purpose of systemd is not to give you an interface. This is
> > provided
> 
>  ↑↑↑ Here, I presume the word intended is
> "systemd-networkd". ↓↓↓
> 
> > by each desktop. Systemd will give you the low level service management -
> > AFAIK it works via dbus.
> 
>Here, I presume that the "network manager" that
> systemd ↓is talking to (via dbus) is systemd-networkd,
>  ↓and *not* NetworkManager (aka network-manager).
>  ↓
>↓↓↓
> 
> > systemd <-> dbus <-> network-manager
> > 
> > network-manager-interface
> > 
> > This is my understanding how it works or should work.
> 
> I don't want a desktop. In fact, wicd doesn't even need X, as it can
> run quite happily on a VC to configure a new AP.
> 
> When I return to somewhere I have been before, wicd (the daemon)
> usually connects before I have typed my passphrase to unlock /home.
> (That assumes I'm logging in.)
> 
> Cheers,
> David.

Re: Buster without systemd? [with backtrack]

[Liam O'Toole]

On Wed, 25 Mar, 2020 at 21:02:17 -0500, David Wright wrote:
> On Wed 25 Mar 2020 at 20:18:29 (+), Liam O'Toole wrote:
> > 
> > For what it's worth, Network Manager doesn't need a GUI either. Tools
> > such as nmcli and nmtui allow you to configure and control network
> > connections from the command line.
> 
> There seems to be some confusion in this subthread. Going back a little:

[...]

My remark was simply a response to yours about wicd not requiring X. I
hope it hasn't added to the confusion.

Re: Buster without systemd? [with backtrack]

[deloptes]

David Wright wrote:

>> But the purpose of systemd is not to give you an interface. This is
>> provided
> 
> ↑↑↑ Here, I presume the word intended is "systemd-networkd".
> ↓↓↓
>> by each desktop. Systemd will give you the low level service management -
>> AFAIK it works via dbus.
> 

Well I do not know o use systemd-networkd, so your assumption is not
correct. It was meant and in the context of some kind of GUI be it CLI or X

> Here, I presume that the "network manager" that systemd
> ↓        is talking to (via dbus) is systemd-networkd,
> ↓        and *not* NetworkManager (aka network-manager).
> ↓
> ↓↓↓
>> systemd <-> dbus <-> network-manager
>> |
>> network-manager-interface
>> 

May be but again I do not know systemd-networkd.

If we go back you would see it was exactly about the combination of systemd
and network-manager

>> This is my understanding how it works or should work.
>> 
> 
> I don't want a desktop. In fact, wicd doesn't even need X, as it can
> run quite happily on a VC to configure a new AP.
> 
> When I return to somewhere I have been before, wicd (the daemon)
> usually connects before I have typed my passphrase to unlock /home.
> (That assumes I'm logging in.)
> 

This is good that you have something useful and you can work with it. Please
do not impose to others. Everybody is free to use whatever suits the needs
the best.

May be 10y ago I've used wicd several time. In my world all GTK is ... well
crap. I do not know why people want to write something like this in C, but
on the other hand if it works - it works.
Anyway this is my personal opinion. If you ask Torvald about C++ he has the
same opinion as me about C.
It is good that there are different opinions

regards

Re: Buster without systemd? [with backtrack]

[David Wright]

On Wed 25 Mar 2020 at 20:18:29 (+), Liam O'Toole wrote:
> 
> For what it's worth, Network Manager doesn't need a GUI either. Tools
> such as nmcli and nmtui allow you to configure and control network
> connections from the command line.

There seems to be some confusion in this subthread. Going back a little:

On Tue 24 Mar 2020 at 12:14:57 (+0200), Andrei POPESCU wrote:
> systemd-networkd is not meant to do very complicated configurations.

On Tue 24 Mar 2020 at 14:49:03 (+0100), deloptes wrote:
> well - for static IP you don't need systemd, but it is a relief for the
> dynamic stuff - i.e. wireless and cabled networks that change.
> I find it amazing how it works  not that I say I am starting to love 
systemd :) 
> […]

On Tue 24 Mar 2020 at 10:44:54 (-0500), David Wright wrote:
> I looked at what documentation I could find, but carried on using
> wicd-curses, and the arch wiki seems to agree with what I found:
>"systemd-networkd does not have a proper interactive management
> interface neither via command-line nor graphical".
> The interface is what matters when you're travelling with a laptop.

So this is a conversation about systemd-networkd, not NetworkManager,
in a thread that's about systemd, not Gnome or any other desktop/DE.

I think the next message in the subthread led to others' confusion,
so I'm going to add two annotations to the quote (which should clarify
what I understand it to mean), and then repeat the reply I gave before:

On Tue 24 Mar 2020 at 20:34:24 (+0100), deloptes wrote:
> David Wright wrote:
> 
> > I looked at what documentation I could find, but carried on using
> > wicd-curses, and the arch wiki seems to agree with what I found:
> > 
> > "systemd-networkd does not have a proper interactive management
> > interface neither via command-line nor graphical".
> > 
> > The interface is what matters when you're travelling with a laptop.
> 
> But the purpose of systemd is not to give you an interface. This is provided

 ↑↑↑ Here, I presume the word intended is 
"systemd-networkd".
   ↓↓↓
> by each desktop. Systemd will give you the low level service management -
> AFAIK it works via dbus.

   Here, I presume that the "network manager" that systemd
   ↓is talking to (via dbus) is systemd-networkd,
   ↓and *not* NetworkManager (aka network-manager).
   ↓
   ↓↓↓
> systemd <-> dbus <-> network-manager
>   |
> network-manager-interface
> 
> This is my understanding how it works or should work.
> 

I don't want a desktop. In fact, wicd doesn't even need X, as it can
run quite happily on a VC to configure a new AP.

When I return to somewhere I have been before, wicd (the daemon)
usually connects before I have typed my passphrase to unlock /home.
(That assumes I'm logging in.)

Cheers,
David.

Re: Buster without systemd?

[Liam O'Toole]

On Tue, 24 Mar, 2020 at 22:04:07 -0500, David Wright wrote:

[...]

> I don't want a desktop. In fact, wicd doesn't even need X, as it can
> run quite happily on a VC to configure a new AP.
> 
> When I return to somewhere I have been before, wicd (the daemon)
> usually connects before I have typed my passphrase to unlock /home.
> (That assumes I'm logging in.)
> 
> Cheers,
> David.
> 

For what it's worth, Network Manager doesn't need a GUI either. Tools
such as nmcli and nmtui allow you to configure and control network
connections from the command line.

Re: Buster without systemd?

[David Wright]

On Wed 25 Mar 2020 at 08:22:33 (+0100), deloptes wrote:
> David Wright wrote:
> 
> > I don't want a desktop. In fact, wicd doesn't even need X, as it can
> > run quite happily on a VC to configure a new AP.
> > 
> 
> I did not tell you what you want. You just complained there is no interface.

I didn't *complain* that systemd-networkd hadn't got an interface—I just
said that I needed one, which is why I've carried on using wicd. There
are occasions when time is of the essence in configuring a wireless
connection, which is why I consider it of overriding importance.

> > When I return to somewhere I have been before, wicd (the daemon)
> > usually connects before I have typed my passphrase to unlock /home.
> > (That assumes I'm logging in.)
> 
> I gave up on the Gnome spooks many years ago. But AFAIK wicd is the same or
> similar as network-manager - well the GNome/GTK way of doing things.

Sorry, I don't want a DE, I don't use one, and I don't know anything
about the "Gnome way of doing things", so I'll refrain from commenting
on how network-manager¹ does what it does. I thought we were discussing
systemd-networkd, hence my quotation about it from the arch wiki.

To be fair, I haven't tried to set up systemd-networkd for the
reason given. Sometimes I've left machines as installed, with
ifupdown running the show (not my laptops). Eventually, though,
I usually end of installing wicd regardless, when I find I can't
communicate with it remotely—wicd just seems better at bringing
and keeping the networking up.

¹ I'll opine that its name is unfortunate—rather like inventing a
  currency and calling it "Currency" or "Money".

Cheers,
David.

Re: Buster without systemd?

[Andrei POPESCU]

On Ma, 24 mar 20, 20:31:17, deloptes wrote:
> Reco wrote:
> 
> > Actually, deloptes is right and you're not.
> > 
> > Kernel's documentation - [1] - describes a way to force a kernel itself
> > to configure a network interface for IPv4, be it static or DHCP, along
> > with the primitive routing table (default gw at most).
> > 
> > Taking IPv6's RA into the account, one does not need anything but the
> > working kernel on the client side to get IPv6 and a primitive routing
> > table.
> >
> [/usr/share/doc/linux-doc-4.19/Documentation/filesystems/nfs/nfsroot.txt.gz
> 
> Thank you Reco,
> you obsoleted my reply.

Ok, I'll bite.

$ cat /proc/cmdline
ip=192.168.1.64::192.168.1.1::a64p:eth0:off root=LABEL=a64p rootwait rw 
rootflags=noatime

The interface is not even up after start and if I bring it up manually 
with 'ip' it doesn't have any IPv4 address configured (only IPv6, which 
is to be expected here).

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Gene Heskett]

On Wednesday 25 March 2020 05:42:12 Kamil Jońca wrote:

> deloptes  writes:
> > Kamil Jońca wrote:
> >>> Kamil Jońca wrote:
>  Forget it. Show me how call script in reaction to DHPC events:
>  BOUND RENEW REBIND TIMEOUT
> >>>
> >>> what for?
> >>
> >> To convince me that systemd has enough functionality.
> >
> > I don't have to do anything.
>
> Of course. And I don't have to take you seriously.
>
> KJ

You probably should, he is one of the more helpfull people about. But you 
need to lose that chip on your shoulder before that will work well now.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: Buster without systemd?

[The Wanderer]

On 2020-03-25 at 07:02, deloptes wrote:


> And BTW I can not find anything like networkd
> 
> $ apt-cache search systemd | grep network
> networkd-dispatcher - Dispatcher service for systemd-networkd connection
> status changes

$ apt-file search system$ apt-file search systemd | grep networkd$
systemd: /lib/systemd/systemd-networkd

> or resolved
> 
> $ apt-cache search systemd | grep resolv
> openvpn-systemd-resolved - integrates OpenVPN with systemd-resolved
> resolvconf-admin - setuid helper program for setting up the local DNS
> libnss-mymachines - nss module to resolve hostnames for local container
> instances
> libnss-resolve - nss module to resolve names via systemd-resolved

$ apt-file search systemd | grep resolved$
openvpn-systemd-resolved: /etc/openvpn/update-systemd-resolved
systemd: /lib/systemd/systemd-resolved

Both are in the main systemd package - so anyone who has systemd
installed (via Debian) at all, has them installed.

-- 
   The Wanderer

The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man. -- George Bernard Shaw



signature.asc
Description: OpenPGP digital signature

Re: Buster without systemd?

[Ansgar]

deloptes writes:
> but I do not have systemd-networkd or -resolved installed

There are no separate packages for systemd-networkd and -resolved;
they are shipped as part of the systemd package, but are not enabled
by default.

You have enable and configure them for them to do anything.

Ansgar

Re: Buster without systemd?

[deloptes]

Andrei POPESCU wrote:

> As far as I can tell network manager is *not* using systemd-networkd in
> the background (it doesn't depend on systemd).

No, it doesn't but both of them use dbus and obviously can talk to each
other there.

So the advantage is when systemd does something network manager can react
and vice versa.

but I do not have systemd-networkd or -resolved installed

$ dpkg -l | grep systemd
ii  libpam-systemd:amd64241-7~deb10u3   
 
amd64system and service manager - PAM module
ii  libsystemd0:amd64   241-7~deb10u3   
 
amd64systemd utility library
ii  systemd 241-7~deb10u3   
 
amd64system and service manager
ii  systemd-sysv:i386   241-7~deb10u3   
 
i386 system and service manager - SysV links

I was going to figure out why systemd-sysv:i386 is selected over
systemd-sysv:amd64, but couldn't allocate time and motivation so far.

