Choice of browser? [WAS Re: how do i configure lynx browser as default browser for html]

[Andrew M.A. Cater]

On Thu, Sep 29, 2022 at 08:02:54AM -0400, The Wanderer wrote:
> On 2022-09-29 at 07:49, Andrew M.A. Cater wrote:
> 
> > There are hints in this thtread as to where to configure your
> > default browser. Update-alternatives is probably your friend here. If
> > all of this is just to run the Wikipedia cleaner application, then
> > you may well want a GUI browser anyway to check how it looks to
> > others in a GUI.
> 
> To be fair, it used to be - and, by some lights, still is - good
> practice to test a Website in multiple types of browser to make sure it
> works to a suitable standard in all of them.
> 
> I certainly wouldn't want to discourage people from including text-mode
> Web browsers in that type of testing.
> 

Hi Wanderer,

Yes, absolutely: as someone peripherally involved in accessibility and
disability, this is completely correct and desirable. If lynx is the 
only text-mode browser you test, that's fine - links and elinks also
exist and may produce different results. If you yourself are constrained to 
use a text mode browser for other reasons, then you are already very
aware that a graphical browser may be very different.

Wikipedia is fairly heavily formatted, as are most wikis so it shouldn't
make a lot of difference.

All the very best, as ever,

Andy Cater

> -- 
>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
> 

Re: choice of web browsers

[songbird]

riveravaldez wrote:
...
> $ dpigs -n2 -SH
> 287.7M linux-signed-amd64
> 195.4M firefox-esr
>
> ;)


  things be different at this machine:

$ dpigs -n8 -SH
 499.3M linux
 359.6M golang-1.16
 357.8M linux-signed-amd64
 325.8M libreoffice
 292.4M qemu
 273.7M rustc
 213.1M python3.9
 203.5M firefox


  songbird

Re: choice of web browsers

[riveravaldez]

On Friday, October 29, 2021, Dan Ritter  wrote:
> kaye n wrote:
>> Maybe you're right about Firefox.  Any "light" browsers you can
recommend?
>
> The problem with web browsers is that they are basically a
> second operating system for computers.
>
> (The great thing about web browsers is that they are a
> nearly-universal operating system.)

$ dpigs -n2 -SH
287.7M linux-signed-amd64
195.4M firefox-esr

;)

Re: choice of web browsers

[Christian Britz]

At 29.10.21 Dan Ritter wrote:

> and then the heavies:
> 
> chromium
> firefox

I was a long time user of Mozilla based browsers (Netscape -> Mozilla
Suite -> SeaMonkey -> Firefox) but lately I found Chromium based
browsers faster and more appealing.
Beat me, but I even prefer Chrome above Chromium. That is because I
actually like the Google integration and security updates from Google
are faster and more often delivered than for the Chromium in Debian.

Regards,
Christian

Re: choice of web browsers

[Roy J. Tellason, Sr.]

On Friday 29 October 2021 09:10:29 am Dan Ritter wrote:
> You can substantially improve the memory usage and speed of
> firefox by installing the extension ublock Origin.
 
Thanks for posting this,  I've installed it and reviewed a fair amount of the 
docs.  It looks to be truly useful...


-- 
Member of the toughest, meanest, deadliest, most unrelenting -- and
ablest -- form of life in this section of space,  a critter that can
be killed but can't be tamed.  --Robert A. Heinlein, "The Puppet Masters"
-
Information is more dangerous than cannon to a society ruled by lies. --James 
M Dakin

choice of web browsers

[Dan Ritter]

kaye n wrote: 
> Maybe you're right about Firefox.  Any "light" browsers you can recommend?

The problem with web browsers is that they are basically a
second operating system for computers.

(The great thing about web browsers is that they are a
nearly-universal operating system.)

The lightest web browsers are also the least capable:

lynx
w3m
elinks
links

then

cog
dillo
epiphany-browser
falkon
luakit
midori
netsurf
surf


and then the heavies:

chromium
firefox

You can substantially improve the memory usage and speed of
firefox by installing the extension ublock Origin.

-dsr-

Re: wlan0: deauthenticating from xx:xx by local choice (Reason 3=Deauth Leaving)

[David BERCOT]

Hello,

J'ai déjà eu une erreur de ce type liée à OpenSSL.
Il y avait un problème de compatibilité (de mémoire, le serveur était
trop ancien et le client trop récent et il m'a fallu autoriser le client
à utiliser une "vieille" version) mais, me concernant, la connexion ne
marchait pas du tout avant ma modification.
Je pense donc que ça vient d'ailleurs.
Néanmoins, si ça peut aider...

David.

Le 12/05/2020 à 14:52, didier gaumet a écrit :
> Le 12/05/2020 à 14:22, MERLIN Philippe a écrit :
> [...]
>> erreur relevée dans syslog :
>> Kernel  wlan0: deauthenticating from xx:xx  by local choice (Reason 3=Deauth 
>> Leaving)
>> Quel paquet émet se message dans syslog ?  il est indiqué Kernel cela 
>> voudrait 
>> dire que c'est le noyau linux , je me trompe ? Et pourquoi ? Cette erreur ne 
>> semble pas bloquante puisque la connexion s'établit la plupart du temps.
>> Google m'a été d'aucun secours.
> [...]
> 
> ça ne va pas t'aider beaucoup: je pense que c'est wpa_supplicant qui
> retourne cet état, mais je n'ai pas trouvé de liste de messages
> d'erreurs ou de statuts sur le net...
> 

wlan0: deauthenticating from xx:xx by local choice (Reason 3=Deauth Leaving)

[MERLIN Philippe]

Bonjour,
Dans la recherche des causes de l'initialisation aléatoire de ma connexion  
Wifi c'est une suite aux messages précédents intitulés (problème de connexion 
avec Wicd,  le paquet pump  chez Debian existe t'il encore?) j'ai que la 
connexion s'initialise ou pas cette erreur relevée dans syslog :
Kernel  wlan0: deauthenticating from xx:xx  by local choice (Reason 3=Deauth 
Leaving)
Quel paquet émet se message dans syslog ?  il est indiqué Kernel cela voudrait 
dire que c'est le noyau linux , je me trompe ? Et pourquoi ? Cette erreur ne 
semble pas bloquante puisque la connexion s'établit la plupart du temps.
Google m'a été d'aucun secours.
Toute idée sur le sujet sera la bienvenue.
Philippe Merlin
P.S. Tes idées Didier Gaumet m'ont parues très intéressantes, je ne les ai pas 
encore testées je les garde en mémoire, me focalisant actuellement sur 
l'analyse des logs.

Re: wlan0: deauthenticating from xx:xx by local choice (Reason 3=Deauth Leaving)

[Fabien R]

On 12/05/2020 14:22, MERLIN Philippe wrote:
> Bonjour,
> Dans la recherche des causes de l'initialisation aléatoire de ma connexion  
> Wifi c'est une suite aux messages précédents intitulés (problème de connexion 
> avec Wicd,  le paquet pump  chez Debian existe t'il encore?) j'ai que la 
> connexion s'initialise ou pas cette erreur relevée dans syslog :
> Kernel  wlan0: deauthenticating from xx:xx  by local choice (Reason 3=Deauth 
> Leaving)
> Quel paquet émet se message dans syslog ?  il est indiqué Kernel cela 
> voudrait 
> dire que c'est le noyau linux , je me trompe ? Et pourquoi ? Cette erreur ne 
Sur ma machine, ce msg s'affiche lorsque je faiskillall -q wpa_supplicant
lorsque je désactive l'interface par ifdown.

--
Fabien

Re: wlan0: deauthenticating from xx:xx by local choice (Reason 3=Deauth Leaving)

[didier gaumet]

Le 12/05/2020 à 14:22, MERLIN Philippe a écrit :
[...]
> erreur relevée dans syslog :
> Kernel  wlan0: deauthenticating from xx:xx  by local choice (Reason 3=Deauth 
> Leaving)
> Quel paquet émet se message dans syslog ?  il est indiqué Kernel cela 
> voudrait 
> dire que c'est le noyau linux , je me trompe ? Et pourquoi ? Cette erreur ne 
> semble pas bloquante puisque la connexion s'établit la plupart du temps.
> Google m'a été d'aucun secours.
[...]

ça ne va pas t'aider beaucoup: je pense que c'est wpa_supplicant qui
retourne cet état, mais je n'ai pas trouvé de liste de messages
d'erreurs ou de statuts sur le net...

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 5:09 AM Tom Browder  wrote:
>
> I am preparing servers to use with OpenSMTPD and Sympa to provide mail
and mailing list service.
...

After considering all the suggestions, I have prepared a plan, using a bit
of pseudo code, to describe what I believe I need to do.

Interested parties can view at this gist:

  https://gist.github.com/tbrowder/0be832c0405d54021fa730b791629643

I appreciate all the help for, IMHO, a very complicated subject.


-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 11:04 AM Henning Follmann
 wrote:
> On Fri, Feb 21, 2020 at 08:47:36AM -0600, Tom Browder wrote:
> > On Fri, Feb 21, 2020 at 08:37 Henning Follmann 
...
> For years I operated a couple of sendmail installations. That and the
> O'Reilly Sendmail book tought me a lot.
> Today I prefer postfix. There is a german book from Peer Heinlein.
> This is an excellent source if you want to set up a multidomain
> mailserver completely driven by a LDAP directory.

That got some good reviews--sorry my old German-speaking roommate is not nearby.

Thanks, Henning.

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Brian]

On Fri 21 Feb 2020 at 12:36:38 -0600, Tom Browder wrote:

> On Fri, Feb 21, 2020 at 12:05 PM Brian  wrote:
> > On Fri 21 Feb 2020 at 11:22:58 -0600, Tom Browder wrote:
> >
> > > On Fri, Feb 21, 2020 at 09:36 Brian  wrote:
> ...
> > Thanks, I will read the page in detail later. Meanwhile, I did a quick
> > search on the page for "mailname" and didn't get anything. Anyway, what
> > I really wanted to know was what function *you* thought /etc/mailname
> > played in OpenSMTPD because it would determine what you chose for it.
> >
> > On Exim it is used to qualify a local part without a domain name. After
> > I installed OpenSMTPD it seems to me that this is also what OpenSMTPD
> > does. In other words, if I had mailname as gmail.com, a mail I send to
> > tombrowder (no domain name) would go to tombrow...@gmail.com.
> 
> Brian, you may be right. I cannot find anything in the man pages, but
> it may have been stated during the installation of the Debian package.
> I'll uninstall and reinstall to see if I can capture the
> instructions...

Instead, you can do 'dpkg-reconfigure opensmtpd'.

-- 
Brian.

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 12:36 PM Tom Browder  wrote:
> On Fri, Feb 21, 2020 at 12:05 PM Brian  wrote:
> > Thanks, I will read the page in detail later. Meanwhile, I did a quick
> > search on the page for "mailname" and didn't get anything. Anyway, what
> > I really wanted to know was what function *you* thought /etc/mailname
> > played in OpenSMTPD because it would determine what you chose for it.
> >
> > On Exim it is used to qualify a local part without a domain name. After
> > I installed OpenSMTPD it seems to me that this is also what OpenSMTPD
> > does. In other words, if I had mailname as gmail.com, a mail I send to
> > tombrowder (no domain name) would go to tombrow...@gmail.com.
>
> Brian, you may be right. I cannot find anything in the man pages, but
> it may have been stated during the installation of the Debian package.
> I'll uninstall and reinstall to see if I can capture the
> instructions...

Okay, here are the instructions, just as you said:

The "mail name" is used as the domain name in the email address for
messages that only have a "local part" (such as
  │  or ). It should be a fully qualified domain
name (FQDN) that you are entitled to use.
  │
  │ For instance, to allow the local host to generate mail with
addresses such as , set the
  │ system mail name to "example.org".

So, I guess I can trust opensmtpd to do the right thing if I set up
its configuration properly. And for the DNS just make every domain
have its own mail server (as long as everything eventually maps the
the correct server IP listening on port 25).

Thanks.

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 12:05 PM Brian  wrote:
> On Fri 21 Feb 2020 at 11:22:58 -0600, Tom Browder wrote:
>
> > On Fri, Feb 21, 2020 at 09:36 Brian  wrote:
...
> Thanks, I will read the page in detail later. Meanwhile, I did a quick
> search on the page for "mailname" and didn't get anything. Anyway, what
> I really wanted to know was what function *you* thought /etc/mailname
> played in OpenSMTPD because it would determine what you chose for it.
>
> On Exim it is used to qualify a local part without a domain name. After
> I installed OpenSMTPD it seems to me that this is also what OpenSMTPD
> does. In other words, if I had mailname as gmail.com, a mail I send to
> tombrowder (no domain name) would go to tombrow...@gmail.com.

Brian, you may be right. I cannot find anything in the man pages, but
it may have been stated during the installation of the Debian package.
I'll uninstall and reinstall to see if I can capture the
instructions...

