Re: Debian 8 and Debian 9 Dual Boot

[Pascal Hambourg]

Le 29/12/2017 à 23:46, Dan Norton a écrit :

On 12/29/2017 08:52 AM, Pascal Hambourg wrote:


The details for other detected OSes are provided by os-prober. The 
entry title for the main OS is derived from the GRUB_DISTRIBUTOR 
variable in /etc/default/grub. You can tweak it to fit your needs. If 
lsb-release is installed, changing -i to -d will provide more details 
such as the Debian version and codename.


Very helpful. Actually the name is "lsb_release".


lsb-release is the name of the Debian package providing the lsb_release 
command.


I need to study the 
shell - not sure about that GRUB_DISTRIBUTOR statement. Will definitely 
change the -i to -d. Why isn't -ds needed?


What do you mean ? -s is already present in the line.
Note that you can define your own static text instead of the shell command.

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 12/29/2017 08:52 AM, Pascal Hambourg wrote:

Le 21/12/2017 à 20:07, Dan Norton a écrit :



Disklabel type: gpt
Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24
Device Start    End    Sectors   Size Type
/dev/sda1   2048 411647 409600   200M BIOS boot

(...)

Is there a problem here?


Yes. /dev/sda1 has the type "BIOS boot" but is actually used as an EFI 
system partition, according to df and /etc/fstab. So it should have 
the type "EFI system".




Ah. Getting this right is a problem for me, but when the installer does 
all the partitioning, the right choices seem to be made.


Before moving on to multiboot with LVM and GPT, I'd like to change 
the menu entries to something more consistent. The last install is 
referred to as "Debian GNU/Linux" but that's ambiguous. Which Debian 
GNU/Linux? If each entry was in the form "Debian GNU/Linux 9 
(stretch) (on /dev/mapper/vol2-root)" that would really be explicit. 
Also I want some more time to mull over which to boot.


The details for other detected OSes are provided by os-prober. The 
entry title for the main OS is derived from the GRUB_DISTRIBUTOR 
variable in /etc/default/grub. You can tweak it to fit your needs. If 
lsb-release is installed, changing -i to -d will provide more details 
such as the Debian version and codename.




Very helpful. Actually the name is "lsb_release". I need to study the 
shell - not sure about that GRUB_DISTRIBUTOR statement. Will definitely 
change the -i to -d. Why isn't -ds needed?

Re: Debian 8 and Debian 9 Dual Boot

[Pascal Hambourg]

Le 21/12/2017 à 20:07, Dan Norton a écrit :



Disklabel type: gpt
Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24
Device Start    End    Sectors   Size Type
/dev/sda1   2048 411647 409600   200M BIOS boot

(...)

Is there a problem here?


Yes. /dev/sda1 has the type "BIOS boot" but is actually used as an EFI 
system partition, according to df and /etc/fstab. So it should have the 
type "EFI system".


Before moving on to multiboot with LVM and GPT, I'd like to change the 
menu entries to something more consistent. The last install is referred 
to as "Debian GNU/Linux" but that's ambiguous. Which Debian GNU/Linux? 
If each entry was in the form "Debian GNU/Linux 9 (stretch) (on 
/dev/mapper/vol2-root)" that would really be explicit. Also I want some 
more time to mull over which to boot.


The details for other detected OSes are provided by os-prober. The entry 
title for the main OS is derived from the GRUB_DISTRIBUTOR variable in 
/etc/default/grub. You can tweak it to fit your needs. If lsb-release is 
installed, changing -i to -d will provide more details such as the 
Debian version and codename.

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 12/28/2017 04:48 AM, Pascal Hambourg wrote:

Le 24/12/2017 à 05:36, Felix Miata a écrit :

Dan Norton composed on 2017-12-23 19:15 (UTC-0500):


The menu inside the box is:
Debian GNU/Linux
Advanced options for Debian GNU/Linux
Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
Advanced options for Debian GNU/Linux 8 (jessie) (on 
/dev/mapper/vol1-root)

Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)
Advanced options for Debian GNU/Linux buster/sid (on 
/dev/mapper/vol3-root)



The first two boot stretch, so they will eventually have "9 (stretch)
(on /dev/mapper/vol2-root)" appended, once the timeout is under 
control.


Based on what I see and what you say, it seems you are modifying the 
timeout for
Stretch (/etc/default/grub on vol2), but actually booting Stretch 
from Jessie's
grub.cfg (/etc/default/grub on vol1), which remains configured to 3 
seconds.


Based on what I see and what Dan wrote, I'd rather say the other way 
around : Dan edited /etc/default/grub in Jessie (update-grub showed 
the system kernel was Jessie's 3.16 and found stretch/9.3 as a foreign 
system) but the GRUB loading at boot time is the one from stretch (the 
first entries boot stretch). So the time-out must be changed from 
stretch.


You can check the result of /etc/default/grub parameters in 
/boot/grub/grub.cfg after running update-grub.
You can also check GRUB variables at boot time in GRUB's shell (press 
"c" to spawn the shell) with the command "set". It will also display 
value of the "prefix" variable which contains the device and path to 
the used /boot/grub directory.




We have a winner! Thanks Pascal.

I checked the GRUB variables for each installation: jessie, stretch, and 
buster. The prefix was identical for all - a hairy, hard to read, 
touch-typing exercise of the form:


lvmid//

Of course, the UUIDs were immediately recognizable ;-) as belonging to 
volume group "vol2" and logical volume "root" where stretch was 
installed. I changed the time-out in /etc/default/grub and ran 
update-grub from stretch and it *changed* .


The old /etc/default/grub was like this (excluding comments):

GRUB_DEFAULT=0
GRUB_TIMEOUT=5
GRUB_DISTRIBUTOR=`lsb_release -i -s 2> /dev/null || echo Debian`
GRUB_CMDLINE_LINUX_DEFAULT="quiet"
GRUB_CMDLINE_LINUX=""

...so the 3 seconds I was seeing was probably due to a run-off of the 5. 
Anyway, I changed 5 to 12 arbitrarily and that was effective.


Thank you, Pascal.

Re: Debian 8 and Debian 9 Dual Boot

[Pascal Hambourg]

Le 24/12/2017 à 05:36, Felix Miata a écrit :

Dan Norton composed on 2017-12-23 19:15 (UTC-0500):


The menu inside the box is:
Debian GNU/Linux
Advanced options for Debian GNU/Linux
Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
Advanced options for Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)
Advanced options for Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)



The first two boot stretch, so they will eventually have "9 (stretch)
(on /dev/mapper/vol2-root)" appended, once the timeout is under control.


Based on what I see and what you say, it seems you are modifying the timeout for
Stretch (/etc/default/grub on vol2), but actually booting Stretch from Jessie's
grub.cfg (/etc/default/grub on vol1), which remains configured to 3 seconds.


Based on what I see and what Dan wrote, I'd rather say the other way 
around : Dan edited /etc/default/grub in Jessie (update-grub showed the 
system kernel was Jessie's 3.16 and found stretch/9.3 as a foreign 
system) but the GRUB loading at boot time is the one from stretch (the 
first entries boot stretch). So the time-out must be changed from stretch.