And BTW I can not find anything like networkd

$ apt-cache search systemd | grep network
networkd-dispatcher - Dispatcher service for systemd-networkd connection
status changes

or resolved

$ apt-cache search systemd | grep resolv
openvpn-systemd-resolved - integrates OpenVPN with systemd-resolved
resolvconf-admin - setuid helper program for setting up the local DNS
libnss-mymachines - nss module to resolve hostnames for local container
instances
libnss-resolve - nss module to resolve names via systemd-resolved

I have no idea what people are complaining of here. Shouldn't I stay at
home, I would not even read or write this ... pure coincidence.

Re: Buster without systemd?

[deloptes]

Kamil Jońca wrote:

> Of course. And I don't have to take you seriously.

Of course. This is your choice, but please stop complaining of self induced
problems and most of all stop blaming others, _please_!

and thank you in advance. We've had here enough of this systemd topic.

Re: Buster without systemd?

[Kamil Jońca]

deloptes  writes:

> Kamil Jońca wrote:
>
>>> Kamil Jońca wrote:
>>>
 Forget it. Show me how call script in reaction to DHPC events: BOUND
 RENEW REBIND TIMEOUT
 
>>>
>>> what for?
>> 
>> To convince me that systemd has enough functionality.
>> 
>
> I don't have to do anything.
>
Of course. And I don't have to take you seriously.

KJ


-- 
http://stopstopnop.pl/stop_stopnop.pl_o_nas.html

Re: Buster without systemd?

[tomas]

On Wed, Mar 25, 2020 at 11:19:22AM +0200, Andrei POPESCU wrote:
> On Ma, 24 mar 20, 10:23:59, Charles Curley wrote:

[...]

> > It is positively Balmeresque in its insistence on metastasizing all over
> > your computer [...]

> It seems to me like you are comparing a FLOSS project with a company 
> that used its dominant market position to squash competition and a 
> terminal disease.

Agreed. I didn't like that either. Whatever the systemd team is doing,
and whether I like it or not -- they're doing free software. Attributes
like "Balmeresque" are doubly insulting, "metastasize" are insulting
(ironically copying Balmer's rhethoric -- remember "Linux is a cancer"?).

Those folks are doing free software. By all means criticize what they
are doing (or, even better: do something yourself), but FFS, take that
into account!

> This would be very insulting to those FLOSS developers, as well as all 
> developers of other FLOSS projects that *chose* to depend on systemd (or 
> parts of it), so I must surely be misunderstanding something.

Thanks. Sometimes I get tired of blowing into that horn. Don't want
systemd? Don't use it. And contribute to alternatives?

And be polite to those who do like systemd.

(And vice-versa, of course).

Cheers
-- t



signature.asc
Description: Digital signature

Re: Buster without systemd?

[Andrei POPESCU]

On Ma, 24 mar 20, 10:23:59, Charles Curley wrote:
> On Tue, 24 Mar 2020 08:30:03 -0500
> Dave Sherohman  wrote:
> 
> > I expect that there would be much, much less resistance to systemd if
> > that were all that it does.  However, it has evolved to take over a
> > number of other functions aside from managing the order that services
> > are started, which makes many people uncomfortable with it.
> 
> It is positively Balmeresque in its insistence on metastasizing all over
> your computer, and in its insistence on being the One True Solution to
> whatever your problem may be.

It seems to me like you are comparing a FLOSS project with a company 
that used its dominant market position to squash competition and a 
terminal disease.

This would be very insulting to those FLOSS developers, as well as all 
developers of other FLOSS projects that *chose* to depend on systemd (or 
parts of it), so I must surely be misunderstanding something.

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Andrei POPESCU]

On Ma, 24 mar 20, 20:34:24, deloptes wrote:
> 
> But the purpose of systemd is not to give you an interface. This is provided
> by each desktop. Systemd will give you the low level service management -
> AFAIK it works via dbus.
> 
> systemd <-> dbus <-> network-manager
>   |
> network-manager-interface
> 
> This is my understanding how it works or should work.

As far as I can tell network manager is *not* using systemd-networkd in 
the background (it doesn't depend on systemd).

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[deloptes]

Kamil Jońca wrote:

>> Kamil Jońca wrote:
>>
>>> Forget it. Show me how call script in reaction to DHPC events: BOUND
>>> RENEW REBIND TIMEOUT
>>> 
>>
>> what for?
> 
> To convince me that systemd has enough functionality.
> 

I don't have to do anything.

> 

 What I know, that if you have a good working DHCP server, you don't
 have to do any hacky stuff and scripts.
>>> Show me.
>>
>> why? You even didn't say "please"!
> 
> To show, that you know what are you saying.

you didn't say "please" again.

I am not your father or teacher or whatever. Do your homework by yourself.

Again: you do not describe your use case. Just demanding that it works the
way you think it should work. I bet you are not the smartest on the planet
Earth. It is up to you what and how you do things.
For myself I have found the perfect combination - no systemd on the servers
and since end of last year I migrated notebook and desktop to systemd.
Recently I also started migrating to UEFI, but it is another topic.

What matters is doing a good research, a good plan and executing the plan.
People (like you) usually don't bother to invest time in this planning
phase. Well ... it's not my business.
The point here was that you do not have to blame something if you do not
like or understand it. The time blaming systemd is over. I've not heard
recently of major issues caused by systemd. IMO it is already in the stable
phase ... but well you can't be sure, so do backups and tests.

regards

Re: Buster without systemd?

[Kamil Jońca]

deloptes  writes:

> Kamil Jońca wrote:
>
>> Forget it. Show me how call script in reaction to DHPC events: BOUND
>> RENEW REBIND TIMEOUT
>> 
>
> what for?

To convince me that systemd has enough functionality.


>>>
>>> What I know, that if you have a good working DHCP server, you don't have
>>> to do any hacky stuff and scripts.
>> Show me.
>
> why? You even didn't say "please"!

To show, that you know what are you saying.

KJ


-- 
http://wolnelektury.pl/wesprzyj/teraz/

Re: Buster without systemd?

[deloptes]

Kamil Jońca wrote:

> Forget it. Show me how call script in reaction to DHPC events: BOUND
> RENEW REBIND TIMEOUT
> 

what for?

>>
>> What I know, that if you have a good working DHCP server, you don't have
>> to do any hacky stuff and scripts.
> Show me.

why? You even didn't say "please"!

Re: Buster without systemd?

[deloptes]

David Wright wrote:

> I don't want a desktop. In fact, wicd doesn't even need X, as it can
> run quite happily on a VC to configure a new AP.
> 

I did not tell you what you want. You just complained there is no interface.

> When I return to somewhere I have been before, wicd (the daemon)
> usually connects before I have typed my passphrase to unlock /home.
> (That assumes I'm logging in.)

I gave up on the Gnome spooks many years ago. But AFAIK wicd is the same or
similar as network-manager - well the GNome/GTK way of doing things.

Re: Buster without systemd?

[Kamil Jońca]

deloptes  writes:

> Kamil Jońca wrote:
>
>> Any hint? I tried (but not very hard, to be honest) to migrate my
>> if-up/if-down/dhclient-hooks scripts to systemd but without succes.
>> For example how to get "domains" from dhcp options and create proper
>> *.network file with "Domains=" setting?
>
> Don't know what you are talking about. I've seen a lot of hacky stuff. The
> best thing is to remove it and replace it with something new.
>
> I don't know your use case or better said. Place your use case here and
> listen to the wisdom of the others.

Forget it. Show me how call script in reaction to DHPC events: BOUND
RENEW REBIND TIMEOUT

>
> What I know, that if you have a good working DHCP server, you don't have to
> do any hacky stuff and scripts.
Show me.

KJ

-- 
http://stopstopnop.pl/stop_stopnop.pl_o_nas.html

Re: Buster without systemd?

[David Wright]

On Tue 24 Mar 2020 at 20:34:24 (+0100), deloptes wrote:
> David Wright wrote:
> 
> > I looked at what documentation I could find, but carried on using
> > wicd-curses, and the arch wiki seems to agree with what I found:
> > 
> > "systemd-networkd does not have a proper interactive management
> > interface neither via command-line nor graphical".
> > 
> > The interface is what matters when you're travelling with a laptop.
> 
> But the purpose of systemd is not to give you an interface. This is provided
> by each desktop. Systemd will give you the low level service management -
> AFAIK it works via dbus.
> 
> systemd <-> dbus <-> network-manager
>   |
> network-manager-interface
> 
> This is my understanding how it works or should work.

I don't want a desktop. In fact, wicd doesn't even need X, as it can
run quite happily on a VC to configure a new AP.

When I return to somewhere I have been before, wicd (the daemon)
usually connects before I have typed my passphrase to unlock /home.
(That assumes I'm logging in.)

Cheers,
David.

Re: Buster without systemd?

[deloptes]

Kamil Jońca wrote:

> Any hint? I tried (but not very hard, to be honest) to migrate my
> if-up/if-down/dhclient-hooks scripts to systemd but without succes.
> For example how to get "domains" from dhcp options and create proper
> *.network file with "Domains=" setting?

Don't know what you are talking about. I've seen a lot of hacky stuff. The
best thing is to remove it and replace it with something new.

I don't know your use case or better said. Place your use case here and
listen to the wisdom of the others.

What I know, that if you have a good working DHCP server, you don't have to
do any hacky stuff and scripts.
It takes time to do things right, but when done, they last forever.

Re: Buster without systemd?

[deloptes]

Kamil Jońca wrote:

> Personal attack detected. Have you anything interesting to say?
> 

Not at all, because I was writing obviously to Ansgar.

> Yes, it is possible that I issue "aptitude update; aptitude
> safe-upgrade", and accept new dependecies without deeper look, but
> that's all. There were NO quuestions, to which I had to answer.
> 
> Of course you do not need to believe me. And of course you can believe
> that were no bugs in installer/updater.

Anyway - do lessons learn next time. Relax, make cup of coffee and go
through the release notes.
If you are bothered to do so, just do a backup, but please do never complain
about a mess you did to yourself with finger pointing to others.

And believe me few years ago I was also not amused by systemd, but not for
what it, but how buggy it was. It should not have been enforced in that
state. Luckily it works better now ... well at least for the size of the
code it is working amazingly well.

Re: Buster without systemd?

[Kamil Jońca]

deloptes  writes:

[...]
>
> AFAIK everything is possible, but the systemd-way. Learn the way to walk it.
> Most probably you will end up with much better solution.

Any hint? I tried (but not very hard, to be honest) to migrate my
if-up/if-down/dhclient-hooks scripts to systemd but without succes.
For example how to get "domains" from dhcp options and create proper
*.network file with "Domains=" setting?

>
> regards

-- 
http://stopstopnop.pl/stop_stopnop.pl_o_nas.html

Re: Buster without systemd?

[Kamil Jońca]

deloptes  writes:

> Ansgar wrote:
>
>> Because it doesn't.
>> 
>>> When I realised existence of it, I turned it off immediately.
>> 
>> Then you must have enabled it first.
>> 
>
> Ansagar - you know this mentality "Next -> Next" and never read anything
> what is written on screen ... I know it from Windows ;-)

Personal attack detected. Have you anything interesting to say?

Yes, it is possible that I issue "aptitude update; aptitude
safe-upgrade", and accept new dependecies without deeper look, but
that's all. There were NO quuestions, to which I had to answer.

Of course you do not need to believe me. And of course you can believe
that were no bugs in installer/updater. 
KJ


-- 
http://stopstopnop.pl/stop_stopnop.pl_o_nas.html

Re: Buster without systemd?