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Brian]

On Fri 21 Feb 2020 at 11:22:58 -0600, Tom Browder wrote:

> On Fri, Feb 21, 2020 at 09:36 Brian  wrote:
> 
> > On Fri 21 Feb 2020 at 05:09:47 -0600, Tom Browder wrote:
> >
> ...
> | > I need to settle on names to define as the "mailname" for each the two
> 
> > > servers I will designate for the DNS MX records for all my mail-enabled
> > > domains. The mailnames should be "fully qualified domain names" (FQDNs)
> > so
> >
> ...
> 
> > I wouldn't use any of those with Exim because I am aware of the function
> > of /etc/mailname in its sending of mail. What does /etc/mailname do on
> > OpenSMTPD?
> 
> 
> The source of my info is in this link:
> 
> 
> https://poolp.org/posts/2019-09-14/setting-up-a-mail-server-with-opensmtpd-dovecot-and-rspamd/?utm_source=share_medium=ios_app_name=iossmf

Thanks, I will read the page in detail later. Meanwhile, I did a quick
search on the page for "mailname" and didn't get anything. Anyway, what
I really wanted to know was what function *you* thought /etc/mailname
played in OpenSMTPD because it would determine what you chose for it.

On Exim it is used to qualify a local part without a domain name. After
I installed OpenSMTPD it seems to me that this is also what OpenSMTPD
does. In other words, if I had mailname as gmail.com, a mail I send to
tombrowder (no domain name) would go to tombrow...@gmail.com.

Apologies if I have missed the point and this is something you have
factored into your planning.

-- 
Brian.

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 09:36 Brian  wrote:

> On Fri 21 Feb 2020 at 05:09:47 -0600, Tom Browder wrote:
>
...
| > I need to settle on names to define as the "mailname" for each the two

> > servers I will designate for the DNS MX records for all my mail-enabled
> > domains. The mailnames should be "fully qualified domain names" (FQDNs)
> so
>
...

> I wouldn't use any of those with Exim because I am aware of the function
> of /etc/mailname in its sending of mail. What does /etc/mailname do on
> OpenSMTPD?


The source of my info is in this link:


https://poolp.org/posts/2019-09-14/setting-up-a-mail-server-with-opensmtpd-dovecot-and-rspamd/?utm_source=share_medium=ios_app_name=iossmf

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Henning Follmann]

On Fri, Feb 21, 2020 at 08:47:36AM -0600, Tom Browder wrote:
> On Fri, Feb 21, 2020 at 08:37 Henning Follmann 
> wrote:
> 
> > On Fri, Feb 21, 2020 at 12:38:45PM +0100, john doe wrote:
> 
> ...
> 
> > Don't do THAT!
> >
> > RFC2181 section 10.3 says you can't point your MX record to a CNAME
> >
> 
> Maybe that's why Namecheap has the MXE record that points to an IP.
> 
> Besides the relevant RFCs, can anyone recommend a good book that covers all
> this. I have a couple of the O'Reilly books on "Postfix" and "DNS and Bind"
> and I am going back to refresh myself, but any other recs?
>

For years I operated a couple of sendmail installations. That and the
O'Reilly Sendmail book tought me a lot.
Today I prefer postfix. There is a german book from Peer Heinlein.
This is an excellent source if you want to set up a multidomain
mailserver completely driven by a LDAP directory.

-H

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

Re: Choice of "mailname" for mail server: suggestions welcome

[Brian]

On Fri 21 Feb 2020 at 05:09:47 -0600, Tom Browder wrote:

> I am preparing servers to use with OpenSMTPD and Sympa to provide mail and
> mailing list service.
> 
> I need to settle on names to define as the "mailname" for each the two
> servers I will designate for the DNS MX records for all my mail-enabled
> domains. The mailnames should be "fully qualified domain names" (FQDNs) so
> they will have names like:
> 
> + mail.example.com
> + smtp.example.com
> + mx.example.com
> 
> Does anyone have strong reasons to use one over another, or any other
> choice?

I wouldn't use any of those with Exim because I am aware of the function
of /etc/mailname in its sending of mail. What does /etc/mailname do on
OpenSMTPD?

-- 
Brian.

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 08:37 Henning Follmann 
wrote:

> On Fri, Feb 21, 2020 at 12:38:45PM +0100, john doe wrote:

...

> Don't do THAT!
>
> RFC2181 section 10.3 says you can't point your MX record to a CNAME
>

Maybe that's why Namecheap has the MXE record that points to an IP.

Besides the relevant RFCs, can anyone recommend a good book that covers all
this. I have a couple of the O'Reilly books on "Postfix" and "DNS and Bind"
and I am going back to refresh myself, but any other recs?

Thanks.

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Henning Follmann]

On Fri, Feb 21, 2020 at 12:38:45PM +0100, john doe wrote:
> On 2/21/2020 12:09 PM, Tom Browder wrote:
> > I am preparing servers to use with OpenSMTPD and Sympa to provide mail and
> > mailing list service.
> >
> > I need to settle on names to define as the "mailname" for each the two
> > servers I will designate for the DNS MX records for all my mail-enabled
> > domains. The mailnames should be "fully qualified domain names" (FQDNs) so
> > they will have names like:
> >
> > + mail.example.com
> > + smtp.example.com
> > + mx.example.com
> >
> > Does anyone have strong reasons to use one over another, or any other
> > choice?
> >
> 
> I might be missing something here but the domain name is what you
> buy/what you get.
> 
> If your planning to use split DNS, 'internal' and 'external' could be
> one option.
> 
> For the lists, 'lists.example.com' might not be a bad idea.
> 
> You can also use cname record to point to the same thing.
> 
> --
> John Doe
>


Don't do THAT!

RFC2181 section 10.3 says you can't point your MX record to a CNAME

-H


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

Re: Choice of "mailname" for mail server: suggestions welcome

[Joe]

On Fri, 21 Feb 2020 06:46:59 -0600
Tom Browder  wrote:

> On Fri, Feb 21, 2020 at 06:29 Joe  wrote:
> 
> > On Fri, 21 Feb 2020 06:00:14 -0600
> > Tom Browder  wrote:  
> 
> ...
> 
> > > As I understand it, a mail server has to use smtp to physically
> > > transfer mail between physical hosts and that requires one name
> > > that will resolve to an IP. Even if the server is hosting multiple
> > > domains, the mail for each still has to use the one "mailname" for
> > > transport.  
> >
> > Not really, you can have an A record for each domain, with the MX
> > record for each domain pointing to its 'own' A record. The only
> > common feature must be the IP address that the A records point to.
> >
> > You may well have only one PTR record for the address, although
> > multiple PTRs on one address are valid, but not many ISPs make
> > provision for that.
> >  
> 
> Interesting--I would prefer that. So with all mail stuff on one
> server, I can have each domain's MX record point to its own mail
> server. And I do have control of the PTR record. I just need to
> ensure everything resolves to the correct host IP.
> 
> So how does all that jibe with the single name required by OpenSMPTD?
> 

Don't know, I'm not familiar with that. But mail server software
shouldn't need to know anything about the DNS records used to route SMTP
mail to it. If so configured, it will look up DNS records of senders,
but not generally its own.

Exim4 has a default mailname, but to the best of my knowledge, it is
used only in the HELO/EHLO banner, and can be overridden for multiple
domains. HELO/EHLO is used to identify the server in logs, and need not
have any relationship to any email domain handled by the server, and is
often (incorrectly) configured as a bare domain name. It is best
configured to the same FQDN that the domain's primary MX record holds.

It is certainly supposed to be a FQDN, and it does need to be
resolvable in public DNS. My mail server, like most, is configured to
reject email from a sender whose HELO is not resolvable, and
particularly those whose HELO appears to be my own domain or public IP
address (yes, some malware does that). DNS servers aren't really
supposed to resolve bare domain names, but most do, aliasing to the
'www' A record, because many people are too lazy to type 'www' in what
should be a web FQDN.

-- 
Joe

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 06:29 Joe  wrote:

> On Fri, 21 Feb 2020 06:00:14 -0600
> Tom Browder  wrote:

...

> > As I understand it, a mail server has to use smtp to physically
> > transfer mail between physical hosts and that requires one name that
> > will resolve to an IP. Even if the server is hosting multiple
> > domains, the mail for each still has to use the one "mailname" for
> > transport.
>
> Not really, you can have an A record for each domain, with the MX
> record for each domain pointing to its 'own' A record. The only common
> feature must be the IP address that the A records point to.
>
> You may well have only one PTR record for the address, although
> multiple PTRs on one address are valid, but not many ISPs make provision
> for that.
>

Interesting--I would prefer that. So with all mail stuff on one server, I
can have each domain's MX record point to its own mail server. And I do
have control of the PTR record. I just need to ensure everything resolves
to the correct host IP.

So how does all that jibe with the single name required by OpenSMPTD?

Thanks, Joe.

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Jonas Smedegaard]

Quoting Tom Browder (2020-02-21 13:17:52)
> On Fri, Feb 21, 2020 at 06:00 Jonas Smedegaard  wrote:
> 
> > Hi tom,
> >
> ...
> 
> > > Does anyone have strong reasons to use one over another, or any other
> >
> > choice?
> >
> ...
> 
> > Depends on the purpose of the name(s).
> 
> ...
> 
> Thanks, Jonas, that makes good sense. Based on that I should use 
> "mail" and maybe "mail2" for my backup mail server.

Only if by "backup" you mean mirror of mail services generally - i.e. 
also for your users to connect to for fetching their mail when the 
primary server is down.

Otherwise, if you mean MX backup then I would use "mx2" for the backup 
host (and I would then consider naming the primary host _both_ "mail" 
and "mx1" so that I can use "mx1" and "mx2" for MX records.


> One of the reasons I asked was I know Gmail used to use something like 
> " smtp.gmail.com" for its smtp server and thought that might be 
> popular among sysadmins with such servers.

Lots of names are popular for various reasons.

Google has numerous hosts serving specifc services, likely with failover 
so that one hostname is even used for multiple hosts behind the scenes.

So if your setup is complex, then name each service, and number it too.

...but if you want simplicity, then beware that for each hostname you 
may (now or later) need to fumble with TLS certificates and/or DNSSEC 
signing keys.

...and beware that your users are not helped by hosts named by services 
but will likely find it geeky that they need to use "smpt2" to send and 
"pop3" to receive (unless of course they are all geeks, where they might 
prefer hosts named by characters in Tolkien books or Star Wars).


As Michael also mentioned, some mail clients blindly assume the world 
uses specific names for user-facing incoming and outgoing services, and 
probe those names before asking the user.  Personally I have found it 
least confusing for my users to tell them that "the server is 
mail-dot-our-domain for all user-facing services - both incoming and 
outgoing", and I then setup hints for those mail clients that wants to 
auto-configure.

 - Jonas

-- 
 * Jonas Smedegaard - idealist & Internet-arkitekt
 * Tlf.: +45 40843136  Website: http://dr.jones.dk/

 [x] quote me freely  [ ] ask before reusing  [ ] keep private

signature.asc
Description: signature

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 06:17 Michael Howard 
wrote:
...

> I don't know your use case but using 'regular' names like smtp.example.com,
> imap.example.com, pop3.example.com etc, help with the auto config
> processes used on devices, thus making it easier for users to setup
> accounts on their devices.
>

I'm not planning on anything at the moment but supporting mailing lists,
but you have a good point for any future expansion.

Thanks, Michael.

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Joe]

On Fri, 21 Feb 2020 06:00:14 -0600
Tom Browder  wrote:

> On Fri, Feb 21, 2020 at 05:39 john doe  wrote:
> ...
> 
> > I might be missing something here but the domain name is what you
> > buy/what you get.  
> 
> 
> Note I own (techically it's more of a rental or lease) multiple
> domains hosted across multiple servers which I also fully control (no
> sharing, full root control).
> 
> As I understand it, a mail server has to use smtp to physically
> transfer mail between physical hosts and that requires one name that
> will resolve to an IP. Even if the server is hosting multiple
> domains, the mail for each still has to use the one "mailname" for
> transport.

Not really, you can have an A record for each domain, with the MX
record for each domain pointing to its 'own' A record. The only common
feature must be the IP address that the A records point to.

You may well have only one PTR record for the address, although
multiple PTRs on one address are valid, but not many ISPs make provision
for that.

> 
> Hence my original question.
> 

I don't really see any reason either way, you can call it
fred.example.com if you wish, as long as the MX records point to it.

You may well have multiple A records for the address, anyway. An
important feature of a mail server is a complementary A-PTR pair, and
not all ISPs allow user setting of the PTR. So you may well have an
A-PTR pair on the IP address, both supplied by the ISP, which bear no
relation to any of your domain names. No problem.

If you have external account users, they will need to set up addresses
for both SMTP and IMAP/POP, and it is slightly more convenient to use
the one name for both, and 'mail' will then be more suitable. It's only
when you get fancy, with separate incoming and outgoing servers, that
you need distinct and meaningful names.

-- 
Joe

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 06:00 Jonas Smedegaard  wrote:

> Hi tom,
>
...