You can check the result of /etc/default/grub parameters in 
/boot/grub/grub.cfg after running update-grub.
You can also check GRUB variables at boot time in GRUB's shell (press 
"c" to spawn the shell) with the command "set". It will also display 
value of the "prefix" variable which contains the device and path to the 
used /boot/grub directory.

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-12-27 18:59 (UTC-0500):

> Felix Miata wrote:

>> Is there more than one directory in /boot/efi/EFI/? If not, it's likely time 
>> for
>> you to explore using /etc/default/grub's GRUB_DISTRIBUTOR= option. I need to
>> (only one Debian, but 3 openSUSEs installed), but have been putting it off,
>> using Stretch's menu for all.

> There is only onedirectory in /boot/efi/EFI/ :

> root@BR914:/# ls -la /boot/efi/EFI
> total 12
> drwx-- 3 root root 4096 Dec  4 22:11 .
> drwx-- 3 root root 4096 Dec 31  1969 ..
> drwx-- 2 root root 4096 Dec  4 22:11 debian

> How would GRUB_DISTRIBUTOR=help the timeout= problem? 

Indirectly, by having a POST-time (F12 menu) choice which installation's Grub to
use.
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-12-27 18:59 (UTC-0500):

> Felix Miata wrote:

>> Based on what I see and what you say, it seems you are modifying the timeout 
>> for
>> Stretch (/etc/default/grub on vol2), but actually booting Stretch from 
>> Jessie's
>> grub.cfg (/etc/default/grub on vol1), which remains configured to 3 seconds.

> That seems reasonable, but I've tried modifying the timeout after 
> booting each of the installations. The timeout remains immutable. It's 
> as if update-grub is not using what is in grub.d for timeout value. But 
> where does 3s come from?   
 There's nothing to do with timeout in /etc/grub.d/ anyone but packager
owner and/or upstream should be touching. If you can't control timeout
exclusively via /etc/default/grub and update-grub, then it must be time to ask
grub people via
https://lists.gnu.org/mailman/listinfo/help-grub or report a bug.
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 12/23/2017 11:36 PM, Felix Miata wrote:


Dan Norton composed on 2017-12-23 19:15 (UTC-0500):


Felix Miata wrote:
The menu inside the box is:
Debian GNU/Linux
Advanced options for Debian GNU/Linux
Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
Advanced options for Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)
Advanced options for Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)
The first two boot stretch, so they will eventually have "9 (stretch)
(on /dev/mapper/vol2-root)" appended, once the timeout is under control.

Based on what I see and what you say, it seems you are modifying the timeout for
Stretch (/etc/default/grub on vol2), but actually booting Stretch from Jessie's
grub.cfg (/etc/default/grub on vol1), which remains configured to 3 seconds.


That seems reasonable, but I've tried modifying the timeout after 
booting each of the installations. The timeout remains immutable. It's 
as if update-grub is not using what is in grub.d for timeout value. But 
where does 3s come from?




Is there more than one directory in /boot/efi/EFI/? If not, it's likely time for
you to explore using /etc/default/grub's GRUB_DISTRIBUTOR= option. I need to
(only one Debian, but 3 openSUSEs installed), but have been putting it off,
using Stretch's menu for all.


There is only onedirectory in /boot/efi/EFI/ :

root@BR914:/# ls -la /boot/efi/EFI
total 12
drwx-- 3 root root 4096 Dec  4 22:11 .
drwx-- 3 root root 4096 Dec 31  1969 ..
drwx-- 2 root root 4096 Dec  4 22:11 debian

How would GRUB_DISTRIBUTOR=help the timeout= problem? OTOH, to change 
the wording of menu entries, I can edit grub.d files.

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-12-23 19:15 (UTC-0500):

> Felix Miata wrote:

> The menu inside the box is:
> Debian GNU/Linux
> Advanced options for Debian GNU/Linux
> Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
> Advanced options for Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
> Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)
> Advanced options for Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)

> The first two boot stretch, so they will eventually have "9 (stretch) 
> (on /dev/mapper/vol2-root)" appended, once the timeout is under control.  
> 

Based on what I see and what you say, it seems you are modifying the timeout for
Stretch (/etc/default/grub on vol2), but actually booting Stretch from Jessie's
grub.cfg (/etc/default/grub on vol1), which remains configured to 3 seconds.

Is there more than one directory in /boot/efi/EFI/? If not, it's likely time for
you to explore using /etc/default/grub's GRUB_DISTRIBUTOR= option. I need to
(only one Debian, but 3 openSUSEs installed), but have been putting it off,
using Stretch's menu for all.
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 12/23/2017 04:35 PM, Felix Miata wrote:


Dan Norton composed on 2017-12-23 15:12 (UTC-0500):


Felix Miata wrote:

[...]
It's not so easy to figure out when POST is over with UEFI. Here, it seems
efibootmgr -t provides extra delay beyond what the BIOS defines for you to make
a selection from its own boot device selection menu, which requires an F12
keystroke here to see.
The timeout after appearance of Grub's menu is supposed to be controlled by
/etc/default/grub's GRUB_TIMEOUT=, which shows up here in Stretch's grub.cfg
first on line 86, a few lines before "### END /etc/grub.d/00_header ###".

Edited /etc/default/grub to change GRUB_TIMEOUT to 11:
root@BR914:/etc/default# nano grub
Observed "If you change this file, run 'update-grub' afterwards..." :
root@BR914:/etc/default# update-grub
Generating grub configuration file ...
Found background image: /usr/share/images/desktop-base/desktop-grub.png
Found linux image: /boot/vmlinuz-3.16.0-4-amd64
Found initrd image: /boot/initrd.img-3.16.0-4-amd64
Found Debian GNU/Linux (9.3) on /dev/mapper/vol2-root
Found Debian GNU/Linux (buster/sid) on /dev/mapper/vol3-root
done
Looking at /boot/grub/grub.cfg, timeout is mentioned as follows:
if [ "${recordfail}" = 1 ] ; then
    set timeout=-1
else
    if [ x$feature_timeout_style = xy ] ; then
      set timeout_style=menu
      set timeout=11
    # Fallback normal timeout code in case the timeout_style feature is
    # unavailable.
    else
      set timeout=11
    fi
fi
...but there is no effect. The timeout when rebooting is still 3
seconds. I'm no shell expert so I don't know how to interpret the above.

What exactly is on the screen during those 3 seconds?



The menu inside the box is:
Debian GNU/Linux
Advanced options for Debian GNU/Linux
Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
Advanced options for Debian GNU/Linux 8 (jessie) (on /dev/mapper/vol1-root)
Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)
Advanced options for Debian GNU/Linux buster/sid (on /dev/mapper/vol3-root)

The first two boot stretch, so they will eventually have "9 (stretch) 
(on /dev/mapper/vol2-root)" appended, once the timeout is under control.


[...]

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-12-23 15:12 (UTC-0500):

> Felix Miata wrote:

>> [...]
>> It's not so easy to figure out when POST is over with UEFI. Here, it seems
>> efibootmgr -t provides extra delay beyond what the BIOS defines for you to 
>> make
>> a selection from its own boot device selection menu, which requires an F12
>> keystroke here to see.