[deloptes]

Kamil Jońca wrote:

> I know. :) I have had dnsmasq (+ some if-up/if-down scripts) configured
> that way.
> This was simple example why I do not like integration several things
> into systemd.
> 

First of all I want to highlight thatthere should be always lessons learned.
Next time you plan major upgrade better and find answers to those questions
before the upgrade happens.
Second you usually test things before rolling out to production (this can be
simple VM or alike, or backup before doing upgrade)

> BTW. Can plug scripts into systemd network events? (interface up/down,
> ipsec tunnel up/down and so on).

AFAIK everything is possible, but the systemd-way. Learn the way to walk it.
Most probably you will end up with much better solution.

regards

Re: Buster without systemd?

[deloptes]

Andrei POPESCU wrote:

> [1] didn't bother to investigate whether these are using 'zeroconf' or
> some other similar technology, they just work.

No they do not. It took me one whole day to find out why the networking was
not working between a HP switch and a RHEL server with LACP enabled.
Found out the guy that installed it did install the default which included
avahi and who knows what else.
In anycase avahi was trying to configure the LACP link which in result was
causing the link to flip-flop between the interfaces.

Always remove the crap!

The other crap you mentioned - the Android - you can also remove. 

Amen!

Re: Buster without systemd?

[deloptes]

Ansgar wrote:

> Because it doesn't.
> 
>> When I realised existence of it, I turned it off immediately.
> 
> Then you must have enabled it first.
> 

Ansagar - you know this mentality "Next -> Next" and never read anything
what is written on screen ... I know it from Windows ;-)

People jump without knowing how high it is - simply amazing

Re: Buster without systemd?

[deloptes]

Kamil Jońca wrote:

> As I said: I did not know that systemd-resolved install with some
> upgrade, and then takes my dns resolving. When I realised existence of
> it, I turned it off immediately.

why would I use a tool without reading the manual?!
I always admire my wife :)
Good that I am reading the manuals for her.

Re: Buster without systemd?

[deloptes]

David Wright wrote:

> I looked at what documentation I could find, but carried on using
> wicd-curses, and the arch wiki seems to agree with what I found:
> 
> "systemd-networkd does not have a proper interactive management
> interface neither via command-line nor graphical".
> 
> The interface is what matters when you're travelling with a laptop.

But the purpose of systemd is not to give you an interface. This is provided
by each desktop. Systemd will give you the low level service management -
AFAIK it works via dbus.

systemd <-> dbus <-> network-manager
  |
network-manager-interface

This is my understanding how it works or should work.

Re: Buster without systemd?

[Joe]

On Tue, 24 Mar 2020 10:44:54 -0500
David Wright  wrote:


> 
> The interface is what matters when you're travelling with a laptop.
> 
[Looks around carefully for signs of a thunderstorm] Don't laugh,
but I've found Network Manager on my mobiles quite useful and reliable
for handling wired, wifi and VPN connections. For the last few years,
anyway, after it stopped being Notwork Manager.

-- 
Joe

Re: Buster without systemd?

[deloptes]

Reco wrote:

> Actually, deloptes is right and you're not.
> 
> Kernel's documentation - [1] - describes a way to force a kernel itself
> to configure a network interface for IPv4, be it static or DHCP, along
> with the primitive routing table (default gw at most).
> 
> Taking IPv6's RA into the account, one does not need anything but the
> working kernel on the client side to get IPv6 and a primitive routing
> table.
> 
> Reco
> 
> [1]
>
[/usr/share/doc/linux-doc-4.19/Documentation/filesystems/nfs/nfsroot.txt.gz

Thank you Reco,
you obsoleted my reply. I will just add few things.
I and many more have been using diskless boot and nfsroot even before the
time systemd emerged.
What I find good is the dynamic management of services -  I do not remember
where I read this, but it was a good article about the need to move forward
from old init. This enlightened me. Thinking over, I came to the conclusion
that the people are right about this need. It seems only the way they did
it was wrong. However, thanks Debian, no one was forced to do anything.
At the end people still complain (my opinion) not because systemd is not
working well, but because they have to learn something new.
I still do not understand why RedHat should use it on the servers as I
personally do not see much of advantage, but it just works, so why not.

Re: Buster without systemd?

[Kenneth Parker]

Thanks.  Star added.

Kenneth Parker

On Tue, Mar 24, 2020, 2:57 PM deloptes  wrote:

> Kenneth Parker wrote:
>
> > On Tue, Mar 24, 2020, 10:12 AM deloptes  wrote:
> >
> >> Kamil Jońca wrote:
> >>
> >> > But always try :) systemd-resolved also.
> >> >
> >> > I lost a lot of time struggling with systemd-resoved  (I wasn't aware
> >> > of its existence), guessing why my dnsmasq stopped working properly.
> >> >
> >> > I simply like unix approach ("do one thing") and systemd is against
> it.
> >>
> >> I think you just did not want to read the documentation before trying
> it.
> >>
> >> After I was reading about systemd for almost an year, I decided to try
> >> it. No issues. See also my previous post
> >>
> >
> > I would love it, if some detailed Books were published on systemd.  I
> > would happily purchase books like "SystemD for Dummies", "The SystemD
> > Bible", or "SystemD in a Nutshell". But Google still shows a blank on any
> > of these.
> >
> > Kenneth Parker
> >
> >>
>
> these are some of the links I get for "systemd how to"
>
> https://wiki.archlinux.org/index.php/systemd
> https://www.linux.com/training-tutorials/understanding-and-using-systemd/
>
> https://www.digitalocean.com/community/tutorials/systemd-essentials-working-with-services-units-and-the-journal
> https://www.golinuxcloud.com/beginners-guide-systemd-tutorial-linux/
>
>

Re: Buster without systemd?

[Kamil Jońca]

Ansgar  writes:

> kjo...@poczta.onet.pl (Kamil Jońca) writes:
>> deloptes  writes:
>>> Kamil Jońca wrote:
 But always try :) systemd-resolved also.
 
 I lost a lot of time struggling with systemd-resoved  (I wasn't aware of
 its existence), guessing why my dnsmasq stopped working properly.
 
 I simply like unix approach ("do one thing") and systemd is against it.
>>>
>>> I think you just did not want to read the documentation before trying it.
>> As I said: I did not know that systemd-resolved install with some
>> upgrade, and then takes my dns resolving.
>
> Because it doesn't.

I am afraid it was.

>> When I realised existence of it, I turned it off immediately.
>
> Then you must have enabled it first.

No, at least intentionally.

>
>> And my task:
>>
>> How can I configure resolved to do something like:
>> eth - interface
>> tun1 - interface to vpn1 with  domains a, b, c
>> tun2 - interface to vpn2 with domains d, e
>>
>> I want to resolve domains a,b,c with dns server in vpn1 and domains d,e
>> with server in vpn2.
>>
>> Can I do it with resolved?
>
> +---
> | Domains=: A whitespace-separated list of domains which should be
> | resolved using the DNS servers on this link.
> +---
What are you cited?
https://www.freedesktop.org/software/systemd/man/resolved.conf.html#
does not have this.
KJ


-- 
http://stopstopnop.pl/stop_stopnop.pl_o_nas.html

Re: Buster without systemd?

[Kamil Jońca]

Greg Wooledge  writes:

> On Tue, Mar 24, 2020 at 06:03:47PM +0100, Kamil Jońca wrote:
>> How can I configure resolved to do something like:
>> eth - interface
>> tun1 - interface to vpn1 with  domains a, b, c
>> tun2 - interface to vpn2 with domains d, e
>> 
>> I want to resolve domains a,b,c with dns server in vpn1 and domains d,e
>> with server in vpn2.
>> 
>> Can I do it with resolved? 
>
> What you want to do would be best solved by a DNS resolver that you
> configure to forward requests to various places based on the requested
> domain.  Any of the major DNS server software packages should be able to
> do it.

I know. :) I have had dnsmasq (+ some if-up/if-down scripts) configured
that way. 
This was simple example why I do not like integration several things
into systemd.

BTW. Can plug scripts into systemd network events? (interface up/down,
ipsec tunnel up/down and so on).
KJ






-- 
http://stopstopnop.pl/stop_stopnop.pl_o_nas.html

Re: Buster without systemd?

[deloptes]

Kenneth Parker wrote:

> On Tue, Mar 24, 2020, 10:12 AM deloptes  wrote:
> 
>> Kamil Jońca wrote:
>>
>> > But always try :) systemd-resolved also.
>> >
>> > I lost a lot of time struggling with systemd-resoved  (I wasn't aware
>> > of its existence), guessing why my dnsmasq stopped working properly.
>> >
>> > I simply like unix approach ("do one thing") and systemd is against it.
>>
>> I think you just did not want to read the documentation before trying it.
>>
>> After I was reading about systemd for almost an year, I decided to try
>> it. No issues. See also my previous post
>>
> 
> I would love it, if some detailed Books were published on systemd.  I
> would happily purchase books like "SystemD for Dummies", "The SystemD
> Bible", or "SystemD in a Nutshell". But Google still shows a blank on any
> of these.
> 
> Kenneth Parker
> 
>>

these are some of the links I get for "systemd how to"

https://wiki.archlinux.org/index.php/systemd
https://www.linux.com/training-tutorials/understanding-and-using-systemd/
https://www.digitalocean.com/community/tutorials/systemd-essentials-working-with-services-units-and-the-journal
https://www.golinuxcloud.com/beginners-guide-systemd-tutorial-linux/

Re: Buster without systemd?

[Sven Joachim]

On 2020-03-24 17:22 +0200, Andrei POPESCU wrote:

> In buster many things are indeed disabled by default, however this will
> likely change for bullseye.
>
> I believe the persistent journal is already enabled by default now

Indeed, but this is easy to change and does not do much harm, except
taking up disk space.

> (it used to forward everything to rsyslog before).

If rsyslog is installed, it still does that.  Whether rsyslog will be
installed by default in bullseye is not yet clear, but on upgrades it
will certainly remain on the system.

Cheers,
   Sven

Re: Buster without systemd?

[Ansgar]

kjo...@poczta.onet.pl (Kamil Jońca) writes:
> deloptes  writes:
>> Kamil Jońca wrote:
>>> But always try :) systemd-resolved also.
>>> 
>>> I lost a lot of time struggling with systemd-resoved  (I wasn't aware of
>>> its existence), guessing why my dnsmasq stopped working properly.
>>> 
>>> I simply like unix approach ("do one thing") and systemd is against it.
>>
>> I think you just did not want to read the documentation before trying it.
> As I said: I did not know that systemd-resolved install with some
> upgrade, and then takes my dns resolving.

Because it doesn't.

> When I realised existence of it, I turned it off immediately.

Then you must have enabled it first.

> And my task:
>
> How can I configure resolved to do something like:
> eth - interface
> tun1 - interface to vpn1 with  domains a, b, c
> tun2 - interface to vpn2 with domains d, e
>
> I want to resolve domains a,b,c with dns server in vpn1 and domains d,e
> with server in vpn2.
>
> Can I do it with resolved?

+---
| Domains=: A whitespace-separated list of domains which should be
| resolved using the DNS servers on this link.
+---

sounds like it might be what you want.

Ansgar

Re: Buster without systemd?

[Greg Wooledge]

On Tue, Mar 24, 2020 at 06:03:47PM +0100, Kamil Jońca wrote:
> How can I configure resolved to do something like:
> eth - interface
> tun1 - interface to vpn1 with  domains a, b, c
> tun2 - interface to vpn2 with domains d, e
> 
> I want to resolve domains a,b,c with dns server in vpn1 and domains d,e
> with server in vpn2.
> 
> Can I do it with resolved? 

What you want to do would be best solved by a DNS resolver that you
configure to forward requests to various places based on the requested
domain.  Any of the major DNS server software packages should be able to
do it.

I don't count systemd-resolved among those.

Note that forwarding of requests means that you send the request to a
DNS resolver/server by IP address.  Not by network interface name.  So,
it's conceivable that you would set up *multiple* custom DNS resolvers --
one that's visible on the LAN to receive the client requests, and one
for each of these tunnel interfaces to have the requests forwarded to
them, and then proceed normally.