> > Does anyone have strong reasons to use one over another, or any other
>
> choice?
>
...

> Depends on the purpose of the name(s).

...

Thanks, Jonas, that makes good sense. Based on that I should use "mail" and
maybe "mail2" for my backup mail server.

One of the reasons I asked was I know Gmail used to use something like "
smtp.gmail.com" for its smtp server and thought that might be popular among
sysadmins with such servers.

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[Michael Howard]

On 21/02/2020 11:09, Tom Browder wrote:
I am preparing servers to use with OpenSMTPD and Sympa to provide mail 
and mailing list service.


I need to settle on names to define as the "mailname" for each the two 
servers I will designate for the DNS MX records for all my 
mail-enabled domains. The mailnames should be "fully qualified domain 
names" (FQDNs) so they will have names like:


+ mail.example.com <http://mail.example.com>
+ smtp.example.com <http://smtp.example.com>
+ mx.example.com <http://mx.example.com>

Does anyone have strong reasons to use one over another, or any other 
choice?


Thanks for any help.

Best regards,

-Tom



I don't know your use case but using 'regular' names like 
smtp.example.com, imap.example.com, pop3.example.com etc, help with the 
auto config processes used on devices, thus making it easier for users 
to setup accounts on their devices.


--
Michael Howard

Re: Choice of "mailname" for mail server: suggestions welcome

[Jonas Smedegaard]

Hi tom,

Quoting Tom Browder (2020-02-21 12:09:47)
> I am preparing servers to use with OpenSMTPD and Sympa to provide mail 
> and mailing list service.
> 
> I need to settle on names to define as the "mailname" for each the two 
> servers I will designate for the DNS MX records for all my 
> mail-enabled domains. The mailnames should be "fully qualified domain 
> names" (FQDNs) so they will have names like:
> 
> + mail.example.com
> + smtp.example.com
> + mx.example.com
> 
> Does anyone have strong reasons to use one over another, or any other 
> choice?

Depends on the purpose of the name(s).

If you run everything on a single host, then mail.example.com.

If you need to distinguish mail routing from other tasks (e.g. when 
running a spam filter on a different host and you want to tell other 
smtp servers to deliver to that instead of directly to your main mail 
server), then mx.example.com (or mx1.example.com) for that.

Similar for other names: Use protocol or other special name to emphasize 
that service, otherwise it is more confusing than helping.

 - Jonas

-- 
 * Jonas Smedegaard - idealist & Internet-arkitekt
 * Tlf.: +45 40843136  Website: http://dr.jones.dk/

 [x] quote me freely  [ ] ask before reusing  [ ] keep private

signature.asc
Description: signature

Re: Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

On Fri, Feb 21, 2020 at 05:39 john doe  wrote:
...

> I might be missing something here but the domain name is what you
> buy/what you get.


Note I own (techically it's more of a rental or lease) multiple domains
hosted across multiple servers which I also fully control (no sharing, full
root control).

As I understand it, a mail server has to use smtp to physically transfer
mail between physical hosts and that requires one name that will resolve to
an IP. Even if the server is hosting multiple domains, the mail for each
still has to use the one "mailname" for transport.

Hence my original question.

-Tom

Re: Choice of "mailname" for mail server: suggestions welcome

[john doe]

On 2/21/2020 12:09 PM, Tom Browder wrote:
> I am preparing servers to use with OpenSMTPD and Sympa to provide mail and
> mailing list service.
>
> I need to settle on names to define as the "mailname" for each the two
> servers I will designate for the DNS MX records for all my mail-enabled
> domains. The mailnames should be "fully qualified domain names" (FQDNs) so
> they will have names like:
>
> + mail.example.com
> + smtp.example.com
> + mx.example.com
>
> Does anyone have strong reasons to use one over another, or any other
> choice?
>

I might be missing something here but the domain name is what you
buy/what you get.

If your planning to use split DNS, 'internal' and 'external' could be
one option.

For the lists, 'lists.example.com' might not be a bad idea.

You can also use cname record to point to the same thing.

--
John Doe

Choice of "mailname" for mail server: suggestions welcome

[Tom Browder]

I am preparing servers to use with OpenSMTPD and Sympa to provide mail and
mailing list service.

I need to settle on names to define as the "mailname" for each the two
servers I will designate for the DNS MX records for all my mail-enabled
domains. The mailnames should be "fully qualified domain names" (FQDNs) so
they will have names like:

+ mail.example.com
+ smtp.example.com
+ mx.example.com

Does anyone have strong reasons to use one over another, or any other
choice?

Thanks for any help.

Best regards,

-Tom

Re: Attributing (was: Choice of VMs under i386 Stretch?)

[David Wright]

On Sat 06 Jul 2019 at 11:16:19 (-0400), Stefan Monnier wrote:
> >> (General observation: it's really quite annoying that you remove all 
> >> attribution when you quote previous emails in your replies.)
> > It really is very annoying, primarily because it's intentional, and so
> > intentionally annoying, which is really, really annoying.
> 
> Interesting.  I never read attributions so find them useless, and most
> of the time my intention is to reply to the particular chunk of text
> that I quote regardless of who wrote it.  Furthermore in the rare cases
> where it's important to know who wrote it, it should be trivial to
> lookup the parents in the thread.
> [ I do find myself having to look at the parents in a thread fairly
>   often, tho not to figure out who wrote it but in order to get more
>   context to better understand what was meant.  ]

If you're reading this list from emails, and are trying to keep
mailboxes at a reasonable size, then it might *not* be trivial to
look up the thread. And that's the basis of netiquette: thinking
of other people in the group.

Hence, we don't top-post, try to quote sensibly sized extracts,
respect quoting syntax conventions, and avoid breaking the
Message-ID threading where possible, and so on.

In a technical list like this, attributions help greatly in weighing the
words of the contributions. Over a period, people build up a picture
of the various contributors' expertise. (Sorry, but I have none.)

Cheers,
David.

Re: Attributing (was: Choice of VMs under i386 Stretch?)

[Curt]

On 2019-07-06, Stefan Monnier  wrote:
>>> (General observation: it's really quite annoying that you remove all 
>>> attribution when you quote previous emails in your replies.)
>> It really is very annoying, primarily because it's intentional, and so
>> intentionally annoying, which is really, really annoying.
>
> Interesting.  I never read attributions so find them useless, and most

It has nothing to do with you finding them useless or not and everything
to do with everybody else reading you, to whom you should show a minimum
of consideration.

https://wiki.debian.org/DebianMailingLists

 Attribute quotes accurately.

You've been told this before, BTW, at least once (it's in the archives) as
you appear to be expressing an element of "surprise" or something.


-- 
"These findings demonstrate that under appropriate conditions the isolated,
intact large mammalian brain possesses an underappreciated capacity for
restoration of microcirculation and molecular and cellular activity after a
prolonged post-mortem interval." From a recent article in *Nature*. Holy shit. 

Attributing (was: Choice of VMs under i386 Stretch?)

[Stefan Monnier]

>> (General observation: it's really quite annoying that you remove all 
>> attribution when you quote previous emails in your replies.)
> It really is very annoying, primarily because it's intentional, and so
> intentionally annoying, which is really, really annoying.

Interesting.  I never read attributions so find them useless, and most
of the time my intention is to reply to the particular chunk of text
that I quote regardless of who wrote it.  Furthermore in the rare cases
where it's important to know who wrote it, it should be trivial to
lookup the parents in the thread.
[ I do find myself having to look at the parents in a thread fairly
  often, tho not to figure out who wrote it but in order to get more
  context to better understand what was meant.  ]


Stefan

Re: Choice of VMs under i386 Stretch?

[Richard Owlett]

On 07/04/2019 08:50 AM, Richard Owlett wrote:

On 07/02/2019 04:05 PM, Linux-Fan wrote:

...

For my local purposes, I have created my own metapackage (not in 
Debian...)




Please send me a copy.
I've a ~10 year old Lenovo T510 whose hardware should be new enough.



Apparently runs fine. No need to add to my hardware pile.

Re: Choice of VMs under i386 Stretch?

[Curt]

On 2019-07-04, Michael Stone  wrote:
> On Thu, Jul 04, 2019 at 11:26:39AM -0400, Stefan Monnier wrote:
>>> If it had been done 10 years ago it wouldn't need to be done now. :)
>
> (General observation: it's really quite annoying that you remove all 
> attribution when you quote previous emails in your replies.)
>

It really is very annoying, primarily because it's intentional, and so
intentionally annoying, which is really, really annoying.

-- 
"These findings demonstrate that under appropriate conditions the isolated,
intact large mammalian brain possesses an underappreciated capacity for
restoration of microcirculation and molecular and cellular activity after a
prolonged post-mortem interval." in Nature. 

Re: Choice of VMs under i386 Stretch?

[Michael Stone]

On Thu, Jul 04, 2019 at 11:26:39AM -0400, Stefan Monnier wrote:

If it had been done 10 years ago it wouldn't need to be done now. :)


(General observation: it's really quite annoying that you remove all 
attribution when you quote previous emails in your replies.)



VMs running on a core 2 definitely have unfixable security issues and aren't
great performers.  That certainly limits the use cases.


AFAIK all CPUs since also have unfixable security issues (spectre).
So I'm not sure it makes much difference.


That's not true, but I really don't want to start another thread against 
CPU FUD so I'll leave it at that.



As for efficiency, AFAIK the production of a laptop takes more
energy than all the electricity it uses in its lifetime, so using my old
laptop is arguably more efficient.


That's a bogus argument when applied to secondhand hardware. When 
comparing two pieces of used hardware and effectively deciding which one 
gets thrown away, there's no particular virtue in choosing the older 
(probably less power efficient) one.



I only pointed out that running a VM in a 32bit system is not silly at all.


Humans do a lot of things that are objectively silly for their own 
amusement.

Re: Choice of VMs under i386 Stretch?

[Stefan Monnier]

> If it had been done 10 years ago it wouldn't need to be done now. :)

[ The initial install was in 2003, FWIW.  ]
I do use 64bit Debian on another machine, but to tell you the truth,
I don't notice any difference at all (other than bigger hex numbers in
GDB which take up more screen real estate and are harder to
remember/decipher ;-), so there's a serious lack of motivation.

> VMs running on a core 2 definitely have unfixable security issues and aren't
> great performers.  That certainly limits the use cases.

AFAIK all CPUs since also have unfixable security issues (spectre).
So I'm not sure it makes much difference.
In any case, my point is that often it's best to use what you have
rather than what is "best".

> Since faster and more efficient hardware can literally be found for free

Regarding faster: my laptop is a Thinkpad T61 with a hand-upgraded Core
2 CPU, and the reason I'm using that is that it's the best 4/3 laptop
I could find.
All the more recent models use short screens and hence don't qualify.

It does deserve to be upgraded, and I'm willing to spend a fair bit of
money for that, but so far noone has decided to re-enter the 4/3
marketplace AFAIK (the "thinkpad" T70 has been the only exception I know
of, but it was very short run).

As for efficiency, AFAIK the production of a laptop takes more
energy than all the electricity it uses in its lifetime, so using my old
laptop is arguably more efficient.

> these days, running on something that old has a strong component of personal
> choice.

Just like choosing something more recent is a personal choice, yes.
With global ecological implications.

> Which is fine, but not a strong argument. Put a different way, it's
> fine if that's what you want to do, but terrible advice for others.

I didn't advise it.  I only pointed out that running a VM in a 32bit
system is not silly at all.


Stefan

Re: Choice of VMs under i386 Stretch?

[Nate Bargmann]

* On 2019 04 Jul 08:51 -0500, Richard Owlett wrote:
> I've a ~10 year old Lenovo T510 whose hardware should be new enough.

I've a T410 running Buster amd64 and have simply installed the
VirtualBox binaries from the Sid repository (manually downloaded and
installed).  I also have QEMU installed for testing some ARM stuff.
Works like a charm.

> BTW my oldest machine is a Kaypro 10 ;/

I'm not that old, my original was a Color Computer 2 a bit over 35 years
ago.  The oldest machine I have now is an IBM T23.  Anything older was
given away or scrapped.

- Nate

-- 

"The optimist proclaims that we live in the best of all
possible worlds.  The pessimist fears this is true."

Web: https://www.n0nb.us  GPG key: D55A8819  GitHub: N0NB


signature.asc
Description: PGP signature

Old computers (Re: Choice of VMs under i386 Stretch?)

[rhkramer]

On Thursday, July 04, 2019 09:50:36 AM Richard Owlett wrote:
> BTW my oldest machine is a Kaypro 10 ;/

Just for kicks, I'll mention that my oldest machine is a Digital Group Z-80, 
circa 1976, assembed from a kit, and with 2K on board RAM plus 2 auxillary 
memory boards with 8K each (iirc) for a total of 18K RAM.  Long term storage 
was via a cassete recorder / player with a speed control to adjust playback 
speed for best result.

I programmed some floating point routines in machine language, and had thoughts 
of writing things like a word processor.

I thought (seriously) about buying (or, more accurately, wished that I could 
afford to buy) a used teletype machine (for about $1500) in order to have a 
means of printing.