>> The timeout after appearance of Grub's menu is supposed to be controlled by
>> /etc/default/grub's GRUB_TIMEOUT=, which shows up here in Stretch's grub.cfg
>> first on line 86, a few lines before "### END /etc/grub.d/00_header ###".

> Edited /etc/default/grub to change GRUB_TIMEOUT to 11:
> root@BR914:/etc/default# nano grub

> Observed "If you change this file, run 'update-grub' afterwards..." :
> root@BR914:/etc/default# update-grub
> Generating grub configuration file ...
> Found background image: /usr/share/images/desktop-base/desktop-grub.png
> Found linux image: /boot/vmlinuz-3.16.0-4-amd64
> Found initrd image: /boot/initrd.img-3.16.0-4-amd64
> Found Debian GNU/Linux (9.3) on /dev/mapper/vol2-root
> Found Debian GNU/Linux (buster/sid) on /dev/mapper/vol3-root
> done

> Looking at /boot/grub/grub.cfg, timeout is mentioned as follows:
> if [ "${recordfail}" = 1 ] ; then
>    set timeout=-1
> else
>    if [ x$feature_timeout_style = xy ] ; then
>      set timeout_style=menu
>      set timeout=11
>    # Fallback normal timeout code in case the timeout_style feature is
>    # unavailable.
>    else
>      set timeout=11
>    fi
> fi

> ...but there is no effect. The timeout when rebooting is still 3 
> seconds. I'm no shell expert so I don't know how to interpret the above.

What exactly is on the screen during those 3 seconds?

NAICT, the first "if" is setting the timeout to infinite if there is nothing
found that could be booted. The next "if" is using 11 if some sort of optional
timeout indication feature is enabled. Otherwise, 11 is used as Grub's own
standard (invisible) timeout "indication".

Do you still have only Jessie installed? If so, maybe its grub-efi is broken,
and going ahead and installing Stretch will replace Jessie's with a working one.
Stretch's is working as expected here. Jessie's I've never had occasion to use.

Another thought is that if there is but one valid Grub stanza, the timeout
setting might be ignored.
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 12/21/2017 05:13 PM, Felix Miata wrote:

[...]
It's not so easy to figure out when POST is over with UEFI. Here, it seems
efibootmgr -t provides extra delay beyond what the BIOS defines for you to make
a selection from its own boot device selection menu, which requires an F12
keystroke here to see.

The timeout after appearance of Grub's menu is supposed to be controlled by
/etc/default/grub's GRUB_TIMEOUT=, which shows up here in Stretch's grub.cfg
first on line 86, a few lines before "### END /etc/grub.d/00_header ###".


Edited /etc/default/grub to change GRUB_TIMEOUT to 11:
root@BR914:/etc/default# nano grub

Observed "If you change this file, run 'update-grub' afterwards..." :
root@BR914:/etc/default# update-grub
Generating grub configuration file ...
Found background image: /usr/share/images/desktop-base/desktop-grub.png
Found linux image: /boot/vmlinuz-3.16.0-4-amd64
Found initrd image: /boot/initrd.img-3.16.0-4-amd64
Found Debian GNU/Linux (9.3) on /dev/mapper/vol2-root
Found Debian GNU/Linux (buster/sid) on /dev/mapper/vol3-root
done

Looking at /boot/grub/grub.cfg, timeout is mentioned as follows:
if [ "${recordfail}" = 1 ] ; then
  set timeout=-1
else
  if [ x$feature_timeout_style = xy ] ; then
    set timeout_style=menu
    set timeout=11
  # Fallback normal timeout code in case the timeout_style feature is
  # unavailable.
  else
    set timeout=11
  fi
fi

...but there is no effect. The timeout when rebooting is still 3 
seconds. I'm no shell expert so I don't know how to interpret the above.

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-12-21 16:53 (UTC-0500):

> Felix Miata wrote:

>> Dan Norton composed on 2017-12-21 14:07 (UTC-0500):

>>> There are still mysteries I have not solved. For some reason, GRUB has
>>> decided that after POST, you only need 3 seconds to choose which
>>> installation to boot. GRUB has resisted my efforts to change that
>>> timeout value. I've been able to change the boot order in NVRAM, but not
>>> the timeout.

>> efibootmgr -t ##

> Yes, it seems like that should work, but currently:

> # efibootmgr
> BootCurrent: 
> Timeout: 11 seconds  # ... and the actual timeout is 3
> BootOrder: 0003,0001,,0002,0006,0004,0005
> Boot* debian
> Boot0001* USB Floppy/CD
> Boot0002* USB Hard Drive
> Boot0003* ATAPI CD-ROM Drive
> Boot0004* Unknown Device
> Boot0005* USB Floppy/CD
> Boot0006* Hard Drive

> Doing it again (See definition of insanity [1])
> # efibootmgr -t 12
> BootCurrent: 
> Timeout: 12 seconds
> BootOrder: 0003,0001,,0002,0006,0004,0005
> Boot* debian
> Boot0001* USB Floppy/CD
> Boot0002* USB Hard Drive
> Boot0003* ATAPI CD-ROM Drive
> Boot0004* Unknown Device
> Boot0005* USB Floppy/CD
> Boot0006* Hard Drive

> Well, that changed it from 3 to 4 (!?). Strange.

> [1] "Insanity is doing the same thing over & over again and expecting a 
> different result." - Einstein

It's not so easy to figure out when POST is over with UEFI. Here, it seems
efibootmgr -t provides extra delay beyond what the BIOS defines for you to make
a selection from its own boot device selection menu, which requires an F12
keystroke here to see.

The timeout after appearance of Grub's menu is supposed to be controlled by
/etc/default/grub's GRUB_TIMEOUT=, which shows up here in Stretch's grub.cfg
first on line 86, a few lines before "### END /etc/grub.d/00_header ###".
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Joe]

On Thu, 21 Dec 2017 16:53:19 -0500
Dan Norton  wrote:


> [1] "Insanity is doing the same thing over & over again and expecting
> a different result." - Einstein
> 

Probably the single most stupid thing he ever said, given that he also
said 'God does not play dice', showing that he knew what dice were and
what they were used for.

He also presumably had never had to fix an intermittent fault...

-- 
Joe

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 12/21/2017 02:54 PM, Felix Miata wrote:

Dan Norton composed on 2017-12-21 14:07 (UTC-0500):


2. You will make extra work for yourself by having a common swap
partition for all installations. With the common swap, each new
installation gave rise to these messages:
    a. "gave up waiting for suspend/resume device"
    b. "a start job is running for dev-disk-by\..."
    c. "failed to connect to lvmetad"
STW can reveal ways to avoid these messages, but they are a PITA and
avoidable by each volume group having its own swap.

Except with a first Linux installation, I tell the partitioner not to use swap,
then add it by LABEL to fstab after installing. The problem is that the
installer insists that the swap partition needs to be formatted, destroying the
validity of swap's UUID in the fstabs of the previous installations.