A DNS-focused mailing list might be a next step, if you can't figure it
out from my horribly mangled attempt to describe the infrastructure.

Re: Buster without systemd?

[Kamil Jońca]

deloptes  writes:

> Kamil Jońca wrote:
>
>> But always try :) systemd-resolved also.
>> 
>> I lost a lot of time struggling with systemd-resoved  (I wasn't aware of
>> its existence), guessing why my dnsmasq stopped working properly.
>> 
>> I simply like unix approach ("do one thing") and systemd is against it.
>
> I think you just did not want to read the documentation before trying it.
As I said: I did not know that systemd-resolved install with some
upgrade, and then takes my dns resolving. When I realised existence of
it, I turned it off immediately.
And my task:

How can I configure resolved to do something like:
eth - interface
tun1 - interface to vpn1 with  domains a, b, c
tun2 - interface to vpn2 with domains d, e

I want to resolve domains a,b,c with dns server in vpn1 and domains d,e
with server in vpn2.

Can I do it with resolved? 

KJ

-- 
http://stopstopnop.pl/stop_stopnop.pl_o_nas.html

Re: Buster without systemd?

[Gene Heskett]

On Tuesday 24 March 2020 09:38:08 David Wright wrote:

> On Tue 24 Mar 2020 at 06:42:44 (-0400), Gene Heskett wrote:
> > On Tuesday 24 March 2020 06:04:10 Andrei POPESCU wrote:
> > > On Lu, 23 mar 20, 18:42:49, Gene Heskett wrote:
> > > > Thats a problem I don't have Greg. I went to a locally defined
> > > > hosts file 30+ years ago for all my private resolutions, and it
> > > > Just Works. Queries that go out on the wire for resolution are
> > > > relayed to the dns services of my provider. Resolution times for
> > > > external sites are sub 100 millisecond as a general rule. Thats
> > > > ALL handled by my router running dd-wrt which I think is using
> > > > dnsmasq.
> > > >
> > > > Yet every time I promote such a structure as a solution to
> > > > someones local network problems I am the idiot according to you
> > > > for not using dhcp globally. I don't enjoy being painted as an
> > > > a-hole for using something that Just Works.  So I've quit unless
> > > > you or someone like you pulls my chain.
> > >
> > > Static IPs and /etc/hosts works just fine here with
> > > systemd-networkd / systemd-resolved and Network Manager.
> > >
> > > > For the "vast majority of system's", as I install them, the
> > > > first thing I have to do is uninstall that stuff as it has yet
> > > > in the last 5 years, to use anything in the routing table but
> > > > the avahi supplied 169.xx.cc.nn address which is not allowed off
> > > > the premises by dd-wrt.
> > >
> > > [citation needed]
> >
> > My own posting about it in just the past year, but that posting was
> > only done after the problem was solved. As its a bit hard to post
> > from the machine that has a bogus routing setup. With dhcpd5 (I
> > think thats the right name) in the system it does no good to put the
> > correct route as a gateway statement in your /e/n/interfaces tree.
>
> Hm, you thought you'd probably fixed the problem IIRC while composing
> your post of Sat, 6 Jul 2019 21:38:10 -0400, after a couple of days of
> back-and-forth here, and after being asked to move the discussion here
> from a different list.
>
> > So that and avahi generally get nuked.
>
> Most of us avoid installing packages, rather than nuking them.

Installed by default, and the defaults are WRONG.

> Oh, but I almost forgot, it wasn't a Debian system in any case.
>
> > IMNSHO both were written to screw up networking,
> > their only possible reason-de-tere.
>
> Shame on you, sir.

Maybe so, but lots of this could be moved to debian IF debian could be so 
kind as to supply us with a buildable preempt-rt kernel that works as 
well as the default kernel does but many microseconds quicker. Raspbian 
does at least supply the src code although they do not otherwise support 
it. Either that, or help the RTAI guy keep up with the changeing 
landscape between cpu architectures.  That would be even better.

> Cheers,
> David.


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: Buster without systemd?

[Charles Curley]

On Tue, 24 Mar 2020 08:30:03 -0500
Dave Sherohman  wrote:

> I expect that there would be much, much less resistance to systemd if
> that were all that it does.  However, it has evolved to take over a
> number of other functions aside from managing the order that services
> are started, which makes many people uncomfortable with it.

It is positively Balmeresque in its insistence on metastasizing all over
your computer, and in its insistence on being the One True Solution to
whatever your problem may be.


-- 
Does anybody read signatures any more?

https://charlescurley.com
https://charlescurley.com/blog/

Re: Buster without systemd?

[Gene Heskett]

On Tuesday 24 March 2020 08:47:32 David wrote:

> On Tue, 24 Mar 2020 at 20:57, Gene Heskett  
wrote:
> > On Tuesday 24 March 2020 04:38:10 to...@tuxteam.de wrote:
> > > On Mon, Mar 23, 2020 at 04:09:14PM -0400, Gene Heskett wrote:
> > > > On Monday 23 March 2020 15:16:33 Joe wrote:
> > >
> > > One basic rule of science and engineering should read "if you want
> > > to criticise something, you better learn first how it works".
> >
> > Which I've not found the docs that would teach me that yet. If such
> > actually exist, where do I find the top of that man tree?
>
> Maybe you missed this previous message:
> https://lists.debian.org/debian-user/2019/06/msg00277.html

I faintly recall it going by but the first screens text was TL;DR and 
looking at it again its still too long to get to the subject.  My fault 
of course...

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: Buster without systemd?

[Klaus Singvogel]

David Wright wrote:
> On Mon 23 Mar 2020 at 20:27:31 (+0100), Klaus Singvogel wrote:
> > My reasons:
[…]
> 
> > My computer "hickups" for 20 seconds at boot. With systemd I don't have
> > the slighest idea which process delays, as several processes start after
> > the delay at once. I had the feeling with Init scripts it was more obvious
> > to see which "startup file" hangs.
> 
> $ systemd-analyze critical-chain
> 
> Is that any help in finding what's waiting for what?

graphical.target @23.269s
└─multi-user.target @23.269s
  └─exim4.service @23.095s +174ms
└─network-online.target @23.092s
  └─network.target @23.086s
└─networking.service @1.383s +21.702s
  └─ifupdown-pre.service @285ms +1.096s
└─systemd-udev-trigger.service @230ms +54ms
  └─systemd-udevd-control.socket @228ms
└─system.slice @218ms
  └─-.slice @218ms

At least I found a hint where to start: networking.service

> > But some important messages are not found in syslog nor messages.
> 
> Which ones are missing?

When daemons are started (at boot) and how long the startup took for example.

> > No easy way to find out: if and what fails. Example: After months I found
> > out that mlocate doesn't update it's database anymore: previously updates
> > were done by cron, but the distri moved to systemd's way (timer) and by
> > default the updatedb was disabled.
> 
> I've not touched the default. What did you have to change to make it run?

Well, this time it was only systemd related, but not Debian. I'm also
running SUSE Linux, and it happened there.

Regards,
Klaus.
-- 
Klaus Singvogel
GnuPG-Key-ID: 1024R/5068792D  1994-06-27

Re: Buster without systemd?

[David Wright]

On Tue 24 Mar 2020 at 14:49:03 (+0100), deloptes wrote:
> Andrei POPESCU wrote:
> 
> > systemd-networkd is not meant to do very complicated configurations.
> > 
> > Why should I install an additional package (e.g. ifupdown, Network
> > Manager, etc.) just for setting up a static IP?
> 
> well - for static IP you don't need systemd, but it is a relief for the
> dynamic stuff - i.e. wireless and cabled networks that change.
> 
> I find it amazing how it works  not that I say I am starting to love
> systemd :) but for the couple of months I am using it on the desktop it
> seems to work fine
> 
> On the servers as mentioned before it is of no use for now.

I looked at what documentation I could find, but carried on using
wicd-curses, and the arch wiki seems to agree with what I found:

   "systemd-networkd does not have a proper interactive management
interface neither via command-line nor graphical".

The interface is what matters when you're travelling with a laptop.

Cheers,
David.

Re: Buster without systemd?

[Kenneth Parker]

On Tue, Mar 24, 2020, 10:12 AM deloptes  wrote:

> Kamil Jońca wrote:
>
> > But always try :) systemd-resolved also.
> >
> > I lost a lot of time struggling with systemd-resoved  (I wasn't aware of
> > its existence), guessing why my dnsmasq stopped working properly.
> >
> > I simply like unix approach ("do one thing") and systemd is against it.
>
> I think you just did not want to read the documentation before trying it.
>
> After I was reading about systemd for almost an year, I decided to try it.
> No issues. See also my previous post
>

I would love it, if some detailed Books were published on systemd.  I would
happily purchase books like "SystemD for Dummies", "The SystemD Bible", or
"SystemD in a Nutshell". But Google still shows a blank on any of these.

Kenneth Parker

>

Re: Buster without systemd?

[Reco]

Hi.

On Tue, Mar 24, 2020 at 05:05:27PM +0200, Andrei POPESCU wrote:
> On Ma, 24 mar 20, 14:49:03, deloptes wrote:
> > Andrei POPESCU wrote:
> > 
> > > systemd-networkd is not meant to do very complicated configurations.
> > > 
> > > Why should I install an additional package (e.g. ifupdown, Network
> > > Manager, etc.) just for setting up a static IP?
> > 
> > well - for static IP you don't need systemd
> 
> Well, "something" has to bring up the interface with the correct IP and 
> set up the default route, even if it's "just" a script calling 'ip', 
> which I would then have to manually plug-in somewhere in the boot 
> process.

Actually, deloptes is right and you're not.

Kernel's documentation - [1] - describes a way to force a kernel itself
to configure a network interface for IPv4, be it static or DHCP, along
with the primitive routing table (default gw at most).

Taking IPv6's RA into the account, one does not need anything but the
working kernel on the client side to get IPv6 and a primitive routing
table.

Reco

[1] /usr/share/doc/linux-doc-4.19/Documentation/filesystems/nfs/nfsroot.txt.gz

Re: Buster without systemd?

[Andrei POPESCU]

On Lu, 23 mar 20, 13:15:13, Greg Wooledge wrote:
> 
> 4) It does way more than just replace /sbin/init.  It has components
>that try to bring up network interfaces, implement DNS resolution,
>implement NTP, implement cron-like functionality, and so on.  Many
>people feel this is over-reaching.  (And to be fair, Debian disables
>most of these optional subsystems by default.)

In buster many things are indeed disabled by default, however this will 
likely change for bullseye.

I believe the persistent journal is already enabled by default now (it 
used to forward everything to rsyslog before).

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Patrick Bartek]

On Tue, 24 Mar 2020 00:04:18 +
Michael Howard  wrote:

> On 23/03/2020 23:20, Patrick Bartek wrote:
> > On Mon, 23 Mar 2020 15:15:14 +
> > Michael Howard  wrote:
> >  
> >> On 23/03/2020 14:28, Patrick Bartek wrote:  
> >>> On Mon, 23 Mar 2020 08:37:47 -0400
> >>> Jude DaShiell  wrote:
> >>> 
>  There is devuan-ascii 2.x but I don't know its equivalent to buster.
>  That system maintained sysv and still has support.  
> >>> ASCII is Stretch without systemd or any of the dependencies.  Devuan
> >>> hasn't release its "Buster" version yet.
> >>>
> >>> B
> >>> 
> >> It can be upgraded to 'beowulf' using apt of course.  
> > I did that some months ago with Ascii running in Virtualbox.  And it did
> > work, but there were enough problems then that it wasn't suitable
> > for general use -- just evaluation. Perhaps there have been
> > improvements since then, but I'll wait for Beowulf to be officially
> > released before checking it out again.  Currently, I'll stick with
> > Stretch running sysvinit until long term support for ceases.
> >
> > B
> >  
> I guess it depends exactly what is installed but I have upgraded
> via this route a number of times.