It (the machine) was working when I retired it, but I should make some fixes if 
I ever un-retired it (I took out the 115 volt AC cooling fan to use as an air 
mover between two rooms of my house),

Aside: the Digital Group machines were pretty neat in that the CPU was on a 
daugherboard, and they made CPU boards with the Z-80, the 6502, the 6800 
(iirc) (and maybe just the plain 8080) -- you could "convert" to any of those 
other CPUs by just buying, assembling, and changing out the CPU daughterboard.

BTW, that wasn't my first computer -- when I was significantly younger (i.e., 
10 
to 15 years before 1976), I acquired two other things (both long gone, I 
think) that were called computers:

   * One was sort of a toy analog computer (one where you input numbers by 
doing things like using potentiometers to set a voltage to represent a 
number), and then read results on a meter (I'm somewhat oversimplifying that 
as I don't really remember much about that.

   * The other was something that I believed had computer in the name, but was 
used to control model trains -- the intent was that you could speak into a 
(built-in) microphone and it was supposed to recognize spoken commands (things 
like "stop" or "go" -- and maybe I'm misremembering the commands, maybe the 
logic was based on detecting and counting syllables in the commands).  It 
never worked very well, but I might actually still have it buried in a box of 
old model train stuff. ;-)

By the way, despite the years mentioned in this email, I am only 29 (or, at 
least, I try to convince women of that) (the same age (or was it 39) that Jack 
Benny always claimed to be, even when he appeared to be much older,

(I woke up with a headache today, and am not "in shape" to do any serious 
work, so I'd doing things like this).

Happy 4th of July (Independence Day) to all of us (in the USA or not).  I hope 
the soldiers that might be marching in Trump's July 4th celebration are not 
"goose stepping".

Re: Choice of VMs under i386 Stretch?

[Richard Owlett]

On 07/02/2019 04:05 PM, Linux-Fan wrote:

...

For my local purposes, I have created my own metapackage (not in Debian...)
-- this is why I suggested a list of packages: I expect that installing all
of the packages yields the same result (except for automatically vs.
manually installed packages...) as if a metapackage were used. In case 
it is

of use to you, I can send you an e-mail with a copy of my metapackage
(it is about 4 KiB as per `du -sh`, but it does not do well for world-wide
publication because it is not Debian Policy compliant etc. thus not
uploaded/attached anywhere yet).



Please send me a copy.
I've a ~10 year old Lenovo T510 whose hardware should be new enough.

BTW my oldest machine is a Kaypro 10 ;/

Re: Oldest Usable x86 CPU for Stretch (Was: Choice of VMs under i386 Stretch?)

[andreimpopescu]

On Jo, 04 iul 19, 01:47:58, Matthew Crews wrote:
> 
> You know, this got me thinking. What *is* the oldest 32-bit x86 CPU that
> we can use in Stretch for a VM host? And assuming we are talking
> out-of-the-box experience, ie I download a standard ISO and fire it up.

This might get you started:
https://www.debian.org/releases/stretch/i386/release-notes/ch-information.en.html#i386-is-now-almost-i686

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


signature.asc
Description: PGP signature

Oldest Usable x86 CPU for Stretch (Was: Choice of VMs under i386 Stretch?)

[Matthew Crews]

On 7/3/19 10:20 AM, Stefan Monnier wrote:
>> That said I do not believe that any existing i386 32-bit-only hardware
>> that is still floating around even supports the virtual machine
>> extensions necessary to run a true VM host.
> 
> I haven't use qemu on my 32bit only i686 machines recently, but I see no
> reason why it wouldn't work any more.

You know, this got me thinking. What *is* the oldest 32-bit x86 CPU that
we can use in Stretch for a VM host? And assuming we are talking
out-of-the-box experience, ie I download a standard ISO and fire it up.

And to be honest, I don't know. Debian seems to support as old as the
Pentium Pro, but I'm not sure what compiler flags that we use, ie what
CPU extensions are required. I remember a couple of years ago I had to
dig up a very old version of Puppy Linux (with a very old Kernel
version) to run an AMD K6-2 laptop because of missing CPU extensions. So
I'm wondering if the kernel that ships with Debian is the same.

Any ideas?

Note: I don't care about the practicality of this.

-Matt



signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Michael Stone]

On Wed, Jul 03, 2019 at 02:05:05PM -0400, Stefan Monnier wrote:

- start by reinstall Debian using a 64bit system this time


If it had been done 10 years ago it wouldn't need to be done now. :) If 
you just don't want to bother that's fine, but not great advice for 
anyone starting out.



In my experience VMs running in a 32bit Debian system on a Core
2 processor (i.e. one of the first amd64-compatible CPU from Intel)
offer adequate performance.  I probably wouldn't use that for a server
under any real load, but VMs can be useful in all kinds of situations.


VMs running on a core 2 definitely have unfixable security issues and 
aren't great performers. That certainly limits the use cases.



It's certainly possible to run software-only virtualization on an
ancient CPU, but at that point (since it would clearly just be to
scratch an itch, not for any practical reason) why ask for opinions
instead of just playing around if playing around is the only goal?


I get the impression that by "if playing around is the only goal" you
mean something like "if it's not within the context of a commercial
deployment".  Sometimes the need is very real (not just for playing
around) but doesn't have anything to do with maximizing hardware
utilization in a datacenter.


Since faster and more efficient hardware can literally be found for free 
these days, running on something that old has a strong component of 
personal choice. Which is fine, but not a strong argument. Put a 
different way, it's fine if that's what you want to do, but terrible 
advice for others. 

Re: Choice of VMs under i386 Stretch?

[Stefan Monnier]

> Because the hardware features to permit efficient virtualization weren't
> available on i386-only CPUs. (And there's really no good reason to run a VM
> host [vs guest] in i386 mode if it can run in amd64 mode.)

By "there's really no good ..." I think you mean "I can't think of any
good ...".

For reference, here's the reason why I did it last time: I needed to
update the small Debian install I have on my USB rescue key.  I had
a few choices:
- boot inside a qemu on my 32bit Debian system.
- reboot my system into the rescue key (hence losing the current state
  of my session and all that stuff and being unable to use my machine
  for anything else while that rescue system was updating itself).
- do the same but hoping that hibernate+resume will work, so while
  I still can't use the machine during the upgrade, at least maybe
  I didn't lose all my session's state (tho resume often fails, in my
  experience).
- start by reinstall Debian using a 64bit system this time, just on the
  hunch that maybe it'll be marginally faster at running the upgrade
  within the VM (tho this is normally limited mostly by the USB key
  write speed).

Now, which of those choices sounds more silly to you?

In my experience VMs running in a 32bit Debian system on a Core
2 processor (i.e. one of the first amd64-compatible CPU from Intel)
offer adequate performance.  I probably wouldn't use that for a server
under any real load, but VMs can be useful in all kinds of situations.

> It's certainly possible to run software-only virtualization on an
> ancient CPU, but at that point (since it would clearly just be to
> scratch an itch, not for any practical reason) why ask for opinions
> instead of just playing around if playing around is the only goal?

I get the impression that by "if playing around is the only goal" you
mean something like "if it's not within the context of a commercial
deployment".  Sometimes the need is very real (not just for playing
around) but doesn't have anything to do with maximizing hardware
utilization in a datacenter.


Stefan

Re: Choice of VMs under i386 Stretch?

[Michael Stone]

On Wed, Jul 03, 2019 at 01:20:06PM -0400, Stefan Monnier wrote:

think in this day and age it is a bit silly to try and run a VM on a
32-bit host


Why?  Often the question is not "which hardware should I pick to run
this VM" but "what kinds of technology should I use to run this software
on this hardware".  When "this hardware" is 32bit, and a VM is needed
between the two, then that's that.


Because the hardware features to permit efficient virtualization weren't 
available on i386-only CPUs. (And there's really no good reason to run a 
VM host [vs guest] in i386 mode if it can run in amd64 mode.)



That said I do not believe that any existing i386 32-bit-only hardware
that is still floating around even supports the virtual machine
extensions necessary to run a true VM host.


I haven't use qemu on my 32bit only i686 machines recently, but I see no
reason why it wouldn't work any more.


It's certainly possible to run software-only virtualization on an 
ancient CPU, but at that point (since it would clearly just be to
scratch an itch, not for any practical reason) why ask for opinions 
instead of just playing around if playing around is the only goal?

Re: Choice of VMs under i386 Stretch?

[Stefan Monnier]

> I have an innate desire to help people, but more importantly I give
> people the benefit of the doubt. Besides I self-taught myself a few
> things along the way, so I consider it a win.

Yes, please (and please remind me of that as well when I fail to follow
it ;-)

>> Now, which one of you is going to tell him that running virtual
>> machines is a bit of a stretch on a 32-bit host?

Huh?  The only systems on which I have ever run virtual machines were
all 32bit Debian systems.  Now admittedly, I tend to count LXC as a VM,
so maybe this is imprecise, but even w.r.t actual VMs they've all
been 32bit.

> think in this day and age it is a bit silly to try and run a VM on a
> 32-bit host

Why?  Often the question is not "which hardware should I pick to run
this VM" but "what kinds of technology should I use to run this software
on this hardware".  When "this hardware" is 32bit, and a VM is needed
between the two, then that's that.

What might be silly is to think that this choice can't be the best one.

> (or for that matter, run a 32-bit host at all if your
> hardware supports 64-bit,

That's what I do on half of my machines, yes (the other half is a mix
of 32bit-only and 64bit-only systems).

> That said I do not believe that any existing i386 32-bit-only hardware
> that is still floating around even supports the virtual machine
> extensions necessary to run a true VM host.

I haven't use qemu on my 32bit only i686 machines recently, but I see no
reason why it wouldn't work any more.

> Containers like Docker?  Sure, those should still work,

Of course they do.  Just like `chroot` (of which they're basically an
extension) they require no special hardware support of any kind.


Stefan

Re: Choice of VMs under i386 Stretch?

[Greg Wooledge]

On Wed, Jul 03, 2019 at 05:01:11AM +0200, Matthew Crews wrote:
> I have no idea what an Owlett thread is,

It's what you're seeing right now.

It's also the reason I end up writing a filtering program to send all
email from certain From: addresses into a spam folder.

Re: Choice of VMs under i386 Stretch?

[tomas]

On Wed, Jul 03, 2019 at 08:17:22AM -, Curt wrote:
> On 2019-07-03,   wrote:
> >
> >> I have an innate desire to help people, but more importantly I give
> >> people the benefit of the doubt. Besides I self-taught myself a few
> >> things along the way, so I consider it a win.
> >
> > I prefer that one, too :)
> 
> People get the benefit of the doubt until they remove all doubt, at
> which point your easy altruism reverses itself into a kind of pandering
> to the selfishness you pretend to decry. 

Let's agree to differ, then.

Cheers
-- t


signature.asc
Description: Digital signature

Re: Choice of VMs under i386 Stretch?

[Curt]

On 2019-07-03,   wrote:
>
>> I have an innate desire to help people, but more importantly I give
>> people the benefit of the doubt. Besides I self-taught myself a few
>> things along the way, so I consider it a win.
>
> I prefer that one, too :)

People get the benefit of the doubt until they remove all doubt, at
which point your easy altruism reverses itself into a kind of pandering
to the selfishness you pretend to decry. 

> Cheers

Re: Choice of VMs under i386 Stretch?

[tomas]

On Wed, Jul 03, 2019 at 05:01:11AM +0200, Matthew Crews wrote:
> On 7/2/19 1:20 PM, Andy Smith wrote:
> > I do feel sorry for you Matthew. You have been enticed into spending
> > considerable time giving a thorough answer in an Owlett thread.
> > Unfortunately Owlett threads are either an ongoing Internet
> > performance art project or a result of severe mental illness (why
> > not both!?), not sincere requests for help.

While this may soundsomewhat funny, it is pretty unfriendly: you
never know how cranky others might find *you* -- so better hope
for some slack in your favour :-)

> I have an innate desire to help people, but more importantly I give
> people the benefit of the doubt. Besides I self-taught myself a few
> things along the way, so I consider it a win.

I prefer that one, too :)

Cheers
-- t


signature.asc
Description: Digital signature

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 7/2/19 1:20 PM, Andy Smith wrote:
> I do feel sorry for you Matthew. You have been enticed into spending
> considerable time giving a thorough answer in an Owlett thread.
> Unfortunately Owlett threads are either an ongoing Internet
> performance art project or a result of severe mental illness (why
> not both!?), not sincere requests for help.

I have an innate desire to help people, but more importantly I give
people the benefit of the doubt. Besides I self-taught myself a few
things along the way, so I consider it a win.

I have no idea what an Owlett thread is, other than it sounds like a
Pokémon or one of the characters from the cartoon PJ Masks. Or is it
just another name for good old fashioned trolling?