There are still mysteries I have not solved. For some reason, GRUB has
decided that after POST, you only need 3 seconds to choose which
installation to boot. GRUB has resisted my efforts to change that
timeout value. I've been able to change the boot order in NVRAM, but not
the timeout.

efibootmgr -t ##


Yes, it seems like that should work, but currently:

# efibootmgr
BootCurrent: 
Timeout: 11 seconds  # ... and the actual timeout is 3
BootOrder: 0003,0001,,0002,0006,0004,0005
Boot* debian
Boot0001* USB Floppy/CD
Boot0002* USB Hard Drive
Boot0003* ATAPI CD-ROM Drive
Boot0004* Unknown Device
Boot0005* USB Floppy/CD
Boot0006* Hard Drive

Doing it again (See definition of insanity [1])
# efibootmgr -t 12
BootCurrent: 
Timeout: 12 seconds
BootOrder: 0003,0001,,0002,0006,0004,0005
Boot* debian
Boot0001* USB Floppy/CD
Boot0002* USB Hard Drive
Boot0003* ATAPI CD-ROM Drive
Boot0004* Unknown Device
Boot0005* USB Floppy/CD
Boot0006* Hard Drive


Well, that changed it from 3 to 4 (!?). Strange.

[1] "Insanity is doing the same thing over & over again and expecting a 
different result." - Einstein

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-12-21 14:07 (UTC-0500):

> 2. You will make extra work for yourself by having a common swap 
> partition for all installations. With the common swap, each new 
> installation gave rise to these messages:

>    a. "gave up waiting for suspend/resume device"

>    b. "a start job is running for dev-disk-by\..."

>    c. "failed to connect to lvmetad"

> STW can reveal ways to avoid these messages, but they are a PITA and
> avoidable by each volume group having its own swap.

Except with a first Linux installation, I tell the partitioner not to use swap,
then add it by LABEL to fstab after installing. The problem is that the
installer insists that the swap partition needs to be formatted, destroying the
validity of swap's UUID in the fstabs of the previous installations.

> There are still mysteries I have not solved. For some reason, GRUB has 
> decided that after POST, you only need 3 seconds to choose which 
> installation to boot. GRUB has resisted my efforts to change that 
> timeout value. I've been able to change the boot order in NVRAM, but not 
> the timeout. 

efibootmgr -t ##
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 12/21/2017 04:36 AM, Felix Miata wrote:

Felix Miata composed on 2017-11-29 13:55 (UTC-0500):


Dan Norton composed on 2017-11-28 22:15 (UTC-0500):

dan@debian8:~$ sudo fdisk /dev/sda
Command (m for help): p
Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24
Device Start    End    Sectors   Size Type
/dev/sda1   2048 411647 409600   200M BIOS boot
/dev/sda2 411648   16783359   16371712   7.8G Linux swap
/dev/sda3   16783360  151001087  134217728    64G Linux LVM
/dev/sda4  151001088  285218815  134217728    64G Linux LVM
/dev/sda5  285218816  419436543  134217728    64G Linux LVM
/dev/sda6  419436544  553654271  134217728    64G Linux LVM
/dev/sda7  553654272 1953525134 1399870863 667.5G Linux filesystem
Is there a problem here?

Maybe. I don't have any GPT-partitioned disks...

No longer the case. I bought a G250 Kaby Lake Intel motherboard. I currently
have Stretch, openSUSE Tumbleweed and openSUSE 15.0 Alpha installed. openSUSE
42.3's installer hangs in the bootloader configuration step.
https://bugzilla.opensuse.org/show_bug.cgi?id=1073201

http://fm.no-ip.com/Tmp/Dfsee/gb250L02.txt  is my partition log. The upper part
is generated by the partitioner I use. The bottom is gpart -l output for 
comparison.

I haven't seen you post the debian-user list in a while. How's multiboot going
for you?


Not bad, actually. I'm nearly ready to try multiboot with GPT again on 
my (elderly) HP desktop machine. It only has a 1T sda, but that seems 
like wretched excess.


Currently jessie, stretch, and buster are installed with primary/logical 
partitioning. Each is in a separate volume group, with logical volumes 
for /, /var, /tmp, /home, and swap. IMHO, the following guidelines are 
helpful:


1. Do all partitioning with the installer. Don't try to prepare the EFI 
for example with other partitioners. Partitioning can be daunting, but 
if you patiently and sometimes repeatedly use the installer UI, you can 
set up the desired partitioning. The installer UI could be improved. :-)


2. You will make extra work for yourself by having a common swap 
partition for all installations. With the common swap, each new 
installation gave rise to these messages:


  a. "gave up waiting for suspend/resume device"

  b. "a start job is running for dev-disk-by\..."

  c. "failed to connect to lvmetad"

STW can reveal ways to avoid these messages, but they are a PITA and 
avoidable by each volume group having its own swap.


There are still mysteries I have not solved. For some reason, GRUB has 
decided that after POST, you only need 3 seconds to choose which 
installation to boot. GRUB has resisted my efforts to change that 
timeout value. I've been able to change the boot order in NVRAM, but not 
the timeout.


Before moving on to multiboot with LVM and GPT, I'd like to change the 
menu entries to something more consistent. The last install is referred 
to as "Debian GNU/Linux" but that's ambiguous. Which Debian GNU/Linux? 
If each entry was in the form "Debian GNU/Linux 9 (stretch) (on 
/dev/mapper/vol2-root)" that would really be explicit. Also I want some 
more time to mull over which to boot.


 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Felix Miata composed on 2017-11-29 13:55 (UTC-0500):

> Dan Norton composed on 2017-11-28 22:15 (UTC-0500):

>> dan@debian8:~$ sudo fdisk /dev/sda
>> Command (m for help): p
>> Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
>> Units: sectors of 1 * 512 = 512 bytes
>> Sector size (logical/physical): 512 bytes / 512 bytes
>> I/O size (minimum/optimal): 512 bytes / 512 bytes
>> Disklabel type: gpt
>> Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24

>> Device Start    End    Sectors   Size Type
>> /dev/sda1   2048 411647 409600   200M BIOS boot
>> /dev/sda2 411648   16783359   16371712   7.8G Linux swap
>> /dev/sda3   16783360  151001087  134217728    64G Linux LVM
>> /dev/sda4  151001088  285218815  134217728    64G Linux LVM
>> /dev/sda5  285218816  419436543  134217728    64G Linux LVM
>> /dev/sda6  419436544  553654271  134217728    64G Linux LVM
>> /dev/sda7  553654272 1953525134 1399870863 667.5G Linux filesystem

>> Is there a problem here?

> Maybe. I don't have any GPT-partitioned disks...  

No longer the case. I bought a G250 Kaby Lake Intel motherboard. I currently
have Stretch, openSUSE Tumbleweed and openSUSE 15.0 Alpha installed. openSUSE
42.3's installer hangs in the bootloader configuration step.
https://bugzilla.opensuse.org/show_bug.cgi?id=1073201

http://fm.no-ip.com/Tmp/Dfsee/gb250L02.txt is my partition log. The upper part
is generated by the partitioner I use. The bottom is gpart -l output for 
comparison.

I haven't seen you post the debian-user list in a while. How's multiboot going
for you?
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/29/2017 04:29 PM, Dan Norton wrote:



On 11/29/2017 03:57 PM, Joe wrote:

On Wed, 29 Nov 2017 15:37:46 -0500
Dan Norton  wrote:


On 11/29/2017 01:36 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-29 13:07 (UTC-0500):

After POST, the following appears:
[...]
PXE-E53: No boot filename received
PXE-MOF: Exiting PXE ROM.
ERROR:No boot disk has been detected or the disk has failed.