I wasn't implying that it wouldn't work.  It's just when I did it
(Fall 2019) Beowulf was less mature and had problems.  I wasn't
expecting it to be stable quality.  I just wanted to evaluate it since
I had problems with Buster trying to keep systemd's init from
wiping out sysvinit and reinstalling itself everytime I installed
something.  Infuriating!

FWIW: My Ascii test install in Virtualbox mirrored my Stretch install:
Basic terminal plus system utilities, X, a window manager (Openbox),
single LXPanel, etc.  No bells and whistles or eye candy. KISS. So, its
upgrade to Beowulf resulted in the same.

B

Re: Buster without systemd?

[Andrei POPESCU]

On Ma, 24 mar 20, 14:49:03, deloptes wrote:
> Andrei POPESCU wrote:
> 
> > systemd-networkd is not meant to do very complicated configurations.
> > 
> > Why should I install an additional package (e.g. ifupdown, Network
> > Manager, etc.) just for setting up a static IP?
> 
> well - for static IP you don't need systemd

Well, "something" has to bring up the interface with the correct IP and 
set up the default route, even if it's "just" a script calling 'ip', 
which I would then have to manually plug-in somewhere in the boot 
process.

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Andrei POPESCU]

On Ma, 24 mar 20, 14:54:15, deloptes wrote:
> Andrei POPESCU wrote:
> 
> > Zeroconf works fine when configured correctly :P
> 
> and why should I use zero conf?

You? I have no idea :)

As far as I'm concerned I do appreciate that the remote control app for 
my Android box can connect without any configuration. Receiving a 
Youtube link on my phone and easily playing it on the Android box 
(hopefully Kodi on Debian in a not so distant future) is also very 
nice[1].

The same would also work for guests, provided I grant them access to my 
wireless network.

I can't think of a straightforward way to do the same from my Debian 
laptop.

On the other hand I tend to remove / prevent installation of avahi / 
libnss-mdsn on my Debian boxes (currently installed on my laptop, but 
not used as far as I can tell).

[1] didn't bother to investigate whether these are using 'zeroconf' or 
some other similar technology, they just work.

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[David Wright]

On Tue 24 Mar 2020 at 14:19:34 (+0100), to...@tuxteam.de wrote:
> On Tue, Mar 24, 2020 at 08:55:32AM -0400, rhkra...@gmail.com wrote:
> > On Tuesday, March 24, 2020 04:38:10 AM to...@tuxteam.de wrote:
> > > There are valid reasons for systemd's binary format. Space savings
> > > is very far off the top of the list (probably it isn't on that list
> > > at all).
> > 
> > Can you (or someone else) elaborate on that a little?
> 
> Strictly defined structure instead of parsing. Indexing, so you can
> easily cross-reference logs. Support for cryptographic integrity
> (think blockchain).
> 
> You might think these features are worth the complexity or they
> aren't, but assuming the designers are idiots is wrong on more than
> one level.
> 
> > (Aside: I don't (yet -- or at least not intentionally) use systemd (well, 
> > or 
> > have any trouble / interaction with it -- I suppose it might have been 
> > installed on either my Jessie or Buster installs by default, and I haven't 
> > had 
> > to dig into those logs, so don't know if they're readable.)
> 
> I can't check right now, but AFAIK Debian's default config forwards the
> binary logs to something logging in text, to be backwards compatible
> (to oldish sysadmins, like me ;-)

Yes, AFAIK that's what my logging is relying on, in which journalctl
only handles the current boot, and text logs store everything.
And AIUI in order to change over to all-binary logging, you have to
create the directory for it: /var/log/journal.

Which is why I don't understand the moaning about systemd's binary logs:
AFAICT, you asked for it, you got it. But what do I know; I just run
plain vanilla Debian.

Cheers,
David.

Re: Buster without systemd?

[deloptes]

Andrei POPESCU wrote:

> Zeroconf works fine when configured correctly :P

and why should I use zero conf?

I hate those Red Hat servers that come up like Windows crap with zero conf
and avahi.

If you ask me those two are even worse than systemd

Re: Buster without systemd?

[deloptes]

Kamil Jońca wrote:

> But always try :) systemd-resolved also.
> 
> I lost a lot of time struggling with systemd-resoved  (I wasn't aware of
> its existence), guessing why my dnsmasq stopped working properly.
> 
> I simply like unix approach ("do one thing") and systemd is against it.

I think you just did not want to read the documentation before trying it.

After I was reading about systemd for almost an year, I decided to try it.
No issues. See also my previous post

Re: Buster without systemd?

[deloptes]

Andrei POPESCU wrote:

> systemd-networkd is not meant to do very complicated configurations.
> 
> Why should I install an additional package (e.g. ifupdown, Network
> Manager, etc.) just for setting up a static IP?

well - for static IP you don't need systemd, but it is a relief for the
dynamic stuff - i.e. wireless and cabled networks that change.

I find it amazing how it works  not that I say I am starting to love
systemd :) but for the couple of months I am using it on the desktop it
seems to work fine

On the servers as mentioned before it is of no use for now.

Re: Buster without systemd?

[David Wright]

On Tue 24 Mar 2020 at 06:42:44 (-0400), Gene Heskett wrote:
> On Tuesday 24 March 2020 06:04:10 Andrei POPESCU wrote:
> > On Lu, 23 mar 20, 18:42:49, Gene Heskett wrote:
> > > Thats a problem I don't have Greg. I went to a locally defined hosts
> > > file 30+ years ago for all my private resolutions, and it Just
> > > Works. Queries that go out on the wire for resolution are relayed to
> > > the dns services of my provider. Resolution times for external sites
> > > are sub 100 millisecond as a general rule. Thats ALL handled by my
> > > router running dd-wrt which I think is using dnsmasq.
> > >
> > > Yet every time I promote such a structure as a solution to someones
> > > local network problems I am the idiot according to you for not using
> > > dhcp globally. I don't enjoy being painted as an a-hole for using
> > > something that Just Works.  So I've quit unless you or someone like
> > > you pulls my chain.
> >
> > Static IPs and /etc/hosts works just fine here with systemd-networkd /
> > systemd-resolved and Network Manager.
> >
> > > For the "vast majority of system's", as I install them, the first
> > > thing I have to do is uninstall that stuff as it has yet in the last
> > > 5 years, to use anything in the routing table but the avahi supplied
> > > 169.xx.cc.nn address which is not allowed off the premises by
> > > dd-wrt.
> >
> > [citation needed]
> 
> My own posting about it in just the past year, but that posting was only 
> done after the problem was solved. As its a bit hard to post from the 
> machine that has a bogus routing setup. With dhcpd5 (I think thats the 
> right name) in the system it does no good to put the correct route as a 
> gateway statement in your /e/n/interfaces tree.

Hm, you thought you'd probably fixed the problem IIRC while composing
your post of Sat, 6 Jul 2019 21:38:10 -0400, after a couple of days of
back-and-forth here, and after being asked to move the discussion here
from a different list.

> So that and avahi generally get nuked.

Most of us avoid installing packages, rather than nuking them.
Oh, but I almost forgot, it wasn't a Debian system in any case.

> IMNSHO both were written to screw up networking, 
> their only possible reason-de-tere.

Shame on you, sir.

Cheers,
David.

Re: Buster without systemd?

[Dave Sherohman]

On Mon, Mar 23, 2020 at 09:59:25AM +0100, Renato Gallo wrote:
> Rly, systemd is, between things, a way to be sure when and after what 
> something starts isn't it ?

I expect that there would be much, much less resistance to systemd if
that were all that it does.  However, it has evolved to take over a
number of other functions aside from managing the order that services
are started, which makes many people uncomfortable with it.

-- 
Dave Sherohman

Re: Buster without systemd?

[tomas]

On Tue, Mar 24, 2020 at 08:55:32AM -0400, rhkra...@gmail.com wrote:
> On Tuesday, March 24, 2020 04:38:10 AM to...@tuxteam.de wrote:
> > There are valid reasons for systemd's binary format. Space savings
> > is very far off the top of the list (probably it isn't on that list
> > at all).
> 
> Can you (or someone else) elaborate on that a little?

Strictly defined structure instead of parsing. Indexing, so you can
easily cross-reference logs. Support for cryptographic integrity
(think blockchain).

You might think these features are worth the complexity or they
aren't, but assuming the designers are idiots is wrong on more than
one level.

> (Aside: I don't (yet -- or at least not intentionally) use systemd (well, or 
> have any trouble / interaction with it -- I suppose it might have been 
> installed on either my Jessie or Buster installs by default, and I haven't 
> had 
> to dig into those logs, so don't know if they're readable.)

I can't check right now, but AFAIK Debian's default config forwards the
binary logs to something logging in text, to be backwards compatible
(to oldish sysadmins, like me ;-)

Cheers
-- t
> 


signature.asc
Description: Digital signature

Re: Buster without systemd?

[rhkramer]

On Tuesday, March 24, 2020 04:38:10 AM to...@tuxteam.de wrote:
> There are valid reasons for systemd's binary format. Space savings
> is very far off the top of the list (probably it isn't on that list
> at all).

Can you (or someone else) elaborate on that a little?

(Aside: I don't (yet -- or at least not intentionally) use systemd (well, or 
have any trouble / interaction with it -- I suppose it might have been 
installed on either my Jessie or Buster installs by default, and I haven't had 
to dig into those logs, so don't know if they're readable.)

Re: Buster without systemd?

[Kamil Jońca]

Andrei POPESCU  writes:

> On Lu, 23 mar 20, 19:51:18, Kamil Jońca wrote:
>> But now systemd wants:
>> - manage network interfaces
>> - do name resolving
>> - and so on.
>> and quite often it cannot do it properly (ie. with little more
>> complicated configuration). 
>
> systemd-networkd is not meant to do very complicated configurations.
But always try :) systemd-resolved also.

I lost a lot of time struggling with systemd-resoved  (I wasn't aware of
its existence), guessing why my dnsmasq stopped working properly.

I simply like unix approach ("do one thing") and systemd is against it.

KJ


-- 
http://stopstopnop.pl/stop_stopnop.pl_o_nas.html

Re: Buster without systemd?

[David]

On Tue, 24 Mar 2020 at 20:57, Gene Heskett  wrote:
> On Tuesday 24 March 2020 04:38:10 to...@tuxteam.de wrote:
> > On Mon, Mar 23, 2020 at 04:09:14PM -0400, Gene Heskett wrote:
> > > On Monday 23 March 2020 15:16:33 Joe wrote:

> > One basic rule of science and engineering should read "if you want
> > to criticise something, you better learn first how it works".

> Which I've not found the docs that would teach me that yet. If such
> actually exist, where do I find the top of that man tree?

Maybe you missed this previous message:
https://lists.debian.org/debian-user/2019/06/msg00277.html

Re: Buster without systemd?

[Dan Ritter]

Andrei POPESCU wrote: 
> For me there's also a clear benefit to having everything well 
> integrated, with the same configuration file format, similar concepts, 
> etc.
> 
> And why should I install a "full" network configuration program, DNS 
> resolver, ntp daemon, etc. when the systemd tools already do what I 
> need?

See, we are different people. We have different needs.

I have more than a hundred Debian servers that already work with
well-understood versions of all of those things. Why should I
have to settle for systemd's new bugs when I already have things
working?


-dsr-

Re: Buster without systemd?

[Andrei POPESCU]

On Ma, 24 mar 20, 12:49:20, Andrei POPESCU wrote:
> On Ma, 24 mar 20, 11:14:47, to...@tuxteam.de wrote:
> > [1] I can't check, since I have no systemd installed. But have
> >you tried "man -k systemd"?
> 
> On stretch with systemd-container installed:

Correction: buster
 
> $ apropos systemd | wc -l
> 180
> 
> The long name helps to pick out what you need from that.