> Now, which one of you is going to tell him that running virtual
> machines is a bit of a stretch on a 32-bit host?
> 
> Better luck next time! :)

I'm not going to discount that someone has a perfectly good reason for
wanting to do this, even if it is for academic purposes. Granted, I
think in this day and age it is a bit silly to try and run a VM on a
32-bit host (or for that matter, run a 32-bit host at all if your
hardware supports 64-bit, but that is another topic).

That said I do not believe that any existing i386 32-bit-only hardware
that is still floating around even supports the virtual machine
extensions necessary to run a true VM host. Containers like Docker?
Sure, those should still work, but I'm not an expert in the subject.

-Matt



signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Linux-Fan]

Richard Owlett writes:


On 07/01/2019 01:43 PM, Linux-Fan wrote:

Richard Owlett writes:

[...]


I do not understand.


It was an attempt to give you a list of packages that may allow you to start
using VMs without further checks ...


That your answer was "a list of packages" is key to the communication
problem.

A restatement of my question might be:

I run the i386 version of Debian 9.8.
Using only contents of that set of installation DVDs, I wish to use a VM
host capable of running multiple VM guests. Although the guests will be
running in command line mode, the host should have a GUI.
What choice metapackages provide that function?


Thank you very much for the rephrasing -- the use of `host` and `guest`
aids the description very well.

TL;DR: There is (AFAICT) no metapackage satisfying your requirements.

* * *

The long version:

As far as I can tell, there is no Debian metapackage to install my preferred
virt-manager (GUI) + libvirt (backend) + KVM (hypervisor) setup:

$ apt rdepends virt-manager
virt-manager
Reverse Depends:
  Depends: mdvl-virtualization

$ aptitude show mdvl-virtualization
Package: mdvl-virtualization
Version: 1.0.0
New: yes
State: installed
Automatically installed: yes
Priority: optional
Section: admin
Maintainer: Linux-Fan 
Architecture: all
Uncompressed Size: 0
Depends: virt-manager, qemu, qemu-kvm, libvirt-clients, libvirt0, 
libvirt-daemon,
 libvirt-daemon-system
Description: MDVL Virtualization efforts
 This metapackage installs common virtualization technology for MDVL, 
i.e. KVM and frontend.

For my local purposes, I have created my own metapackage (not in Debian...)
-- this is why I suggested a list of packages: I expect that installing all
of the packages yields the same result (except for automatically vs.
manually installed packages...) as if a metapackage were used. In case it is
of use to you, I can send you an e-mail with a copy of my metapackage
(it is about 4 KiB as per `du -sh`, but it does not do well for world-wide
publication because it is not Debian Policy compliant etc. thus not
uploaded/attached anywhere yet).

HTH
Linux-Fan


pgpgBLFXwZZ9u.pgp
Description: PGP signature

Re: Choice of VMs under i386 Stretch?

[Andy Smith]

Hello Curt and Matthew,

On Tue, Jul 02, 2019 at 12:04:36PM -, Curt wrote:
> On 2019-07-02, Richard Owlett  wrote:
> >
> > A restatement of my question might be:

[…]

> What do we win if we provide the correct answer? A year's supply of
> invective?

I do feel sorry for you Matthew. You have been enticed into spending
considerable time giving a thorough answer in an Owlett thread.
Unfortunately Owlett threads are either an ongoing Internet
performance art project or a result of severe mental illness (why
not both!?), not sincere requests for help.

Now, which one of you is going to tell him that running virtual
machines is a bit of a stretch on a 32-bit host?

Better luck next time! :)

Cheers,
Andy

-- 
https://bitfolk.com/ -- No-nonsense VPS hosting

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 7/2/19 8:34 AM, Stefan Monnier wrote:
>>> I'm partial to VirtualBox.   Is that on any of the Debian DVD'S?
>> No, because it isn't free software.
>> See  for details.
> 
> I believe what you wrote is slightly misleading: the base VirtualBox
> software seems to satisfy the definition of Free Software just fine.
> The above wiki page mentions two problems for Debian packaging:
> - There's a proprietary "add-on" for it, but that doesn't affect the
>   status of the base system. 

Technically Debian places Virtualbox in Contrib since version 4.2
because a non-free component is required to build the BIOS.

> - There's a lack of interest from the main developers to provide
>   long-term support for older releases which made it too much trouble
>   for Debian's security team to provide support for it, so it's not in
>   Debian any more.

This is the primary driver as to why Stretch places Virtualbox in
backports, and why Buster will not have Virtualbox packages in the
Debian repo at all.

Additionally, Oracle no longer supports i386 versions of Virtualbox, not
since version 6.0.0.

For that reason I would not recommend using Virtualbox unless you are
intending to use the Oracle-supplied packages, and I certainly would not
recommend using the i386 versions.

-Matt



signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 7/2/19 8:34 AM, Stefan Monnier wrote:
>>> I'm partial to VirtualBox.   Is that on any of the Debian DVD'S?
>> No, because it isn't free software.
>> See  for details.
>
> I believe what you wrote is slightly misleading: the base VirtualBox
> software seems to satisfy the definition of Free Software just fine.
> The above wiki page mentions two problems for Debian packaging:
> - There's a proprietary "add-on" for it, but that doesn't affect the
>   status of the base system.

Technically Debian places Virtualbox in Contrib since version 4.2
because a non-free component is required to build the BIOS.

> - There's a lack of interest from the main developers to provide
>   long-term support for older releases which made it too much trouble
>   for Debian's security team to provide support for it, so it's not in
>   Debian any more.

This is the primary driver as to why Stretch places Virtualbox in
backports, and why Buster will not have Virtualbox packages in the
Debian repo at all.

Additionally, Oracle no longer supports i386 versions of Virtualbox, not
since version 6.0.0.

For that reason I would not recommend using Virtualbox unless you are
intending to use the Oracle-supplied packages, and I certainly would not
recommend using the i386 versions.

-Matt





signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 7/2/19 8:34 AM, Stefan Monnier wrote:
>>> I'm partial to VirtualBox.   Is that on any of the Debian DVD'S?
>> No, because it isn't free software.
>> See  for details.
>
> I believe what you wrote is slightly misleading: the base VirtualBox
> software seems to satisfy the definition of Free Software just fine.
> The above wiki page mentions two problems for Debian packaging:
> - There's a proprietary "add-on" for it, but that doesn't affect the
>   status of the base system.

Technically Debian places Virtualbox in Contrib since version 4.2
because a non-free component is required to build the BIOS.

> - There's a lack of interest from the main developers to provide
>   long-term support for older releases which made it too much trouble
>   for Debian's security team to provide support for it, so it's not in
>   Debian any more.

This is the primary driver as to why Stretch places Virtualbox in
backports, and why Buster will not have Virtualbox packages in the
Debian repo at all.

Additionally, Oracle no longer supports i386 versions of Virtualbox, not
since version 6.0.0.

For that reason I would not recommend using Virtualbox unless you are
intending to use the Oracle-supplied packages, and I certainly would not
recommend using the i386 versions.

-Matt





signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Stefan Monnier]

>> I'm partial to VirtualBox.   Is that on any of the Debian DVD'S?
> No, because it isn't free software.
> See  for details.

I believe what you wrote is slightly misleading: the base VirtualBox
software seems to satisfy the definition of Free Software just fine.
The above wiki page mentions two problems for Debian packaging:
- There's a proprietary "add-on" for it, but that doesn't affect the
  status of the base system. 
- There's a lack of interest from the main developers to provide
  long-term support for older releases which made it too much trouble
  for Debian's security team to provide support for it, so it's not in
  Debian any more.


Stefan

Re: Choice of VMs under i386 Stretch?

[Greg Wooledge]

On Tue, Jul 02, 2019 at 11:17:06AM -0400, Kenneth Parker wrote:
> I'm partial to VirtualBox.   Is that on any of the Debian DVD'S?

No, because it isn't free software.

See  for details.

Re: Choice of VMs under i386 Stretch?

[Kenneth Parker]

On Tue, Jul 2, 2019, 8:05 AM Curt  wrote:

> On 2019-07-02, Richard Owlett  wrote:
> >
> > A restatement of my question might be:
> >
> > I run the i386 version of Debian 9.8.
> > Using only contents of that set of installation DVDs, I wish to use a VM
> > host capable of running multiple VM guests. Although the guests will be
> > running in command line mode, the host should have a GUI.
> > What choice metapackages provide that function?
>

I'm partial to VirtualBox.   Is that on any of the Debian DVD'S?

>
> What do we win if we provide the correct answer? A year's supply of

invective?
>

An Invection Oven?:-D

Kenneth Parker

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 7/2/19 6:04 AM, Matthew Crews wrote:

> To determine *which* installation DVD contains one of these programs,
> you will need to look at the individual wiki page for each VM host
> program, see which package or packages you need, and check the DVD .list
> files to see which DVD contains that package [2]. You should also be
> aware that each package may have untold number of dependencies, so I
> suggest you look at each individual package [3], figure out the
> dependencies, and then figure out which dvd those dependencies are on.

Addendum:

An alternative method is to look at each individual package, figure out
which packages you need, download the .deb files for each package, put
them on removable media (CD-R/RW, DVD-R/RW, or a usb flash drive), put
the removable media in your host machine, and install from the removable
media.

On an internet-capable host, you can download the .deb for apps foo and
bar this way (without root access):

$ apt download foo bar

make sure you pay attention to the file name afterwards:

$ pwd
/path/to/file
$ ls *.deb
foo_3.14.deb  bar_2.18.deb

To install a .deb file, you can either use apt to point to the file:

$ sudo apt install /path/to/file/foo_3.14.deb /path/to/file/bar_2.18.deb

Or you can use dpkg:

$ sudo dpkg -i /path/to/file/foo_3.14.deb /path/to/file/bar_2.18.deb

-Matt



signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 7/2/19 4:30 AM, Richard Owlett wrote:

> A restatement of my question might be:
> 
> I run the i386 version of Debian 9.8.
> Using only contents of that set of installation DVDs, I wish to use a VM
> host capable of running multiple VM guests. Although the guests will be
> running in command line mode, the host should have a GUI.
> What choice metapackages provide that function?

You have your answer already in previous emails.

The Wiki provides a list of software capable of turning Debian into a VM
host [1]. They include full hardware virtualization such as Qemu,
containers such as Docker, and pseudo-virtualization such as Schroot.
Its up to you to determine which one best suits your needs, as all of
these VM host programs can do what you want.

1:
https://wiki.debian.org/SystemVirtualization#Using_Debian_to_host_Virtual_Computers

To determine *which* installation DVD contains one of these programs,
you will need to look at the individual wiki page for each VM host
program, see which package or packages you need, and check the DVD .list
files to see which DVD contains that package [2]. You should also be
aware that each package may have untold number of dependencies, so I
suggest you look at each individual package [3], figure out the
dependencies, and then figure out which dvd those dependencies are on.

2: https://cdimage.debian.org/cdimage/release/current/i386/list-dvd/
3: https://packages.debian.org/

Is this a lot of work? Yes.

Normally I would recommend VirtualBox for its ease of use, but
Virtualbox is *not* found on the installation DVDs.

Depending on your hardware, native virtualization might not even be
possible. I suggest consulting your CPU manufacturer to determine if it
is. If you are using an Intel CPU, you can visit their product page to
see if it supports VT-x [4]. I can't speak for AMD CPUs or other brand
CPUs, though it should be revealed in the output of lscpu

4: https://ark.intel.com

If native virtualization is not available, you may be required to use a
container or a pseudo-virtualization option.

-Matt



signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Curt]

On 2019-07-02, Richard Owlett  wrote:
>
> A restatement of my question might be:
>
> I run the i386 version of Debian 9.8.
> Using only contents of that set of installation DVDs, I wish to use a VM 
> host capable of running multiple VM guests. Although the guests will be 
> running in command line mode, the host should have a GUI.
> What choice metapackages provide that function?

What do we win if we provide the correct answer? A year's supply of
invective?

Re: Choice of VMs under i386 Stretch?

[Richard Owlett]

On 07/01/2019 01:43 PM, Linux-Fan wrote:

Richard Owlett writes:

[...]


I do not understand.


It was an attempt to give you a list of packages that may allow you to 
start using VMs without further checks ...




That your answer was "a list of packages" is key to the communication 
problem.


A restatement of my question might be:

I run the i386 version of Debian 9.8.
Using only contents of that set of installation DVDs, I wish to use a VM 
host capable of running multiple VM guests. Although the guests will be 
running in command line mode, the host should have a GUI.

What choice metapackages provide that function?

Re: Choice of VMs under i386 Stretch?

[Curt]

On 2019-07-01, Matthew Crews  wrote:
> On 7/1/19 10:35 AM, Curt wrote:
>> On 2019-07-01, Matthew Crews  wrote:
>>>
>>> At a cursory glance, it does NOT appear that DVD-1 contains any VM Host
>>> software, other than perhaps nspawn (which is part of Systemd).
>> 
>> Isn't nspawn a chroot container?
>> 
>
> Indeed, but depending on the needs, it might work.
>

Well, now, Mr. Owlett's a real stickler for disallowing any deviation
whatsoever in the answers proposed from the precise terms of the
question as formulated by Mr. Owlett himself, so I fear you are now
venturing upon very, very dangerous ground.