It tries to PXE boot because it finds no bootable storage device:
there is no active partition, or no boot code in the MBR sector of
the device that you expect to boot. Installing Grub to a partition
leaves the MBR untouched, so it might not yet contain anything
other than partition data. The partitioner I use installs MBR code
automatically. The one in the installer may have needed to have
this step explicitly asked for. It can be added post-install
manually. 'man install-mbr'.

Isn't that because the primary is not mounted to /boot?

Mounting happens well after the point you have reached.

Nothing found for "man install-mbr" but web search yields "How to
Install Grub Onto Your MBR [1] which I will try.

[1] http://www.av8n.com/computer/htm/grub-reinstall.htm


You've said that you can boot your system with the aid of a boot
utility disc: that bypasses a lot of trouble, and you should be able to
go directly to step 8 from within your working system. It's harder to
do if you have to work from a different environment.


Good. Actually, I started the procedure on my booted jessie system but 
stopped and booted the live cd when I re-read [1].

As I said, I believe you should also do update-grub, which will
certainly do no harm. You don't yet know that grub is configured with
the necessary information for an unaided boot.


I ran update-grub recently but will be glad to do it again.


dan@debian8:~$ sudo update-grub
[sudo] password for dan:
Generating grub configuration file ...
Found background image: /usr/share/images/desktop-base/desktop-grub.png
Found linux image: /boot/vmlinuz-3.16.0-4-amd64
Found initrd image: /boot/initrd.img-3.16.0-4-amd64
done

Re: Debian 8 and Debian 9 Dual Boot

[Joe]

On Wed, 29 Nov 2017 16:06:53 -0500
Dan Norton  wrote:


> >
> > [1] http://www.av8n.com/computer/htm/grub-reinstall.htm  
> 
> Well that won't fly. Booted Debian-Live 8.8.0 amd64 Standard and
> reached Step 8 in section 1.1 of [1] which called for "grub-install 
> --root-directory=/x $drive" but grub-install is not found.
> 

/usr/sbin/grub-install is part of grub2-common, which is of optional
priority, but really ought to be part of a live CD distribution.

-- 
Joe 

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/29/2017 03:57 PM, Joe wrote:

On Wed, 29 Nov 2017 15:37:46 -0500
Dan Norton  wrote:


On 11/29/2017 01:36 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-29 13:07 (UTC-0500):

After POST, the following appears:
[...]
PXE-E53: No boot filename received
PXE-MOF: Exiting PXE ROM.
ERROR:No boot disk has been detected or the disk has failed.

It tries to PXE boot because it finds no bootable storage device:
there is no active partition, or no boot code in the MBR sector of
the device that you expect to boot. Installing Grub to a partition
leaves the MBR untouched, so it might not yet contain anything
other than partition data. The partitioner I use installs MBR code
automatically. The one in the installer may have needed to have
this step explicitly asked for. It can be added post-install
manually. 'man install-mbr'.

Isn't that because the primary is not mounted to /boot?

Mounting happens well after the point you have reached.
  

Nothing found for "man install-mbr" but web search yields "How to
Install Grub Onto Your MBR [1] which I will try.

[1] http://www.av8n.com/computer/htm/grub-reinstall.htm


You've said that you can boot your system with the aid of a boot
utility disc: that bypasses a lot of trouble, and you should be able to
go directly to step 8 from within your working system. It's harder to
do if you have to work from a different environment.


Good. Actually, I started the procedure on my booted jessie system but 
stopped and booted the live cd when I re-read [1].

As I said, I believe you should also do update-grub, which will
certainly do no harm. You don't yet know that grub is configured with
the necessary information for an unaided boot.


I ran update-grub recently but will be glad to do it again.



Note that grub2 is still a work in progress, and many of the boot
problem tutorials you find on the Net are no longer completely
accurate. This one should be OK, I think. It's usually worth adding
'debian' to your search keywords, if you turn up something on the
debian.org site there's a good chance it's up to date.

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/29/2017 03:37 PM, Dan Norton wrote:



On 11/29/2017 01:36 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-29 13:07 (UTC-0500):

After POST, the following appears:
[...]
PXE-E53: No boot filename received
PXE-MOF: Exiting PXE ROM.
ERROR:No boot disk has been detected or the disk has failed.
It tries to PXE boot because it finds no bootable storage device: 
there is no
active partition, or no boot code in the MBR sector of the device 
that you
expect to boot. Installing Grub to a partition leaves the MBR 
untouched, so it
might not yet contain anything other than partition data. The 
partitioner I use
installs MBR code automatically. The one in the installer may have 
needed to
have this step explicitly asked for. It can be added post-install 
manually. 'man

install-mbr'.


Isn't that because the primary is not mounted to /boot?

Mounting happens well after the point you have reached.



Nothing found for "man install-mbr" but web search yields "How to 
Install Grub Onto Your MBR [1] which I will try.


[1] http://www.av8n.com/computer/htm/grub-reinstall.htm


Well that won't fly. Booted Debian-Live 8.8.0 amd64 Standard and reached 
Step 8 in section 1.1 of [1] which called for "grub-install 
--root-directory=/x $drive" but grub-install is not found.

Re: Debian 8 and Debian 9 Dual Boot

[Joe]

On Wed, 29 Nov 2017 15:37:46 -0500
Dan Norton  wrote:

> On 11/29/2017 01:36 PM, Felix Miata wrote:
> > Dan Norton composed on 2017-11-29 13:07 (UTC-0500):  
> >> After POST, the following appears:
> >> [...]
> >> PXE-E53: No boot filename received
> >> PXE-MOF: Exiting PXE ROM.
> >> ERROR:No boot disk has been detected or the disk has failed.  
> > It tries to PXE boot because it finds no bootable storage device:
> > there is no active partition, or no boot code in the MBR sector of
> > the device that you expect to boot. Installing Grub to a partition
> > leaves the MBR untouched, so it might not yet contain anything
> > other than partition data. The partitioner I use installs MBR code
> > automatically. The one in the installer may have needed to have
> > this step explicitly asked for. It can be added post-install
> > manually. 'man install-mbr'. 
> >> Isn't that because the primary is not mounted to /boot?  
> > Mounting happens well after the point you have reached.
> >  
> 
> Nothing found for "man install-mbr" but web search yields "How to 
> Install Grub Onto Your MBR [1] which I will try.
> 
> [1] http://www.av8n.com/computer/htm/grub-reinstall.htm
> 

You've said that you can boot your system with the aid of a boot
utility disc: that bypasses a lot of trouble, and you should be able to
go directly to step 8 from within your working system. It's harder to
do if you have to work from a different environment.

As I said, I believe you should also do update-grub, which will
certainly do no harm. You don't yet know that grub is configured with
the necessary information for an unaided boot.

Note that grub2 is still a work in progress, and many of the boot
problem tutorials you find on the Net are no longer completely
accurate. This one should be OK, I think. It's usually worth adding
'debian' to your search keywords, if you turn up something on the
debian.org site there's a good chance it's up to date.