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Andrei POPESCU]

On Ma, 24 mar 20, 06:42:44, Gene Heskett wrote:
> On Tuesday 24 March 2020 06:04:10 Andrei POPESCU wrote:
> > On Lu, 23 mar 20, 18:42:49, Gene Heskett wrote:
> >
> > > For the "vast majority of system's", as I install them, the first
> > > thing I have to do is uninstall that stuff as it has yet in the last
> > > 5 years, to use anything in the routing table but the avahi supplied
> > > 169.xx.cc.nn address which is not allowed off the premises by
> > > dd-wrt.
> >
> > [citation needed]
> 
> My own posting about it in just the past year, but that posting was only 
> done after the problem was solved. As its a bit hard to post from the 
> machine that has a bogus routing setup. With dhcpd5 (I think thats the 
> right name) in the system it does no good to put the correct route as a 
> gateway statement in your /e/n/interfaces tree. So that and avahi 
> generally get nuked. IMNSHO both were written to screw up networking, 
> their only possible reason-de-tere.

Zeroconf works fine when configured correctly :P

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Andrei POPESCU]

On Lu, 23 mar 20, 13:12:02, Dan Ritter wrote:
> 
> systemd started with a good idea: let's make an init system that solves
> the problems of sysvinit. They then proceeded to ignore the long
> history of people writing software to do that, and chose:
> 
> - a heavyweight implementation
> - written as a series of executables that interlock with each
> - which try to handle:
> 
> - process 0 init existence
> - system startup and shutdown 
> - daemon start, status check and stop
> 
> Everything above this line is generally agreed to be the
> province of an init system. Systemd also wants to take over:
> 
> - system logging
> - network interface configuration, including DHCP
> - DNS resolver selection
> - network time protocol
> - cron
> - login management and authentication/authorization
> - setup of virtual machines
> - package management
> 
> I see the attitude of most people as being "I don't care, as
> long as it works."  That's a sane attitude.

For me there's also a clear benefit to having everything well 
integrated, with the same configuration file format, similar concepts, 
etc.

And why should I install a "full" network configuration program, DNS 
resolver, ntp daemon, etc. when the systemd tools already do what I 
need?

There is probably also the philosophical question of "loose" versus 
"tight" integration.

In my opinion "loose" integration is better in theory, while "tight" 
integration is more successful in practice.

Just look at the most successful projects in the FLOSS ecosystem: GCC, 
Emacs, Vim, X.org, Linux, LaTeX, (neo)mutt, GNOME, KDE, GIMP, Firefox, 
LibreOffice, Kodi, etc.

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Greg Wooledge]

On Tue, Mar 24, 2020 at 06:42:44AM -0400, Gene Heskett wrote:
> My own posting about it in just the past year, but that posting was only 
> done after the problem was solved. As its a bit hard to post from the 
> machine that has a bogus routing setup. With dhcpd5 (I think thats the 
> right name) in the system it does no good to put the correct route as a 
> gateway statement in your /e/n/interfaces tree. So that and avahi 
> generally get nuked. IMNSHO both were written to screw up networking, 
> their only possible reason-de-tere.

Debian does not use dhcpcd5 by default.

You know who does?  Raspbian.

Re: Buster without systemd?

[tomas]

On Tue, Mar 24, 2020 at 12:49:20PM +0200, Andrei POPESCU wrote:
> On Ma, 24 mar 20, 11:14:47, to...@tuxteam.de wrote:
> > [1] I can't check, since I have no systemd installed. But have
> >you tried "man -k systemd"?
> 
> On stretch with systemd-container installed:
> 
> $ apropos systemd | wc -l
> 180

:-)

Yep, thanks. Something like that. Plenty of docs. And judging
by what I've seen, they are well-written.

> The long name helps to pick out what you need from that.

Thanks for checking, cheers
-- tomás


signature.asc
Description: Digital signature

Re: Buster without systemd?

[Andrei POPESCU]

On Ma, 24 mar 20, 11:14:47, to...@tuxteam.de wrote:
> [1] I can't check, since I have no systemd installed. But have
>you tried "man -k systemd"?

On stretch with systemd-container installed:

$ apropos systemd | wc -l
180

The long name helps to pick out what you need from that.

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Gene Heskett]

On Tuesday 24 March 2020 06:04:10 Andrei POPESCU wrote:

> On Lu, 23 mar 20, 18:42:49, Gene Heskett wrote:
> > Thats a problem I don't have Greg. I went to a locally defined hosts
> > file 30+ years ago for all my private resolutions, and it Just
> > Works. Queries that go out on the wire for resolution are relayed to
> > the dns services of my provider. Resolution times for external sites
> > are sub 100 millisecond as a general rule. Thats ALL handled by my
> > router running dd-wrt which I think is using dnsmasq.
> >
> > Yet every time I promote such a structure as a solution to someones
> > local network problems I am the idiot according to you for not using
> > dhcp globally. I don't enjoy being painted as an a-hole for using
> > something that Just Works.  So I've quit unless you or someone like
> > you pulls my chain.
>
> Static IPs and /etc/hosts works just fine here with systemd-networkd /
> systemd-resolved and Network Manager.
>
> > For the "vast majority of system's", as I install them, the first
> > thing I have to do is uninstall that stuff as it has yet in the last
> > 5 years, to use anything in the routing table but the avahi supplied
> > 169.xx.cc.nn address which is not allowed off the premises by
> > dd-wrt.
>
> [citation needed]

My own posting about it in just the past year, but that posting was only 
done after the problem was solved. As its a bit hard to post from the 
machine that has a bogus routing setup. With dhcpd5 (I think thats the 
right name) in the system it does no good to put the correct route as a 
gateway statement in your /e/n/interfaces tree. So that and avahi 
generally get nuked. IMNSHO both were written to screw up networking, 
their only possible reason-de-tere.

> Kind regards,
> Andrei

To you too Andrei.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: Buster without systemd?

[Andrei POPESCU]

On Lu, 23 mar 20, 19:51:18, Kamil Jońca wrote:
> But now systemd wants:
> - manage network interfaces
> - do name resolving
> - and so on.
> and quite often it cannot do it properly (ie. with little more
> complicated configuration). 

systemd-networkd is not meant to do very complicated configurations.

Why should I install an additional package (e.g. ifupdown, Network 
Manager, etc.) just for setting up a static IP?

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[tomas]

On Tue, Mar 24, 2020 at 05:41:17AM -0400, Gene Heskett wrote:
> On Tuesday 24 March 2020 04:38:10 to...@tuxteam.de wrote:
> 
> > On Mon, Mar 23, 2020 at 04:09:14PM -0400, Gene Heskett wrote:
> > > On Monday 23 March 2020 15:16:33 Joe wrote:
> >
> > [...]
> >
> > > > You might add 5a): Why?
> > > >
> > > > There's not exactly a burning need to compress text into
> > > > illegibility to save disc space [...]
> > >
> > > NONE Thats my Main objection to it [...]
> >
> > And this is why we can't have nice things.
> >
> > One basic rule of science and engineering should read "if you want
> > to criticise something, you better learn first how it works".
> >
> Which I've not found the docs that would teach me that yet. If such 
> actually exist, where do I find the top of that man tree?

Come on, Gene. There's a lot of things (at least from my point
of view [0]) to criticize of systemd. Lack of documentation
isn't one. It comes with a very complete set of man pages [1].
Entering "systemd documentation" in your search engine of choice
yields as top results things like [2] or [3]. Start reading,
if that's your thirst.

Now you could argue that you haven't got the time to digest a
full new system if you already have one you feel comfy with;
this would be absolutely OK. But spreading half-truths (or
even lies) about that system you don't know (in this thread's
case: that systemd's binary log format is there to save disk
space) -- this is *not* OK.

> > Folks. This ain't twitter.
> 
> No, its quite a bit more verbose.

And hopefully more thoughtful. Hopefully.

> Thanks for pulling my chain, Tomas, and makeing me discover that docs do 
> indeed exist.

Here's n the hope of having made your life more interesting.
You make ours more interesting, too -- so it's giving back.

Cheers
[0] And no, no self-med. Henning: I'm still angry at you about
   that.
[1] I can't check, since I have no systemd installed. But have
   you tried "man -k systemd"?
[2] https://www.freedesktop.org/wiki/Software/systemd/
[3] https://systemd.io/
-- tomás


signature.asc
Description: Digital signature

Re: Buster without systemd?

[Andrei POPESCU]

On Lu, 23 mar 20, 18:42:49, Gene Heskett wrote:
>
> Thats a problem I don't have Greg. I went to a locally defined hosts file 
> 30+ years ago for all my private resolutions, and it Just Works. Queries 
> that go out on the wire for resolution are relayed to the dns services 
> of my provider. Resolution times for external sites are sub 100 
> millisecond as a general rule. Thats ALL handled by my router running 
> dd-wrt which I think is using dnsmasq.
> 
> Yet every time I promote such a structure as a solution to someones local 
> network problems I am the idiot according to you for not using dhcp 
> globally. I don't enjoy being painted as an a-hole for using something 
> that Just Works.  So I've quit unless you or someone like you pulls my 
> chain.

Static IPs and /etc/hosts works just fine here with systemd-networkd / 
systemd-resolved and Network Manager.
 
> For the "vast majority of system's", as I install them, the first thing I 
> have to do is uninstall that stuff as it has yet in the last 5 years, to 
> use anything in the routing table but the avahi supplied 169.xx.cc.nn 
> address which is not allowed off the premises by dd-wrt.

[citation needed]

Kind regards,
Andrei
-- 
http://wiki.debian.org/FAQsFromDebianUser


signature.asc
Description: PGP signature

Re: Buster without systemd?

[Gene Heskett]

On Tuesday 24 March 2020 04:38:10 to...@tuxteam.de wrote:

> On Mon, Mar 23, 2020 at 04:09:14PM -0400, Gene Heskett wrote:
> > On Monday 23 March 2020 15:16:33 Joe wrote:
>
> [...]
>
> > > You might add 5a): Why?
> > >
> > > There's not exactly a burning need to compress text into
> > > illegibility to save disc space [...]
> >
> > NONE Thats my Main objection to it [...]
>
> And this is why we can't have nice things.
>
> One basic rule of science and engineering should read "if you want
> to criticise something, you better learn first how it works".
>
Which I've not found the docs that would teach me that yet. If such 
actually exist, where do I find the top of that man tree?

> There are valid reasons for systemd's binary format. Space savings
> is very far off the top of the list (probably it isn't on that list
> at all).

Show me. Or show myself. Now I see there is actually such a manpage now. 
It didn't exist the last time I looked which was probably 3+ years ago 
after it came up at about jessie in my awareness. Its gonna take a while 
to absorb all that, the manpage starts out a bit "abstract" and will 
take a goodly number of re-reads to connect all the dots.  Lots of see 
also's to absorb.  And having become aware of the manpages for that, I 
withdraw my objections until such time as I do understand it.

> Folks. This ain't twitter.

No, its quite a bit more verbose.

Thanks for pulling my chain, Tomas, and makeing me discover that docs do 
indeed exist.
> Cheers
> -- tomás


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: Buster without systemd?

[tomas]

On Mon, Mar 23, 2020 at 01:15:28PM -0400, Henning Follmann wrote:
> On Mon, Mar 23, 2020 at 04:31:33PM +, Tony van der Hoff wrote:
> >
> [...]
> > I know it's a sensitive subject, and I really don't want to upset the list,
> > there's been enough of that already, but why are some people so afraid of
> > systemd?
> 
> Who said "Jehovah"?

:-)

> > I have  used it since the beginning of jessie, through stretch, and now
> > buster, and have had no problems with it.
> > 
> > I ask the question in all innocence, purely to understand whypeople seem to
> > want to jump through hoops to avoid it.
> 
> Maybe lack of self medication?

My reply to this one would definitely run afoul of list policy
(and against my own principles).

Luckily your creative imagination can make up for that.

non cheers
-- t


signature.asc
Description: Digital signature

Re: Buster without systemd?

[tomas]

On Mon, Mar 23, 2020 at 04:09:14PM -0400, Gene Heskett wrote:
> On Monday 23 March 2020 15:16:33 Joe wrote:

[...]