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 7/1/19 10:35 AM, Curt wrote:
> On 2019-07-01, Matthew Crews  wrote:
>>
>> At a cursory glance, it does NOT appear that DVD-1 contains any VM Host
>> software, other than perhaps nspawn (which is part of Systemd).
> 
> Isn't nspawn a chroot container?
> 

Indeed, but depending on the needs, it might work.

Re: Choice of VMs under i386 Stretch?

[Linux-Fan]

Richard Owlett writes:

[...]


On 06/30/2019 10:44 AM, Linux-Fan wrote:

Matthew Crews writes:


On 6/30/19 5:34 AM, Richard Owlett wrote:
> I'm considering using a VM for some experiments.
> Although my web searches have turned up articles about particular
> aspects of particular VMs, I've found no inventory of what VMs are
> available in Debian.
>
> My firm requirement is that all required software be in set of install
> DVDs for i386 Debian 9.8.
>
> What are my options?

Are you asking what virtual machine hosts are available on Debian?


*explicitly*


Good :)

[...]


Did I *NOT* say:

...all required software be in set of install DVDs for i386 Debian 9.8.


I know you wrote that and I attempted to respond to that by the following by
giving a shortcut to a very likely working solution ( :) ):


> If the DVDs contain all of the

`main`
repository, then these packages can be used to start with a
Libvirt/Virt-Manager/KVM setup (this is what I install on my systems):

virt-manager qemu qemu-kvm libvirt-clients libvirt0 libvirt-daemon 
libvirt-daemon-system

The frontend `virt-manager` can then be run to graphically manage VMs.
`virsh` can be used to manage them at the commandline.


I do not understand.


It was an attempt to give you a list of packages that may allow you to start
using VMs without further checks because they are likely to be on the DVDs
(I am still confident that they are and I still have not and will not check
it). If you are interested in finding out that/if all packages are on the
DVDs, use the search engine:

https://cdimage-search.debian.org/

By this I found out that `virt-manager` is on DVD2. So part of my assumption
is right... it's probably easiest if you just try to install the packages?

If it is something else that is not understood in the paragraph I can try to
write it differently if wanted?


Btw. when reading what "VMs are available in Debian" I first thought the
question was about prebuilt images for virtual machines? ...


My *explicit* question is:
What 32 bit VMs which DO NOT depend on non-FOSS components are available for
32-bit Debian?


Not so explicit for me as non-native English speaker due to ambigous use of
word `VM`. `VM` is virtual machine but there are multiple things to
associate with it in this context:

* A `VM` as I commonly understand it on my local system is a hard disk
  image and some metadata allowing it to be run by means of a hypervisor.
  Prebuilt images can be downloaded separately and are not in Debian DVDs.
  This is not an issue for you since you have the DVDs and can ``create''
  your own VMs by using the installer inside an ``empty/new'' VM which most
  management softwares can create easily.

* As an alternative notion, a `VM` is just another server that happens to
  run on a hypervisor and I as the user do not care about what image it
  comes from...

* From the line with `*explicitly*` above I gather that you are asking
  for a software to manage VMs. It is rarely called a `VM` here although
  the linux kernel's hypervisor is called `KVM` for Kernel Virtual Machine.
  All the necessary software is probably available on Debian DVDs.
  If unsure, consult above mentioned URL and list of packages...

* (Other uses of the word may exist...)

HTH
Linux-Fan


pgpGUXOxBXuUl.pgp
Description: PGP signature

Re: Choice of VMs under i386 Stretch?

[Curt]

On 2019-07-01, Matthew Crews  wrote:
>
> At a cursory glance, it does NOT appear that DVD-1 contains any VM Host
> software, other than perhaps nspawn (which is part of Systemd).

Isn't nspawn a chroot container?

> Hope this helps.
>

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 6/30/19 12:12 PM, Richard Owlett wrote:
> On 06/30/2019 10:44 AM, Linux-Fan wrote:
>> Matthew Crews writes:
>>> Are you asking what virtual machine hosts are available on Debian?
> 
> *explicitly*

So clearly I struck a nerve, and I apologize. The way you asked your
question was somewhat unclear to me at first. It is clear now.

I do not know the entire list of software that is found on the different
DVDs, but you can find out by looking on the official DVD lists. Each
DVD has a corresponding .list.gz you can browse, all the way up to DVD 14.

https://cdimage.debian.org/cdimage/release/current/i386/list-dvd/

At a cursory glance, it does NOT appear that DVD-1 contains any VM Host
software, other than perhaps nspawn (which is part of Systemd).

Hope this helps.

-Matt



signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Richard Hector]

On 1/07/19 7:12 AM, Richard Owlett wrote:
> What 32 bit VMs which DO NOT depend on non-FOSS components are available
> for 32-bit Debian?
> 


LEARN TO WRITE!!

A VM is a Virtual Machine. You appear not to be asking for a Virtual
Machine, but software to host a Virtual Machine!!!1!

If you ask the wrong question, you're not going to be happy with the answer!


Please, Richard, remember people are trying to help, but
miscommunications are bound to happen. Nobody deliberately does that
(well, maybe a few do), so please assume that a misinterpretation is
probably a mistake, and _politely_ re-word your question to make it less
ambiguous. You might have to think for a bit to work out how it was
misinterpreted first.

I, like some others, use QEMU/KVM, managed by libvirt. virsh is the
command-line tool to manage VMs. virt-manager is the graphical tool,
which you will probably want if you're using graphical interfaces (X,
Wayland, even Windows) in the VM.

If you're going to use that, you probably want your CPU to support
hardware virtualisation. See here:
https://www.linux-kvm.org/page/Processor_support

Unfortunately, I don't think you'll find that support on 32-bit hardware.

If you're stuck on 32-bit hardware, you might like to try LXC - it's not
full virtualisation, but more like an enhanced chroot. I use that too
(but only on 64-bit hardware at the moment).

There are others, but I haven't used them much.

In order to easily provision disk space for either type, I would
recommend learning to use LVM, if you haven't already.

The VMs themselves, I create.

Richard (a different one)



signature.asc
Description: OpenPGP digital signature

Re: Choice of VMs under i386 Stretch?

[Dan Ritter]

Richard Owlett wrote: 
> *CAVEAT LECTOR*
> If I can't win,
>may I aspire to "break even"?
> Recently I was chastised for expecting people to *read*.
> if no hmr, *DENT* rd frthr ;/
> caveat finis ;/
> 
> > > > What are my options?
> > > >
> > > Are you asking what virtual machine hosts are available on Debian?
> 
> *explicitly*
> 
> > > Because Debian can easily run as a virtual machine guest.
> 
> If not host
>   why ask?  

While debian-user is conducted in English, it is not the first
language of all participants. When a question is asked in an
ambiguous fashion, it's good to ask clarifying questions to make
sure that you understand what is being asked.

And there are some participants who appear to natively read and
write a language which is nearly English, but deliberately or
unconsciously constructed to be as ambiguous and frustrating as 
possible.

> My *explicit* question is:
> What 32 bit VMs which DO NOT depend on non-FOSS components are available for
> 32-bit Debian?

An uncountable but finite number, bounded in a nutshell called the Internet.

As official Debian images: the only official Debian VM images I
am aware of are produced under the Debian Live moniker. However,
the tools to create VM images are available, and many, many
people have taken advantage of them.

-dsr-

Re: Choice of VMs under i386 Stretch?

[Richard Owlett]

*CAVEAT LECTOR*
If I can't win,
   may I aspire to "break even"?
Recently I was chastised for expecting people to *read*.
if no hmr, *DENT* rd frthr ;/
caveat finis ;/

On 06/30/2019 10:44 AM, Linux-Fan wrote:

Matthew Crews writes:


On 6/30/19 5:34 AM, Richard Owlett wrote:
> I'm considering using a VM for some experiments.
> Although my web searches have turned up articles about particular
> aspects of particular VMs, I've found no inventory of what VMs are
> available in Debian.
>
> My firm requirement is that all required software be in set of install
> DVDs for i386 Debian 9.8.
>
> What are my options?
>
>
Are you asking what virtual machine hosts are available on Debian?


*explicitly*


Because Debian can easily run as a virtual machine guest.


If not host
  why ask?  



Debian has several available virtual machine hosts at its disposal.
Check out the wiki:

https://wiki.debian.org/SystemVirtualization#Using_Debian_to_host_Virtual_Computers 



https://wiki.debian.org/SystemVirtualization/ prompted question



That page gives a good overview,


Though "overviews" explicitly identified as irrelevant


although it mentions quite a few solutions
which are possibly not in the DVD set.


Did I *NOT* say:
...all required software be in set of install DVDs for i386 Debian 9.8. 


> If the DVDs contain all of the

`main`
repository, then these packages can be used to start with a
Libvirt/Virt-Manager/KVM setup (this is what I install on my systems):

virt-manager qemu qemu-kvm libvirt-clients libvirt0 libvirt-daemon 
libvirt-daemon-system


The frontend `virt-manager` can then be run to graphically manage VMs.
`virsh` can be used to manage them at the commandline.


I do not understand.



Btw. when reading what "VMs are available in Debian" I first thought the
question was about prebuilt images for virtual machines? ...


My *explicit* question is:
What 32 bit VMs which DO NOT depend on non-FOSS components are available 
for 32-bit Debian?

Re: Choice of VMs under i386 Stretch?

[Linux-Fan]

Matthew Crews writes:


On 6/30/19 5:34 AM, Richard Owlett wrote:
> I'm considering using a VM for some experiments.
> Although my web searches have turned up articles about particular
> aspects of particular VMs, I've found no inventory of what VMs are
> available in Debian.
>
> My firm requirement is that all required software be in set of install
> DVDs for i386 Debian 9.8.
>
> What are my options?
>
>
Are you asking what virtual machine hosts are available on Debian?
Because Debian can easily run as a virtual machine guest.

Debian has several available virtual machine hosts at its disposal.
Check out the wiki:

https://wiki.debian.org/SystemVirtualization#Using_Debian_to_host_Virtual_Computers


That page gives a good overview, although it mentions quite a few solutions
which are possibly not in the DVD set. If the DVDs contain all of the `main`
repository, then these packages can be used to start with a
Libvirt/Virt-Manager/KVM setup (this is what I install on my systems):

virt-manager qemu qemu-kvm libvirt-clients libvirt0 libvirt-daemon 
libvirt-daemon-system

The frontend `virt-manager` can then be run to graphically manage VMs.
`virsh` can be used to manage them at the commandline.

Btw. when reading what "VMs are available in Debian" I first thought the
question was about prebuilt images for virtual machines? Images can be used
to bypass the installation at the expense of needing more download
bandwidth and being less configurable -- such images are not part of the
Debian main repository as far as I know. It is not an issue for me since I
either clone existing VMs or install them anew with the installer most of
the times...

HTH
Linux-Fan


pgpKx027eD78W.pgp
Description: PGP signature

Re: Choice of VMs under i386 Stretch?

[john doe]

On 6/30/2019 3:45 PM, Matthew Crews wrote:
> On 6/30/19 5:34 AM, Richard Owlett wrote:
>> I'm considering using a VM for some experiments.
>> Although my web searches have turned up articles about particular
>> aspects of particular VMs, I've found no inventory of what VMs are
>> available in Debian.
>>
>> My firm requirement is that all required software be in set of install
>> DVDs for i386 Debian 9.8.
>>
>> What are my options?
>>
>>
> Are you asking what virtual machine hosts are available on Debian?
> Because Debian can easily run as a virtual machine guest.
>
> Debian has several available virtual machine hosts at its disposal.
> Check out the wiki:
>
> https://wiki.debian.org/SystemVirtualization#Using_Debian_to_host_Virtual_Computers
>
> Cheers,
>

Depending on what you want to do you could also consider  using chroot:

https://wiki.debian.org/chroot

--
John Doe

Re: Choice of VMs under i386 Stretch?

[Matthew Crews]

On 6/30/19 5:34 AM, Richard Owlett wrote:
> I'm considering using a VM for some experiments.
> Although my web searches have turned up articles about particular
> aspects of particular VMs, I've found no inventory of what VMs are
> available in Debian.
> 
> My firm requirement is that all required software be in set of install
> DVDs for i386 Debian 9.8.
> 
> What are my options?
> 
> 
Are you asking what virtual machine hosts are available on Debian?
Because Debian can easily run as a virtual machine guest.

Debian has several available virtual machine hosts at its disposal.
Check out the wiki:

https://wiki.debian.org/SystemVirtualization#Using_Debian_to_host_Virtual_Computers

Cheers,

-Matt



signature.asc
Description: OpenPGP digital signature

Choice of VMs under i386 Stretch?

[Richard Owlett]

I'm considering using a VM for some experiments.
Although my web searches have turned up articles about particular 
aspects of particular VMs, I've found no inventory of what VMs are 
available in Debian.


My firm requirement is that all required software be in set of install 
DVDs for i386 Debian 9.8.