-- 
Joe

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/29/2017 01:36 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-29 13:07 (UTC-0500):

After POST, the following appears:
[...]
PXE-E53: No boot filename received
PXE-MOF: Exiting PXE ROM.
ERROR:No boot disk has been detected or the disk has failed.

It tries to PXE boot because it finds no bootable storage device: there is no
active partition, or no boot code in the MBR sector of the device that you
expect to boot. Installing Grub to a partition leaves the MBR untouched, so it
might not yet contain anything other than partition data. The partitioner I use
installs MBR code automatically. The one in the installer may have needed to
have this step explicitly asked for. It can be added post-install manually. 'man
install-mbr'.


Isn't that because the primary is not mounted to /boot?

Mounting happens well after the point you have reached.



Nothing found for "man install-mbr" but web search yields "How to 
Install Grub Onto Your MBR [1] which I will try.


[1] http://www.av8n.com/computer/htm/grub-reinstall.htm

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Felix Miata composed on 2017-11-29 13:55 (UTC-0500):

> Dan Norton composed on 2017-11-28 22:15 (UTC-0500):

>> dan@debian8:~$ sudo fdisk /dev/sda
>> Command (m for help): p
>> Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
>> Units: sectors of 1 * 512 = 512 bytes
>> Sector size (logical/physical): 512 bytes / 512 bytes
>> I/O size (minimum/optimal): 512 bytes / 512 bytes
>> Disklabel type: gpt
>> Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24

>> Device Start    End    Sectors   Size Type
>> /dev/sda1   2048 411647 409600   200M BIOS boot
>> /dev/sda2 411648   16783359   16371712   7.8G Linux swap
>> /dev/sda3   16783360  151001087  134217728    64G Linux LVM
>> /dev/sda4  151001088  285218815  134217728    64G Linux LVM
>> /dev/sda5  285218816  419436543  134217728    64G Linux LVM
>> /dev/sda6  419436544  553654271  134217728    64G Linux LVM
>> /dev/sda7  553654272 1953525134 1399870863 667.5G Linux filesystem

>> Is there a problem here?

> Maybe. I don't have any GPT-partitioned disks to check whether there is a
> missing column. On DOS-partitioned disks, there is a column in between 
> "Device"
> and "Start" labeled "Boot", where any boot flags that exist are shown.

> Since your HD is <2GB, one way to make all this trouble go away would be to
> start all over, selecting DOS as the partitioning type, make a primary as a 
> home
> for Grub, keep LVM on logicals, and put swap either on primary or logical. The
> primary with Grub you would mount on /boot for your first installation, but
> until you understand better the foibles of multiboot, not those subsequent.
Additional considerations that may be at work here:
https://www.rodsbooks.com/gdisk/bios.html

How old is the BIOS/Firmware on this HP PC? Is an upgrade available?
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-11-28 22:15 (UTC-0500):

> dan@debian8:~$ sudo fdisk /dev/sda
> Command (m for help): p
> Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
> Units: sectors of 1 * 512 = 512 bytes
> Sector size (logical/physical): 512 bytes / 512 bytes
> I/O size (minimum/optimal): 512 bytes / 512 bytes
> Disklabel type: gpt
> Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24

> Device Start    End    Sectors   Size Type
> /dev/sda1   2048 411647 409600   200M BIOS boot
> /dev/sda2 411648   16783359   16371712   7.8G Linux swap
> /dev/sda3   16783360  151001087  134217728    64G Linux LVM
> /dev/sda4  151001088  285218815  134217728    64G Linux LVM
> /dev/sda5  285218816  419436543  134217728    64G Linux LVM
> /dev/sda6  419436544  553654271  134217728    64G Linux LVM
> /dev/sda7  553654272 1953525134 1399870863 667.5G Linux filesystem

> Is there a problem here?
Maybe. I don't have any GPT-partitioned disks to check whether there is a
missing column. On DOS-partitioned disks, there is a column in between "Device"
and "Start" labeled "Boot", where any boot flags that exist are shown.

Since your HD is <2GB, one way to make all this trouble go away would be to
start all over, selecting DOS as the partitioning type, make a primary as a home
for Grub, keep LVM on logicals, and put swap either on primary or logical. The
primary with Grub you would mount on /boot for your first installation, but
until you understand better the foibles of multiboot, not those subsequent.
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Joe]

On Wed, 29 Nov 2017 18:19:13 +
Joe  wrote:

Wrong button. Let's try again:

> On Wed, 29 Nov 2017 13:07:01 -0500
> Dan Norton  wrote:
> 

> > 
> > After POST, the following appears:
> > 
> > [...]
> > PXE-E53: No boot filename received
> > PXE-MOF: Exiting PXE ROM.
> > ERROR:No boot disk has been detected or the disk has failed.
> > 
> > Isn't that because the primary is not mounted to /boot?

This goes back to before anything even knows what or where /boot is,
or partitions, or LVM, or anything. Some code needs to be written to the
drive to get the whole process started and to call grub. The hardware
will reset to the very beginning of the drive, and it needs to find the
first stage of a bootloader there. At the moment, it isn't.

> > 
> > Sorry to be so dense, but how do I "have the primary mounted
> > to /boot?"

At this point, with apparently nothing to lose, I'd get back into the
new installation with the boot disc, then try:

grub-install /dev/
 (which will get some bootloader code in place)

update-grub
 (which will look around, configure grub to find its own / and /boot
 and any other installed OSes, and recreate an initial ramdisk)

It looks as if no bootloader code was written to the drive during
installation. Let us hope that whatever prevented it does not prevent
grub-install from working.

-- 
Joe

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-11-29 13:07 (UTC-0500):
> After POST, the following appears:

> [...]
> PXE-E53: No boot filename received
> PXE-MOF: Exiting PXE ROM.
> ERROR:No boot disk has been detected or the disk has failed.
It tries to PXE boot because it finds no bootable storage device: there is no
active partition, or no boot code in the MBR sector of the device that you
expect to boot. Installing Grub to a partition leaves the MBR untouched, so it
might not yet contain anything other than partition data. The partitioner I use
installs MBR code automatically. The one in the installer may have needed to
have this step explicitly asked for. It can be added post-install manually. 'man
install-mbr'.

> Isn't that because the primary is not mounted to /boot?

Mounting happens well after the point you have reached.