> > You might add 5a): Why?
> >
> > There's not exactly a burning need to compress text into illegibility
> > to save disc space [...]

> NONE Thats my Main objection to it [...]

And this is why we can't have nice things.

One basic rule of science and engineering should read "if you want
to criticise something, you better learn first how it works".

There are valid reasons for systemd's binary format. Space savings
is very far off the top of the list (probably it isn't on that list
at all).

Folks. This ain't twitter.

Cheers
-- tomás


signature.asc
Description: Digital signature

Re: Buster without systemd?

[David Wright]

On Mon 23 Mar 2020 at 20:27:31 (+0100), Klaus Singvogel wrote:
> Tony van der Hoff wrote:
> > I know it's a sensitive subject, and I really don't want to upset the list,
> > there's been enough of that already, but why are some people so afraid of
> > systemd?
> 
> My reasons:

[…]

> Logging (journald) disables searching in /var/log: "grep -r" not really
> working there.

I commented on this already. grep -r is alive and well.

> My computer "hickups" for 20 seconds at boot. With systemd I don't have
> the slighest idea which process delays, as several processes start after
> the delay at once. I had the feeling with Init scripts it was more obvious
> to see which "startup file" hangs.

$ systemd-analyze critical-chain

Is that any help in finding what's waiting for what?

> Its flooding syslog/messages with unnecessary information (type: INFO).

Does removing "info" from the .conf file not work?

> But some important messages are not found in syslog nor messages.

Which ones are missing?

> No easy way to find out: if and what fails. Example: After months I found
> out that mlocate doesn't update it's database anymore: previously updates
> were done by cron, but the distri moved to systemd's way (timer) and by
> default the updatedb was disabled.

I've not touched the default. What did you have to change to make it run?

$ ls -l /var/lib/mlocate/mlocate.db 
-rw-r- 1 root mlocate 46870853 Mar 23 08:18 /var/lib/mlocate/mlocate.db
$ 

Cheers,
David.

Re: Buster without systemd?

[David Wright]

On Mon 23 Mar 2020 at 19:16:33 (+), Joe wrote:
> On Mon, 23 Mar 2020 13:15:13 -0400 Greg Wooledge  wrote:

> > 5) It does not write logs in human-readable files.  You need systemd's
> >tools to read systemd's logs.  This makes post mortem diagnostics
> >much more difficult.
> 
> You might add 5a): Why? 
> 
> There's not exactly a burning need to compress text into illegibility to
> save disc space these days. How many rational purposes are there to
> make logs readable only through systemd's own code?

All my buster systems log in the usual manner. Yes, I have some
incantations that I've copied from this list, but I don't normally
use them. eg:

$ journalctl -xe _SYSTEMD_UNIT=ssh.service | lessx
-- Logs begin at Mon 2020-03-23 08:11:58 CDT, end at Mon 2020-03-23 21:10:01 
CDT. --
Mar 23 08:12:35 wren sshd[701]: Server listening on 0.0.0.0 port 22.
Mar 23 08:12:35 wren sshd[701]: Server listening on :: port 22.
$ grep 'Server listening' /var/log/auth.log
Mar 22 08:10:57 wren sshd[976]: Server listening on 0.0.0.0 port 22.
Mar 22 08:10:57 wren sshd[976]: Server listening on :: port 22.
Mar 23 08:12:35 wren sshd[701]: Server listening on 0.0.0.0 port 22.
Mar 23 08:12:35 wren sshd[701]: Server listening on :: port 22.
$ 

Looks the same to me.

Cheers,
David.

Re: Buster without systemd?

[Michael Howard]

On 23/03/2020 23:20, Patrick Bartek wrote:

On Mon, 23 Mar 2020 15:15:14 +
Michael Howard  wrote:


On 23/03/2020 14:28, Patrick Bartek wrote:

On Mon, 23 Mar 2020 08:37:47 -0400
Jude DaShiell  wrote:
  

There is devuan-ascii 2.x but I don't know its equivalent to buster.
That system maintained sysv and still has support.

ASCII is Stretch without systemd or any of the dependencies.  Devuan
hasn't release its "Buster" version yet.

B
  

It can be upgraded to 'beowulf' using apt of course.

I did that some months ago with Ascii running in Virtualbox.  And it did
work, but there were enough problems then that it wasn't suitable
for general use -- just evaluation. Perhaps there have been
improvements since then, but I'll wait for Beowulf to be officially
released before checking it out again.  Currently, I'll stick with
Stretch running sysvinit until long term support for ceases.

B


I guess it depends exactly what is installed but I have upgraded
via this route a number of times.

--
Michael Howard

Re: Buster without systemd?

[Patrick Bartek]

On Mon, 23 Mar 2020 15:15:14 +
Michael Howard  wrote:

> On 23/03/2020 14:28, Patrick Bartek wrote:
> > On Mon, 23 Mar 2020 08:37:47 -0400
> > Jude DaShiell  wrote:
> >  
> >> There is devuan-ascii 2.x but I don't know its equivalent to buster.
> >> That system maintained sysv and still has support.  
> > ASCII is Stretch without systemd or any of the dependencies.  Devuan
> > hasn't release its "Buster" version yet.
> >
> > B
> >  
> It can be upgraded to 'beowulf' using apt of course.

I did that some months ago with Ascii running in Virtualbox.  And it did
work, but there were enough problems then that it wasn't suitable
for general use -- just evaluation. Perhaps there have been
improvements since then, but I'll wait for Beowulf to be officially
released before checking it out again.  Currently, I'll stick with
Stretch running sysvinit until long term support for ceases.

B

Re: Buster without systemd?

[Toni Mas]

I did. I just did not use "full-upgrade" option. I upgraded package by
package resolving all dependencies and I had to install elogind but it
is not needed to start X system. It was just for dependencies.

Please, could you explain the race of conditions risk race?

Thanks.

Toni Mas

Missatge de Renato Gallo  del dia dl., 23 de
març 2020 a les 9:06:
>
>
> linux without systemd = race condition risks = why in hell anyone would want 
> to do it ?
>
> - Messaggio originale -
> Da: "Felix Miata" 
> A: "debian-user" 
> Inviato: Lunedì, 23 marzo 2020 8:08:28
> Oggetto: Re: Buster without systemd?
>
> Marc Shapiro composed on 2020-03-22 18:21 (UTC-0700):
>
> > after 21 to 22 years of using
> > Debian (since Bo), do I have to switch to another linux distro?
>
> AFAIK, no one has ever died as a consequence of using an OS with systemd. So, 
> no,
> you don't "have to" switch to another distro. You can do as most have done, 
> fondly
> or not so fondly remember sysvinit, and accept the change, whether for better 
> or
> worse.
>
> OTOH, would switching to Devuan really be "switching" to another distro? 
> That's
> like "switching" to any of the zillion distros based on Debian that include 
> Debian
> repos in sources.list. They're mostly Debian but with different defaults,
> different far more the interface than the guts that make Debian debian.
> --
> Evolution as taught in public schools is religion, not science.
>
>  Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!
>
> Felix Miata  ***  http://fm.no-ip.com/
>

Re: Buster without systemd?

[Gene Heskett]

On Monday 23 March 2020 16:34:22 Greg Wooledge wrote:

> On Mon, Mar 23, 2020 at 04:09:14PM -0400, Gene Heskett wrote:
> > That which I have managed to get my head wrapped around, do seem to
> > be a better idea. But thats far from all of it at the present time.
> > For starters, lets say you know you are having a dhcpd problem, so
> > you go read man dhcpd, and it has nothing to say about it if its now
> > managed by systemd. Hells bells, people, at least put a "see also
> > man xxx" pointing at the systemd managed version of its man
> > page.
>
> Please remember that Debian does not use systemd's network interface
> bringing-up features by default.  If you've gone out of your way to do
> so locally, that's on you.
>
Thats a problem I don't have Greg. I went to a locally defined hosts file 
30+ years ago for all my private resolutions, and it Just Works. Queries 
that go out on the wire for resolution are relayed to the dns services 
of my provider. Resolution times for external sites are sub 100 
millisecond as a general rule. Thats ALL handled by my router running 
dd-wrt which I think is using dnsmasq.

Yet every time I promote such a structure as a solution to someones local 
network problems I am the idiot according to you for not using dhcp 
globally. I don't enjoy being painted as an a-hole for using something 
that Just Works.  So I've quit unless you or someone like you pulls my 
chain.

> For a Debian system, the path of inquiry is still the same as before
> -- /etc/network/interfaces and then Network Manager, for the vast
> majority of systems.

For the "vast majority of system's", as I install them, the first thing I 
have to do is uninstall that stuff as it has yet in the last 5 years, to 
use anything in the routing table but the avahi supplied 169.xx.cc.nn 
address which is not allowed off the premises by dd-wrt.

I have repeatedly said I'll use what works, and NM/Avahi hasn't yet in 
all the years it has existed.

> Raspbian is of course a separate distribution, and you'd have to ask a
> Raspbian mailing list if your issue is with something of theirs.

I will, just as soon as one is instigated.  All they have is a forum with 
a broken search engine, if you actually find an answer, its the result 
of reading half the msgs ever posted to that forum. Fortunately I read 
at a good clip. Thats how I got to be a C.E.T., and a very well paid 
broadcast engineer/Chief Operator on an 8th grade education. I got that 
way the hard way, I never asked for a raise, I just got them.


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: Buster without systemd?

[Charles Curley]

On Mon, 23 Mar 2020 19:51:18 +0100
kjo...@fastmail.com (Kamil Jońca) wrote:

> Well. I am quite happy when systemd start to govern
> services/daemons.

Good for you.


I don't suppose you use gpsd under systemd. The gpsd maintainers hate
systemd with a passion because systemd makes assumptions about daemons
which simply do not apply to gpsd. And every time the gpsd maintainers
cobble together some way of beating systemd into submission, systemd
changes the rules and we're back to the drawing board.


> 
> I also understand timers - cron and anacron sometimes are too "weak".

What does "weak" mean?

cron has its problems. So someone -- debian? -- glued anacron on top of
that, giving us the subsequent mess. Now along comes systemd with more
of the same ill-thought-out design? No thanks.


> But now systemd wants:
> - manage network interfaces
> - do name resolving
> - and so on.
> and quite often it cannot do it properly (ie. with little more
> complicated configuration). 

And there's perfectly good code out there, some of which has been in
use since before Bill Gates was in diapers.

-- 
Does anybody read signatures any more?

https://charlescurley.com
https://charlescurley.com/blog/

Re: Buster without systemd?

[deloptes]

Gene Heskett wrote:

> That along would ease the transition, and reduce the the systemd hate and
> discontent on ALL of these lists.  And justify the reason for the
> individual change. None of us like being kept in the dark.

Absolutely agree with this. It took me about an year to read myself in to
the systemd and understand the idea behind.

On the server and the firewall it is buster and still sysvinit. On the
desktop it is systemd - and this is where it belongs.
Meanwhile I must admit it does not cause troubles at all (or at least I do
not notice them, if there are such).

I think it is a very good choice to put it on the desktop. On the server I
do not know what it is good for ... perhaps depends on what kind of server
it is.

Re: Buster without systemd?

[Greg Wooledge]

On Mon, Mar 23, 2020 at 04:09:14PM -0400, Gene Heskett wrote:
> That which I have managed to get my head wrapped around, do seem to be a 
> better idea. But thats far from all of it at the present time. For 
> starters, lets say you know you are having a dhcpd problem, so you go 
> read man dhcpd, and it has nothing to say about it if its now managed by 
> systemd. Hells bells, people, at least put a "see also man xxx" 
> pointing at the systemd managed version of its man page.

Please remember that Debian does not use systemd's network interface
bringing-up features by default.  If you've gone out of your way to do
so locally, that's on you.

For a Debian system, the path of inquiry is still the same as before --
/etc/network/interfaces and then Network Manager, for the vast majority
of systems.