What are my options?

Re: New laptop: need advice on choice of file system types

[Jonathan Dowland]

On Tue, Apr 16, 2019 at 08:39:25PM +0200, Pascal Hambourg wrote:
If you intend to use guided partitioning on the whole disk, I repeat 
that LVM is worthless unless you plan to add disks in the future.


I'd agree that It's utility is very much diminished by d-i allocating the
entire VG with its guided schemes.

Re: New laptop: need advice on choice of file system types

[Jonas Smedegaard]

Quoting Pascal Hambourg (2019-04-16 20:39:25)
> Le 15/04/2019 à 16:38, Tom Browder a écrit :
> > 
> > I have decided to use the Deb installer and select LVM during
> > the clean installation, and accept the FS default (I assume it will be
> > ext4, but if not, I will select it).
> 
> If you intend to use guided partitioning on the whole disk, I repeat 
> that LVM is worthless unless you plan to add disks in the future.

It is true that debian-installer currently by default uses up all space, 
but LVM allows you to adjust space allocation later - if only you chose 
a partitioning layout with multiple partitions: Log in as root (without 
being logged in as regular user, so that /home is not actively used), 
and run something like this:

  lvreduce --resizefs --size 5G /dev/myhost/home

 - Jonas

-- 
 * Jonas Smedegaard - idealist & Internet-arkitekt
 * Tlf.: +45 40843136  Website: http://dr.jones.dk/

 [x] quote me freely  [ ] ask before reusing  [ ] keep private


signature.asc
Description: signature

Re: New laptop: need advice on choice of file system types

[Felix Miata]

Jonathan Dowland composed on 2019-04-16 09:17 (UTC+0100):

> On Mon, Apr 15, 2019 at 01:38:12PM -0400, Felix Miata wrote:

>> Both DFSee and IBM
>>BM use the last sector on the first track for data storage, including useful
>>cataloging data. Even when not having IBM BM installed, its data sector is
>>(optionally) used by DFSee, by me, always.

> So I gather that you believe DFSee (or some functionality of DFSee that you
> rely upon) is incompatible with LVM.

>From it's logging I have filesystem/operating system inventory that facilitates
clone, backup, restore, move, etc., all available wherever a disk happens to be
accessible, without requiring mounting anything or support for any particular
filesystem:
http://fm.no-ip.com/Tmp/Dfsee/gb250L07.txt
http://fm.no-ip.com/Tmp/Dfsee/p5bseL05.txt

I can't imagine that Gnu LVM wouldn't escalate complications of the complicated
environment that multiboot is.

> But I didn't understand this particular
> point about the last sector on the first track… is that a region of a physical
> drive prior to the first physical partition (assuming an MBR partitioning
> scheme)?

Typically it's either LBA sector 0x1F or 0x3E.

> This alone would not prevent you using LVM, which one would typically
> layer on top of traditional partitions.

That sector contains information not available elsewhere, originally intended 
for
constructing IBM BM's boot menu and functionality, with OS/2 LVM data later 
added.
It wouldn't prevent using Gnu LVM, but neither could it assist with cataloging 
of
Gnu LVM content. Gnu LVM would be an extra layer only accessible via Linux 
tools,
entirely outside the scope of what works for me now.

All this disregards the issue of LVM's physical discontinuity when sector level
rescue operations are indicated by the gap between failure and most recent 
backup.
This alone would keep me from considering its addition here. I'm well beyond a
point in time when any change in routine could be considered anything but 
hostile.
I strongly resist fixing what ain't broke.
-- 
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: New laptop: need advice on choice of file system types

[Pascal Hambourg]

Le 15/04/2019 à 16:38, Tom Browder a écrit :


I have decided to use the Deb installer and select LVM during
the clean installation, and accept the FS default (I assume it will be
ext4, but if not, I will select it).


If you intend to use guided partitioning on the whole disk, I repeat 
that LVM is worthless unless you plan to add disks in the future.


If you intend to use manual partitioning, there is no default filesystem 
type.

Re: New laptop: need advice on choice of file system types

[Jonathan Dowland]

On Mon, Apr 15, 2019 at 01:38:12PM -0400, Felix Miata wrote:

Both DFSee and IBM
BM use the last sector on the first track for data storage, including useful
cataloging data. Even when not having IBM BM installed, its data sector is
(optionally) used by DFSee, by me, always.


So I gather that you believe DFSee (or some functionality of DFSee that you
rely upon) is incompatible with LVM. But I didn't understand this particular
point about the last sector on the first track… is that a region of a physical
drive prior to the first physical partition (assuming an MBR partitioning
scheme)? This alone would not prevent you using LVM, which one would typically
layer on top of traditional partitions.


--

⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Jonathan Dowland
⢿⡄⠘⠷⠚⠋⠀ https://jmtd.net
⠈⠳⣄ Please do not CC me, I am subscribed to the list.

Re: New laptop: need advice on choice of file system types

[Felix Miata]

Jonathan Dowland composed on 2019-04-15 10:28 (UTC-0400):

> On Fri, Apr 12, 2019 at 05:36:00PM -0400, Felix Miata wrote:

>>LVM's extra layer(s) would render my backup/restore system that depends in 
>>large
>>part on cloning useless.

> I don't quite understand this, would you care to elaborate? Thanks! 

My system's inception dates back more than two decades, before I ever installed
Linux, before I became conscious of FOSS. I began then using cross-platform 
DFSee
exclusively as my partitioner and IBM Boot Manager manager when its only 
binaries
were DOS, OS/2 & Windows. Later were added Linux, then MacOS. Both DFSee and IBM
BM use the last sector on the first track for data storage, including useful
cataloging data. Even when not having IBM BM installed, its data sector is
(optionally) used by DFSee, by me, always. As DFSee evolved, so did my system, 
and
PC collection. Portions of DFSee's automatic logging double as my inventory 
system
for dozens of multi- multiboot PCs, some with upwards of 30 operating systems.
DFSee is a lot of things, but GNU LVM manager from an OS/2 or floppy boot isn't
one of them.
-- 
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: New laptop: need advice on choice of file system types

[Tom Browder]

On Fri, Apr 12, 2019 at 9:50 AM Tom Browder  wrote:
>
> I have used ext4 for many years while I have been watching zfs and
...

Thanks to all who have given me advice on selecting the file system
for a new laptop. After considering all suggestions and my use of the
laptop, I have decided to use the Deb installer and select LVM during
the clean installation, and accept the FS default (I assume it will be
ext4, but if not, I will select it).

Warm regards, and many thanks, to all.

-Tom

Re: New laptop: need advice on choice of file system types

[Jonathan Dowland]

On Fri, Apr 12, 2019 at 09:50:23AM -0500, Tom Browder wrote:

I have used ext4 for many years while I have been watching zfs and
btrfs being developed. I am now considering using one or both on at
least one partion during my upcoming new Debian installation.


What in particular do you find attractive about them?


Can anyone recommend either one for a normal (non-developer,
non-hobbyiest) user who does backups and values his data and wants
reasonable reliability?


Ext4 on top of LVM.

If you want to play around with new tech then I'd give ZFS on Linux
a try, but if you just want your machine to work stick to what you
know. I'd avoid BTRFS entirely unless you are an aspiring BTRFS
developer.

--

⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Jonathan Dowland
⢿⡄⠘⠷⠚⠋⠀ https://jmtd.net
⠈⠳⣄ Please do not CC me, I am subscribed to the list.

Re: New laptop: need advice on choice of file system types

[Jonathan Dowland]

On Fri, Apr 12, 2019 at 05:36:00PM -0400, Felix Miata wrote:

LVM's extra layer(s) would render my backup/restore system that depends in large
part on cloning useless.


I don't quite understand this, would you care to elaborate? Thanks!

--

⢀⣴⠾⠻⢶⣦⠀
⣾⠁⢠⠒⠀⣿⡁ Jonathan Dowland
⢿⡄⠘⠷⠚⠋⠀ https://jmtd.net
⠈⠳⣄ Please do not CC me, I am subscribed to the list.

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[Peter Wiersig]

Dan Ritter  writes:

> Peter Wiersig wrote: 
>
> ZFS is now in two incompatible versions: Oracle's, and ZFSonLinux,
> which is now effectively the parent for all the other efforts including
> FreeBSD's ZFS.

The biggest problem is the incompatible license which makes the code
untouchable.

How Oracle is able to distribute a Linux kernel with ZFS support without
either releasing the whole thing under a GPL compatible license or
violating the kernel GPL is a miracle I didn't found the time or need to
investigate.  I consider ZFS a poisoned gift from Oracle to the Linux
community and will not be surprised when they go for one vendor, on
their whim.  If they wanted to, they could release their code under a
compatible license.

My knowledge on the whole affair is bit vague as I had no real interest
past reading LWN or cks articles on ZFS, and I noted the problems around
the CDDL once Illumos was released.

I understand from cks that the real development nowadays happens on
ZFSonLinux as the other OSes were stopped in development, but I would
not touch that patch on the future prospect on a lawsuit analog to the
SCO one.

from https://github.com/zfsonlinux/zfs/wiki/FAQ#licensing
"""Licensing

ZFS is licensed under the Common Development and Distribution License
(CDDL), and the Linux kernel is licensed under the GNU General Public
License Version 2 (GPLv2). While both are free open source licenses they
are restrictive licenses. The combination of them causes problems
because it prevents using pieces of code exclusively available under one
license with pieces of code exclusively available under the other in the
same binary. In the case of the kernel, this prevents us from
distributing ZFS on Linux as part of the kernel binary. However, there
is nothing in either license that prevents distributing it in the form
of a binary module or in the form of source code."""

Peter

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[Dan Ritter]

Peter Wiersig wrote: 
> Matthew Crews  writes:
> >
> > Here is a good talk on the subject by Michael Lucas, one of the premier
> > experts on ZFS. Its worth noting that a lot of the concepts apply to
> > BTRFS to varying degrees:
> >
> > https://www.youtube.com/watch?v=x9A0dX2WqW8
> 
> I don't have time yet, I think I will watch the whole thing later the
> week.
> 
> But I hate the ZFS pushing going around, I think it's a almost fine idea
> from the technical standpoint, it will work if you decide to use it, but
> 
> I surely don't trust the commercial parties behind ZFS as
> - they have acted malicious in the past,

There are no commercial parties behind ZFS. The code came from
Sun, who open-sourced it. Oracle bought Sun, and closed further
development. Open Solaris was forked as Illumos, Open Indiana,
and others, and continued ZFS development. FreeBSD adopted it
fairly quickly. All these efforts came together in OpenZFS.

ZFS is now in two incompatible versions: Oracle's, and ZFSonLinux,
which is now effectively the parent for all the other efforts including
FreeBSD's ZFS.

Oracle? Yeah, you can't trust Oracle any further than you can
spit Larry Ellison. They aren't involved in anything we do with
ZFS anymore.

> I would never put my business on ZFS, I'd always go the "more backups,
> and maybe less perfect GPL solution" route.

You have the choice.

-dsr-

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[Peter Wiersig]

Matthew Crews  writes:
>
> Here is a good talk on the subject by Michael Lucas, one of the premier
> experts on ZFS. Its worth noting that a lot of the concepts apply to
> BTRFS to varying degrees:
>
> https://www.youtube.com/watch?v=x9A0dX2WqW8

I don't have time yet, I think I will watch the whole thing later the
week.

But I hate the ZFS pushing going around, I think it's a almost fine idea
from the technical standpoint, it will work if you decide to use it, but

I surely don't trust the commercial parties behind ZFS as
- they have acted malicious in the past,

- cost us in the linux community a whole lot of energy with
that lawsuit

- were bought by the most twisting corporation which put Microsofts
embrace, extend and extinguish campaigns back on the amateur level.

I would never put my business on ZFS, I'd always go the "more backups,
and maybe less perfect GPL solution" route.

I would really like if the hostile Linux patchset would be off-topic
here and sorry BSD guys, while I know that there were (or are) things
like debian-kfreebsd, but I'm just your average (A/L)GPL-fanboy which
only grudgingly accepts BSD licensed code.  Because FSFs license
protects me better as a developer and as a user, is a non-malicous
cancer from my point of view.  And yes, the FSF GFDL is a mistake.