> Sorry to be so dense, but how do I "have the primary mounted to /boot?"
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Joe]

On Wed, 29 Nov 2017 13:07:01 -0500
Dan Norton  wrote:

> On 11/16/2017 03:13 PM, Felix Miata wrote:
> > Joe composed on 2017-11-16 19:27 (UTC):
> >  
> >> I see the date of the page is 2015,  
> > IIRC, that 2015 update was all about updating and/or replacing
> > broken links. 
> >> Multiple OSes will still use one overall bootloader, which may or
> >> may not redirect to other bootloaders. It is common, therefore, to
> >> use a single physical partition for /boot, common to all *nix
> >> OSes  
> > Mounting the same single primary partition as /boot to all *nix
> > OSes invites virtually the same trouble that installing Grub to MBR
> > causes. Absent user intervention, each OS thinks it owns the space,
> > so at updates times, each stomps on whatever is there at the time,
> > eventually exhausting its freespace with a multitude of diverse
> > kernels and initrds. Only one OS, or no OS, should mount the
> > bootable physical partition to /boot. Others can either have their
> > own bootloader installed to their own / filesystem, or be used
> > without their own bootloader.
> >
> > On most of my PCs, the bootable primary is never mounted to /boot.
> > I manage it manually myself, providing direct loading of kernels
> > and initrds, as well as chainloading entries, and configfile
> > entries, the latter two used primarily when prior kernels are to be
> > booted, the former for most booting.
> >
> > For someone doing multiboot the first time, it's easiest, if not a
> > requirement, for the first installation to have the primary mounted
> > to /boot. The subsequent installations either need to have no
> > bootloader or bootloader only installed to /; or, the newer can be
> > configured to mount the primary on /boot, and the original, prior
> > to its next boot, preferably immediately prior to starting the new
> > installation, reconfigured to not mount the primary on /boot, not
> > manage the bootloader installed there, and relocate its content to
> > its own / filesystem.  
> 
> After POST, the following appears:
> 
> [...]
> PXE-E53: No boot filename received
> PXE-MOF: Exiting PXE ROM.
> ERROR:No boot disk has been detected or the disk has failed.
> 
> Isn't that because the primary is not mounted to /boot?
> 
> Sorry to be so dense, but how do I "have the primary mounted
> to /boot?"
> 
> 
> > Multiboot setup is complex, one of the reasons why virtualization
> > became popular. There is no single right way to get where you want
> > to go. For a beginner it helps a lot to have an extra PC to do the
> > learning, and for doc access while doing the installations and
> > configurations on the other. :-)  
> 

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/16/2017 03:13 PM, Felix Miata wrote:

Joe composed on 2017-11-16 19:27 (UTC):


I see the date of the page is 2015,

IIRC, that 2015 update was all about updating and/or replacing broken links.


Multiple OSes will still use one overall bootloader, which may or may
not redirect to other bootloaders. It is common, therefore, to use a
single physical partition for /boot, common to all *nix OSes

Mounting the same single primary partition as /boot to all *nix OSes invites
virtually the same trouble that installing Grub to MBR causes. Absent user
intervention, each OS thinks it owns the space, so at updates times, each stomps
on whatever is there at the time, eventually exhausting its freespace with a
multitude of diverse kernels and initrds. Only one OS, or no OS, should mount
the bootable physical partition to /boot. Others can either have their own
bootloader installed to their own / filesystem, or be used without their own
bootloader.

On most of my PCs, the bootable primary is never mounted to /boot. I manage it
manually myself, providing direct loading of kernels and initrds, as well as
chainloading entries, and configfile entries, the latter two used primarily when
prior kernels are to be booted, the former for most booting.

For someone doing multiboot the first time, it's easiest, if not a requirement,
for the first installation to have the primary mounted to /boot. The subsequent
installations either need to have no bootloader or bootloader only installed to
/; or, the newer can be configured to mount the primary on /boot, and the
original, prior to its next boot, preferably immediately prior to starting the
new installation, reconfigured to not mount the primary on /boot, not manage the
bootloader installed there, and relocate its content to its own / filesystem.


After POST, the following appears:

[...]
PXE-E53: No boot filename received
PXE-MOF: Exiting PXE ROM.
ERROR:No boot disk has been detected or the disk has failed.

Isn't that because the primary is not mounted to /boot?

Sorry to be so dense, but how do I "have the primary mounted to /boot?"



Multiboot setup is complex, one of the reasons why virtualization became
popular. There is no single right way to get where you want to go. For a
beginner it helps a lot to have an extra PC to do the learning, and for doc
access while doing the installations and configurations on the other. :-)

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/29/2017 03:59 AM, Joe wrote:

On Tue, 28 Nov 2017 22:15:22 -0500
Dan Norton  wrote:


On 11/13/2017 01:55 PM, Joe wrote:

On Mon, 13 Nov 2017 11:01:27 -0500
Dan Norton  wrote:

  

Although I didn't say so, each install would have its own set of
directories. Please say more about how to mount the other
installation and share data. How to mount things in another volume
group?
  

Good advice so far, but to add a bit: all LVM groups will be seen at
boot, and /dev will know about them. See man lvm2 and also here:

https://wiki.debian.org/LVM  for complete information about the
commands you have available. There are also numerous tutorials on
the Net which show basic usage of the simpler commands. It's worth
having a look when you have some spare time, as one day you'll need
to know some of this and won't have any spare time.

Reading the wiki reveals "Grub and ?
LiLo are not compatible with LVM,
so /bootshould be outside the storage disk managed by LVM."

Not many people are using lilo today, and a recent grub2 should have no
problem at all with lvm. But there's no harm in a separate /boot, and
I've always done it that way.


Here's
what I have:

Attempts to boot normally do not work. But using Super Grub2 on a
bootable cd and selecting "Boot manually" and picking the hd1 entry
brings up the jessie system that the installer reports as
successfully installed on sda3. Using fdisk to take a look:

dan@debian8:~$ sudo fdisk /dev/sda
Command (m for help): p
Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24

Device Start    End    Sectors   Size Type
/dev/sda1   2048 411647 409600   200M BIOS boot
/dev/sda2 411648   16783359   16371712   7.8G Linux swap
/dev/sda3   16783360  151001087  134217728    64G Linux LVM
/dev/sda4  151001088  285218815  134217728    64G Linux LVM
/dev/sda5  285218816  419436543  134217728    64G Linux LVM
/dev/sda6  419436544  553654271  134217728    64G Linux LVM
/dev/sda7  553654272 1953525134 1399870863 667.5G Linux filesystem

Is there a problem here?

Not obviously so. But you know that you have a working installation,
just one with a booting problem, which suggests this is all plausible.


dan@debian8:~$ df -h
Filesystem    Size  Used Avail Use% Mounted on
/dev/dm-0 9.1G  3.0G  5.7G  35% /
udev   10M 0   10M   0% /dev
tmpfs 775M  9.0M  766M   2% /run
tmpfs 1.9G   68K  1.9G   1% /dev/shm
tmpfs 5.0M  4.0K  5.0M   1% /run/lock
tmpfs 1.9G 0  1.9G   0% /sys/fs/cgroup
/dev/sda1 992K  142K  851K  15% /boot/efi
/dev/mapper/debian8--vg-var   8.2G  1.3G  6.4G  17% /var
/dev/mapper/debian8--vg-home  9.1G  356M  8.3G   5% /home
/dev/mapper/debian8--vg-tmp   268M  2.1M  247M   1% /tmp
tmpfs 388M  4.0K  388M   1% /run/user/115
tmpfs 388M   12K  388M   1% /run/user/1000


Doesn't this satisfy the statement that "/boot should be outside the
storage disk managed by LVM" since it is on sda1?


Look in /etc/fstab for lines beginning /dev/mapper/[volume] which
will be the volumes mounted in the running installation. The
'mapper' is turning LVM volumes into things which look like
partitions for many purposes.