Raspbian is of course a separate distribution, and you'd have to ask a
Raspbian mailing list if your issue is with something of theirs.

Re: Buster without systemd?

[Gene Heskett]

On Monday 23 March 2020 15:16:33 Joe wrote:

> On Mon, 23 Mar 2020 13:15:13 -0400
>
> Greg Wooledge  wrote:
> > 5) It does not write logs in human-readable files.  You need
> > systemd's tools to read systemd's logs.  This makes post mortem
> > diagnostics much more difficult.
>
> You might add 5a): Why?
>
> There's not exactly a burning need to compress text into illegibility
> to save disc space these days. How many rational purposes are there to
> make logs readable only through systemd's own code?

NONE Thats my Main objection to it. If it can't leave logs in 
plain "This is what you need to do text", in the users first language 
with instructions right in the log, one of us needs to get up and leave 
the party.

The current practice of posting the error, and praying someone can make 
ones native language out of it is the stuff found on the ground behind 
the male of the bovine specie.

That which I have managed to get my head wrapped around, do seem to be a 
better idea. But thats far from all of it at the present time. For 
starters, lets say you know you are having a dhcpd problem, so you go 
read man dhcpd, and it has nothing to say about it if its now managed by 
systemd. Hells bells, people, at least put a "see also man xxx" 
pointing at the systemd managed version of its man page.

That along would ease the transition, and reduce the the systemd hate and 
discontent on ALL of these lists.  And justify the reason for the 
individual change. None of us like being kept in the dark.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 

Re: Buster without systemd?

[Klaus Singvogel]

Tony van der Hoff wrote:
> I know it's a sensitive subject, and I really don't want to upset the list,
> there's been enough of that already, but why are some people so afraid of
> systemd?

My reasons:

Can't debug the start / stop of daemons as good as before. "(ba)sh -x" no
more working.

Not able to move easily a more complex SysV init script to systemd. Lost
functionality by my home-made packages.

Logging (journald) disables searching in /var/log: "grep -r" not really
working there.

My computer "hickups" for 20 seconds at boot. With systemd I don't have
the slighest idea which process delays, as several processes start after
the delay at once. I had the feeling with Init scripts it was more obvious
to see which "startup file" hangs.

As others say: systemd is too fat and has too much functionality. It
throws too fast the experienced software overboard, replacing it by its
own version.

Its flooding syslog/messages with unnecessary information (type: INFO).
But some important messages are not found in syslog nor messages.

No easy way to find out: if and what fails. Example: After months I found
out that mlocate doesn't update it's database anymore: previously updates
were done by cron, but the distri moved to systemd's way (timer) and by
default the updatedb was disabled.

Default values without files: configuration files are not required, but
then some default values are used. You have to know which file by which
filename has to be created and how the variables/values are named. Not to
mention the lack of documentation for some configuration files.

Best regards,
Klaus.
-- 
Klaus Singvogel
GnuPG-Key-ID: 1024R/5068792D  1994-06-27

Re: Buster without systemd?

[Joe]

On Mon, 23 Mar 2020 13:15:13 -0400
Greg Wooledge  wrote:


> 
> 5) It does not write logs in human-readable files.  You need systemd's
>tools to read systemd's logs.  This makes post mortem diagnostics
>much more difficult.
> 

You might add 5a): Why? 

There's not exactly a burning need to compress text into illegibility to
save disc space these days. How many rational purposes are there to
make logs readable only through systemd's own code?

-- 
Joe

Re: Buster without systemd?

[Kamil Jońca]

Tony van der Hoff  writes:

> On 23/03/2020 15:15, Michael Howard wrote:
>> On 23/03/2020 14:28, Patrick Bartek wrote:
>>> On Mon, 23 Mar 2020 08:37:47 -0400
>>> Jude DaShiell  wrote:
>>>
> 
>
> I know it's a sensitive subject, and I really don't want to upset the
> list, there's been enough of that already, but why are some people so
> afraid of systemd?
Well. I am quite happy when systemd start to govern services/daemons.

I also understand timers - cron and anacron sometimes are too "weak".
But now systemd wants:
- manage network interfaces
- do name resolving
- and so on.
and quite often it cannot do it properly (ie. with little more
complicated configuration). 

KJ

-- 
http://wolnelektury.pl/wesprzyj/teraz/

Re: Buster without systemd?

[Kenneth Parker]

On Mon, Mar 23, 2020, 12:48 PM Tony van der Hoff 
wrote:

>
> I know it's a sensitive subject, and I really don't want to upset the
> list, there's been enough of that already, but why are some people so
> afraid of systemd?
>

I consider myself Neutral on the subject, actually wanting to educate
myself on, both systemd and its alternatives.  So, since I have multiple
Partitions, on multiple Systems, I have room to experiment with both.  So
my two, main Operating systems are Buster, and Devuan Ascii.  And I have
Development tools on both, to help my Education.

Mainly, my goal, right now is to use each correctly.

Also, I can answer questions on both frameworks.

Kenneth Parker

>

Re: Buster without systemd?

[Dan Ritter]

Henning Follmann wrote: 
> On Mon, Mar 23, 2020 at 04:31:33PM +, Tony van der Hoff wrote:
> > I ask the question in all innocence, purely to understand whypeople seem to
> > want to jump through hoops to avoid it.
> > 
> 
> Maybe lack of self medication?
> Duck...

Not only was that not helpful, you knew it wasn't going to be
helpful. And you still posted it.

-dsr-

Re: Buster without systemd?

[Henning Follmann]

On Mon, Mar 23, 2020 at 04:31:33PM +, Tony van der Hoff wrote:
>
[...]
> I know it's a sensitive subject, and I really don't want to upset the list,
> there's been enough of that already, but why are some people so afraid of
> systemd?
>

Who said "Jehovah"?


> I have  used it since the beginning of jessie, through stretch, and now
> buster, and have had no problems with it.
> 
> I ask the question in all innocence, purely to understand whypeople seem to
> want to jump through hoops to avoid it.
> 

Maybe lack of self medication?
Duck...


-- 
Henning Follmann   | hfollm...@itcfollmann.com

Re: Buster without systemd?

[Greg Wooledge]

On Mon, Mar 23, 2020 at 04:31:33PM +, Tony van der Hoff wrote:
> I know it's a sensitive subject, and I really don't want to upset the list,
> there's been enough of that already, but why are some people so afraid of
> systemd?

Well, there seem to be several reasons.  In no particular order:

1) It's new.  Some people hate new things, simply because they've
   become accustomed to the old things.

2) Because it's new, it has not had a chance to reach maturity.  There
   are still flaws being worked out.

3) It's extremely large.  Compared to sysvinit, it is a sprawling
   leviathan of software.  Not only does size = bugs, but some people
   object to the increase in memory usage, disk space usage, etc.

4) It does way more than just replace /sbin/init.  It has components
   that try to bring up network interfaces, implement DNS resolution,
   implement NTP, implement cron-like functionality, and so on.  Many
   people feel this is over-reaching.  (And to be fair, Debian disables
   most of these optional subsystems by default.)

5) It does not write logs in human-readable files.  You need systemd's
   tools to read systemd's logs.  This makes post mortem diagnostics
   much more difficult.

6) It is more complex than sysvinit.  There's a lot to learn.  However, it
   should be noted that there is an enormous degree of *hidden* complexity
   in sysv-rc (decades worth of hacks developed to work around sysv-rc's
   fundamental flaws), and these things go away with systemd.  So, really
   it's trading one set of complexity for a different set of complexity.
   But some people don't see it that way.

7) Some people have developed a personal dislike of the author.

I'm sure there are other reasons, but these are the main ones of which
I'm aware.

Re: Buster without systemd?

[Dan Ritter]

Tony van der Hoff wrote: 
> On 23/03/2020 15:15, Michael Howard wrote:
> > On 23/03/2020 14:28, Patrick Bartek wrote:
> > > On Mon, 23 Mar 2020 08:37:47 -0400
> > > Jude DaShiell  wrote:
> > > 
> 
> 
> I know it's a sensitive subject, and I really don't want to upset the list,
> there's been enough of that already, but why are some people so afraid of
> systemd?
 
Nobody, to a first approximation, is afraid of systemd.

> I ask the question in all innocence, purely to understand whypeople seem to
> want to jump through hoops to avoid it.


systemd started with a good idea: let's make an init system that solves
the problems of sysvinit. They then proceeded to ignore the long
history of people writing software to do that, and chose:

- a heavyweight implementation
- written as a series of executables that interlock with each
- which try to handle:

- process 0 init existence
- system startup and shutdown 
- daemon start, status check and stop

Everything above this line is generally agreed to be the
province of an init system. Systemd also wants to take over:

- system logging
- network interface configuration, including DHCP
- DNS resolver selection
- network time protocol
- cron
- login management and authentication/authorization
- setup of virtual machines
- package management

I see the attitude of most people as being "I don't care, as
long as it works."  That's a sane attitude.

The problem is that for me -- and many other people -- there is
a long history of not wanting to change a working system unless
the benefits of that change are clear and worthwhile. I judge
that systemd is less reliable than existing systems, causes
needless chaos, and does not deliver benefits that are not
available from other, less disruptive systems.

Not only does it have to work, it has to work well. 

-dsr-

Re: Buster without systemd?

[Michael Howard]

On 23/03/2020 16:31, Tony van der Hoff wrote:

On 23/03/2020 15:15, Michael Howard wrote:

On 23/03/2020 14:28, Patrick Bartek wrote:

On Mon, 23 Mar 2020 08:37:47 -0400
Jude DaShiell  wrote:




I know it's a sensitive subject, and I really don't want to upset the 
list, there's been enough of that already, but why are some people so 
afraid of systemd?

Afraid?

--
Michael Howard

Re: Buster without systemd?

[tomas]

On Mon, Mar 23, 2020 at 04:31:33PM +, Tony van der Hoff wrote:
> On 23/03/2020 15:15, Michael Howard wrote:
> >On 23/03/2020 14:28, Patrick Bartek wrote:
> >>On Mon, 23 Mar 2020 08:37:47 -0400
> >>Jude DaShiell  wrote:
> >>
> 
> 
> I know it's a sensitive subject, and I really don't want to upset
> the list, there's been enough of that already, but why are some
> people so afraid of systemd?
> 
> I have  used it since the beginning of jessie, through stretch, and
> now buster, and have had no problems with it.
> 
> I ask the question in all innocence, purely to understand whypeople
> seem to want to jump through hoops to avoid it.

I think you'll get as many answers as people out there. In my case,
I'd like to keep alternatives to systemd viable. At the same time,
I want the discussion to stay civil and don't want to see anyone
denigrated over it. Proponents and opponents likewise.

Cheers
-- tomás


signature.asc
Description: Digital signature

Re: Buster without systemd?

[Tony van der Hoff]

On 23/03/2020 15:15, Michael Howard wrote:

On 23/03/2020 14:28, Patrick Bartek wrote:

On Mon, 23 Mar 2020 08:37:47 -0400
Jude DaShiell  wrote:




I know it's a sensitive subject, and I really don't want to upset the 
list, there's been enough of that already, but why are some people so 
afraid of systemd?


I have  used it since the beginning of jessie, through stretch, and now 
buster, and have had no problems with it.


I ask the question in all innocence, purely to understand whypeople seem 
to want to jump through hoops to avoid it.


--
Tony van der Hoff| mailto:t...@vanderhoff.org
Buckinghamshire, England |

Re: Buster without systemd?

[Michael Howard]

On 23/03/2020 14:28, Patrick Bartek wrote:

On Mon, 23 Mar 2020 08:37:47 -0400
Jude DaShiell  wrote:


There is devuan-ascii 2.x but I don't know its equivalent to buster.
That system maintained sysv and still has support.

ASCII is Stretch without systemd or any of the dependencies.  Devuan
hasn't release its "Buster" version yet.

B


It can be upgraded to 'beowulf' using apt of course.

--
Michael Howard