Peter

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[Anders Andersson]

On Sat, Apr 13, 2019 at 7:08 PM Felix Miata  wrote:
>
> Anders Andersson composed on 2019-04-13 17:31 (UTC+0200):
>
> > Felix Miata wrote:
>
> >> Because of its snapshotting, BTRFS requires considerably more space than 
> >> older
> >> filesystems, as much as double.
>
> > A btrfs snapshot takes approximately zero space. Where did you get
> > this idea from?
>
> (not an exhaustive list)
>
> 1: "Disk Space Full Because of Snapper" on https://en.opensuse.org/SDB:BTRFS
>
> 2: Since 2015, BTRFS has been the default / filesystem on openSUSE, which
> recommends minimum / filesystem size of 20GB for EXT4, compared to 40GB for 
> BTRFS.
>
> 3: Much more common / filesystem freespace exhausted threads on mailing lists 
> and
> web forums from BTRFS users compared to EXT4 users, with the usual 
> recommendation
> to delete one or more snapshots to free space.
>
> 4: 
> https://btrfs.wiki.kernel.org/index.php/FAQ#Help.21_I_ran_out_of_disk_space.21
>
> 5: 
> https://btrfs.wiki.kernel.org/index.php/FAQ#Why_is_free_space_so_complicated.3F
>
> 6:
> https://www.linuxquestions.org/questions/linux-containers-122/docker-on-btrfs-using-much-space-in-var-lib-docker-btrfs-4175622037/#post5811463



Of course you will run out of space if you keep taking snapshots!
Btrfs never does this, but you can do it *manually* or with
third-party tools. When you take a snapshot, btrfs will keep
everything until it's deleted. That's why everyone says that if you
run out of space, you can delete snapshots. If the advise to the user
is to delete a snapshot, it is something that the user did *because*
they wanted to retain those files.

The "Snapper" user has *installed* a tool that takes snapshots all the
time - obviously you will run out of space because every file you
modify or delete will still be stored in its original version until
you remove the snapshot.

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[Matthew Crews]

On 4/13/19 5:40 PM, Peter Wiersig wrote:
> Peter Wiersig  writes:
>>
>> I would be pissed if my OS removes snapshots I might or might not need
>> in the future.  That's a release critical bug in my eyes.  Yeah, I know
>> Microsoft and Apple do that automatically if your capacity runs out, but
>> that's also why I don't recommend them at all.
> 
> Ok, I checked https://en.opensuse.org/Portal:Snapper and they do support
> LVM and ext4, and they have a bullet of auto-removing old snapshots.  I
> hope they did it right, perhaps I need to make a new test drive with the
> latest release.
> 
> Snapshots on ZFS can't be zero cost, so you need to account for them
> there, too.
> 
> Peter
> 

ZFS Snapshots are nearly zero cost to create the snapshot, since ZFS
(and likewise BTRFS) are copy-on-write file systems. What it does is
records the deltas after the snapshots.

This is a good thinkg as it saves on disk space.

For example:

You create a random 10 MB file, and take a snapshot. You then alter the
5 tail MBs and add five more MBs at the tail. You are left with:

Pre-Snapshot | 5MB chunk #1 | 5MB chunk #2 |  - total 10MBs
Post-Snapshot | 5MB chunk #1 | 5MB chunk #3 | 5MB chunk #4 | - total 15MBs

Actual disk usage will be 20 MBs, since the 5MB chunk #1 is only
recorded on disk once, not twice.

Here is a good talk on the subject by Michael Lucas, one of the premier
experts on ZFS. Its worth noting that a lot of the concepts apply to
BTRFS to varying degrees:

https://www.youtube.com/watch?v=x9A0dX2WqW8

-Matt

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[Peter Wiersig]

Peter Wiersig  writes:
>
> I would be pissed if my OS removes snapshots I might or might not need
> in the future.  That's a release critical bug in my eyes.  Yeah, I know
> Microsoft and Apple do that automatically if your capacity runs out, but
> that's also why I don't recommend them at all.

Ok, I checked https://en.opensuse.org/Portal:Snapper and they do support
LVM and ext4, and they have a bullet of auto-removing old snapshots.  I
hope they did it right, perhaps I need to make a new test drive with the
latest release.

Snapshots on ZFS can't be zero cost, so you need to account for them
there, too.

Peter

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[Peter Wiersig]

Felix Miata  writes:

> Anders Andersson composed on 2019-04-13 17:31 (UTC+0200):
>
>> Felix Miata wrote:
>
>>> Because of its snapshotting, BTRFS requires considerably more space than 
>>> older
>>> filesystems, as much as double.
>
>> A btrfs snapshot takes approximately zero space. Where did you get
>> this idea from? 
>
> 1: "Disk Space Full Because of Snapper" on https://en.opensuse.org/SDB:BTRFS

i only skimmed that, good resource for anyone even not on SUSE.

> 2: Since 2015, BTRFS has been the default / filesystem on openSUSE, which
> recommends minimum / filesystem size of 20GB for EXT4, compared to 40GB for 
> BTRFS.

>From my experience I think they only use snapshotting on BTRFS volumes,
I don't know if they support it with LVM, but is it even possible with
ext4 only?

And yeah, no surprise, if you want snapshots, you'll need more capacity,
it would probably the same recommendation with LVM

And the effects of using snapshots is assuring, you simply set the
system to auto-update everything, and if your system doesn't boot, you
simply select the older snapshot from the grub menu and have your system
running in no time. (This was my scenario with a SUSE desktop, I tried
tumbleweed aka SUSEs unstable/experimental distribution, where some
instability was expected)

I would never recommend something like that for server updates/upgrades,
as there a other far more procedures you can follow to test verify and
prevent service loss on failed updates, if you simply cluster your
services.

I would be pissed if my OS removes snapshots I might or might not need
in the future.  That's a release critical bug in my eyes.  Yeah, I know
Microsoft and Apple do that automatically if your capacity runs out, but
that's also why I don't recommend them at all.

Have a monitoring on all your systems, track each and every possible
value in compact rrd databases, calculate trends from those values and
you'll never be surprised by filled up disks, growing defects detected
by SMART etc.

I'd like a debian desktop/notebook/tablet where snapshots were
implemented in a openSUSE manner, so that I can simply forget about
updates, have them installed in background und can go to the snapshot if
problems arise.

Peter

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[deloptes]

Felix Miata wrote:

> Anders Andersson composed on 2019-04-13 17:31 (UTC+0200):
> 
>> Felix Miata wrote:
> 
>>> Because of its snapshotting, BTRFS requires considerably more space than
>>> older filesystems, as much as double.
> 
>> A btrfs snapshot takes approximately zero space. Where did you get
>> this idea from?
> 
> (not an exhaustive list)
> 
> 1: "Disk Space Full Because of Snapper" on
> https://en.opensuse.org/SDB:BTRFS
> 
> 2: Since 2015, BTRFS has been the default / filesystem on openSUSE, which
> recommends minimum / filesystem size of 20GB for EXT4, compared to 40GB
> for BTRFS.
> 
> 3: Much more common / filesystem freespace exhausted threads on mailing
> lists and web forums from BTRFS users compared to EXT4 users, with the
> usual recommendation to delete one or more snapshots to free space.
> 
> 4:
>
https://btrfs.wiki.kernel.org/index.php/FAQ#Help.21_I_ran_out_of_disk_space.21
> 
> 5:
>
https://btrfs.wiki.kernel.org/index.php/FAQ#Why_is_free_space_so_complicated.3F
> 
> 6:
>
https://www.linuxquestions.org/questions/linux-containers-122/docker-on-btrfs-using-much-space-in-var-lib-docker-btrfs-4175622037/#post5811463

This time I like you :) and agree with you fully.

ext4 or xfs - these are most convenient, although also zfs might be an
option if one can manage it

regards

Re: BTRFS snapshot space consumption (was: New laptop: need advice on choice...)

[Felix Miata]

Anders Andersson composed on 2019-04-13 17:31 (UTC+0200):

> Felix Miata wrote:

>> Because of its snapshotting, BTRFS requires considerably more space than 
>> older
>> filesystems, as much as double.

> A btrfs snapshot takes approximately zero space. Where did you get
> this idea from? 

(not an exhaustive list)

1: "Disk Space Full Because of Snapper" on https://en.opensuse.org/SDB:BTRFS

2: Since 2015, BTRFS has been the default / filesystem on openSUSE, which
recommends minimum / filesystem size of 20GB for EXT4, compared to 40GB for 
BTRFS.

3: Much more common / filesystem freespace exhausted threads on mailing lists 
and
web forums from BTRFS users compared to EXT4 users, with the usual 
recommendation
to delete one or more snapshots to free space.

4: 
https://btrfs.wiki.kernel.org/index.php/FAQ#Help.21_I_ran_out_of_disk_space.21

5: 
https://btrfs.wiki.kernel.org/index.php/FAQ#Why_is_free_space_so_complicated.3F

6:
https://www.linuxquestions.org/questions/linux-containers-122/docker-on-btrfs-using-much-space-in-var-lib-docker-btrfs-4175622037/#post5811463
-- 
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: New laptop: need advice on choice of file system types

[Anders Andersson]

On Fri, Apr 12, 2019 at 4:51 PM Tom Browder  wrote:
>
> I have used ext4 for many years while I have been watching zfs and
> btrfs being developed. I am now considering using one or both on at
> least one partion during my upcoming new Debian installation.
>
> Can anyone recommend either one for a normal (non-developer,
> non-hobbyiest) user who does backups and values his data and wants
> reasonable reliability?

I've used btrfs on every machine for a many years, and I'm very happy
with it. Run it on my server with various disks, desktop, and laptop.
It has saved my data from silent corruption due to bad hardware where
ext4 would just return bad data, and it's a breeze to take very quick
backups thanks to the snapshot and btrfs-send functions. You just have
to learn the new tools.

I've replaced disks more or less live on my sever and it works well.

I'm not going to play with hacking in a non-native filesystem. ZFS is
great if you run FreeBSD, but I run debian main and prefer not to
support contrib and non-free.

Re: New laptop: need advice on choice of file system types

[Anders Andersson]

On Fri, Apr 12, 2019 at 11:36 PM Felix Miata  wrote:
>
> Tom Browder composed on 2019-04-12 09:50 (UTC-0500):
>
> > I have used ext4 for many years while I have been watching zfs and
> > btrfs being developed. I am now considering using one or both on at
> > least one partion during my upcoming new Debian installation.
>
> Because of its snapshotting, BTRFS requires considerably more space than older
> filesystems, as much as double.

A btrfs snapshot takes approximately zero space. Where did you get
this idea from?

Re: New laptop: need advice on choice of file system types

[Dan Ritter]

Curt wrote: 
> On 2019-04-12, Thomas D Dial  wrote:
> >
> > ZFS for /home makes sense, especially for anyone not already somewhat
> > familiar with ZFS.
> 
> Well, if ZFS is this big sixteen-wheeler that you might crash into the
> concrete embankment if you're not careful, what are the benefits that
> outweigh or override these risks for the casual Linux enthusiast, when
> she can just drive an easy default ext4 automatic vehicle and know
> she'll get from here to there safely without worrying about it?

ZFS is not for the casual, right now. For example, my father has
been using Debian for more than 15 years now, but he's never
typed a sudo command without explicit instruction from me. He
just wants to use his computer. If I put him on a ZFS system, I
would continue to manage it for him the way I manage his ext4
backups right now.

The data management concerns that prompted the creation of ZFS are
increasingly within the scope of normal hardware purchases.  People are
buying drives that perform like sports cars and have the capacities of
trucks. Unlike a commercial truck or a fancy sports car, ZFS is free. So
it can make sense to more people to spend the time necessary to learn
how to drive it properly.

ZFS unifies disk management, storage allocation, RAID, caching, data
safety, snapshotting, compression and deduplication. The simpler your
scenario, the less attractive ZFS is.

For a single-disk workstation, ZFS offers flexible filesystem allocations
(/home and /var can share the disk without either one being limited by
a partition size), on-the-fly compression, checksumming of all data,
and a live scrub (fsck) method that fixes any issues without taking your
system offline. There's also the possibility of snapshotting the system
for recovery or backup over the network. Snapshots are extremely fast
and can be done automatically via cron.  Zsend/zrecv are snapshot-aware.

For a 2-4 disk desktop, ZFS does all that plus RAID1, RAID10, or improved
RAID5 variants (RAIDZ1 through Z3).

For a multi-disk server, add in a separate intent log (write cache), read
cache, and possible deduplication. (Dedupe is a giant RAM hog,
however, and very difficult to turn off if you haven't prepared
properly.)

In the near future, ZFS adds native encryption.

-dsr-

Re: New laptop: need advice on choice of file system types

[Curt]

On 2019-04-12, Thomas D Dial  wrote:
>
> ZFS for /home makes sense, especially for anyone not already somewhat
> familiar with ZFS.

Well, if ZFS is this big sixteen-wheeler that you might crash into the
concrete embankment if you're not careful, what are the benefits that
outweigh or override these risks for the casual Linux enthusiast, when
she can just drive an easy default ext4 automatic vehicle and know
she'll get from here to there safely without worrying about it?

> Regards,
> Tom Dial
>
>> 
>> -dsr-
>

Re: New laptop: need advice on choice of file system types

[Default User]

On Fri, Apr 12, 2019, 18:07 Dan Ritter  wrote:

> Default User wrote:
> > On Fri, Apr 12, 2019, 12:43 Dan Ritter  wrote:
>
> > And what about Btrfs?
>
> I don't currently recommend it in any situation where ZFS is an
> option. That comes from 2 years of working with btrfs where
> doing normal maintenance ended up destroying data more than
> once. It may be better now; I haven't looked back.
>
> -dsr-
>



Dan, thanks for the feedback.