Here is fstab:
# /etc/fstab: static file system information.
#
# Use 'blkid' to print the universally unique identifier for a
# device; this may be used with UUID= as a more robust way to name
devices # that works even if disks are added and removed. See
fstab(5). #
#           
/dev/mapper/debian8--vg-root /  ext4 errors=remount-ro
0   1 # /boot/efi was on /dev/sda1 during installation
UUID=B07E-1F0B  /boot/efi   vfat    umask=0077  0   1
/dev/mapper/debian8--vg-home /home  ext4 defaults
0   2 /dev/mapper/debian8--vg-tmp /tmp    ext4
defaults    0   2 /dev/mapper/debian8--vg-var /var
ext4 defaults    0   2 # swap was on /dev/sda2 during
installation UUID=6aa1846f-34dd-424d-b02c-dbd0af037a23
none    swap sw  0   0
/dev/sr0    /media/cdrom0   udf,iso9660 user,noauto 0   0


Why won't it boot normally, that is without using the bootable Grub2
cd?


I've had enormous amounts of trouble with grub/grub2 over the years.
Could be anything.

We have to start from the exact screen messages you get when you try
booting. There are different classes of grub problem, and it will drop
you in different shells depending on what it has found.

Please, the error messages, word for word...



It doesn't get as far as grub. These are coming from bios(?) 

Re: Debian 8 and Debian 9 Dual Boot

[Jonathan Dowland]

On Mon, Nov 13, 2017 at 09:38:59PM -0500, Dan Norton wrote:
How do you set up LVM so that install does not clobber one of the 
siblings? During installation, for partitioning, do you pick "Manual" 
or "Guided -*" ?


Manual. Guided -* is very likely to result in you clobbering the
existing partitions or volume group by accident. (No offense intended.
I've done it myself very recently.)

--

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

Re: Debian 8 and Debian 9 Dual Boot

[Joe]

On Tue, 28 Nov 2017 22:15:22 -0500
Dan Norton  wrote:

> On 11/13/2017 01:55 PM, Joe wrote:
> > On Mon, 13 Nov 2017 11:01:27 -0500
> > Dan Norton  wrote:
> >
> >  
> >> Although I didn't say so, each install would have its own set of
> >> directories. Please say more about how to mount the other
> >> installation and share data. How to mount things in another volume
> >> group?
> >>  
> > Good advice so far, but to add a bit: all LVM groups will be seen at
> > boot, and /dev will know about them. See man lvm2 and also here:
> >
> > https://wiki.debian.org/LVM  for complete information about the
> > commands you have available. There are also numerous tutorials on
> > the Net which show basic usage of the simpler commands. It's worth
> > having a look when you have some spare time, as one day you'll need
> > to know some of this and won't have any spare time.  
> 
> Reading the wiki reveals "Grub and ?
> LiLo are not compatible with LVM,
> so /bootshould be outside the storage disk managed by LVM."

Not many people are using lilo today, and a recent grub2 should have no
problem at all with lvm. But there's no harm in a separate /boot, and
I've always done it that way. 

> Here's
> what I have:
> 
> Attempts to boot normally do not work. But using Super Grub2 on a 
> bootable cd and selecting "Boot manually" and picking the hd1 entry 
> brings up the jessie system that the installer reports as
> successfully installed on sda3. Using fdisk to take a look:
> 
> dan@debian8:~$ sudo fdisk /dev/sda
> Command (m for help): p
> Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
> Units: sectors of 1 * 512 = 512 bytes
> Sector size (logical/physical): 512 bytes / 512 bytes
> I/O size (minimum/optimal): 512 bytes / 512 bytes
> Disklabel type: gpt
> Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24
> 
> Device Start    End    Sectors   Size Type
> /dev/sda1   2048 411647 409600   200M BIOS boot
> /dev/sda2 411648   16783359   16371712   7.8G Linux swap
> /dev/sda3   16783360  151001087  134217728    64G Linux LVM
> /dev/sda4  151001088  285218815  134217728    64G Linux LVM
> /dev/sda5  285218816  419436543  134217728    64G Linux LVM
> /dev/sda6  419436544  553654271  134217728    64G Linux LVM
> /dev/sda7  553654272 1953525134 1399870863 667.5G Linux filesystem
> 
> Is there a problem here?

Not obviously so. But you know that you have a working installation,
just one with a booting problem, which suggests this is all plausible.

> 
> dan@debian8:~$ df -h
> Filesystem    Size  Used Avail Use% Mounted on
> /dev/dm-0 9.1G  3.0G  5.7G  35% /
> udev   10M 0   10M   0% /dev
> tmpfs 775M  9.0M  766M   2% /run
> tmpfs 1.9G   68K  1.9G   1% /dev/shm
> tmpfs 5.0M  4.0K  5.0M   1% /run/lock
> tmpfs 1.9G 0  1.9G   0% /sys/fs/cgroup
> /dev/sda1 992K  142K  851K  15% /boot/efi
> /dev/mapper/debian8--vg-var   8.2G  1.3G  6.4G  17% /var
> /dev/mapper/debian8--vg-home  9.1G  356M  8.3G   5% /home
> /dev/mapper/debian8--vg-tmp   268M  2.1M  247M   1% /tmp
> tmpfs 388M  4.0K  388M   1% /run/user/115
> tmpfs 388M   12K  388M   1% /run/user/1000
> 
> 
> Doesn't this satisfy the statement that "/boot should be outside the 
> storage disk managed by LVM" since it is on sda1?
> 
> > Look in /etc/fstab for lines beginning /dev/mapper/[volume] which
> > will be the volumes mounted in the running installation. The
> > 'mapper' is turning LVM volumes into things which look like
> > partitions for many purposes.  
> 
> Here is fstab:
> # /etc/fstab: static file system information.
> #
> # Use 'blkid' to print the universally unique identifier for a
> # device; this may be used with UUID= as a more robust way to name
> devices # that works even if disks are added and removed. See
> fstab(5). #
> #           
> /dev/mapper/debian8--vg-root /  ext4 errors=remount-ro
> 0   1 # /boot/efi was on /dev/sda1 during installation
> UUID=B07E-1F0B  /boot/efi   vfat    umask=0077  0   1
> /dev/mapper/debian8--vg-home /home  ext4 defaults
> 0   2 /dev/mapper/debian8--vg-tmp /tmp    ext4
> defaults    0   2 /dev/mapper/debian8--vg-var /var
> ext4 defaults    0   2 # swap was on /dev/sda2 during
> installation UUID=6aa1846f-34dd-424d-b02c-dbd0af037a23
> none    swap sw  0   0
> /dev/sr0    /media/cdrom0   udf,iso9660 user,noauto 0   0 
> 
> 
> Why won't it boot normally, that is without using the bootable Grub2
> cd?
> 

I've had enormous amounts of trouble with grub/grub2 over the years.
Could be anything.

We have to start from the exact screen messages you get when you try
booting. There are different 

Re: Debian 8 and Debian 9 Dual Boot

[Michael Fothergill]

On 29 November 2017 at 03:15, Dan Norton  wrote:

>
> On 11/13/2017 01:55 PM, Joe wrote:
>
> On Mon, 13 Nov 2017 11:01:27 -0500
> Dan Norton   wrote:
>
>
>
> Although I didn't say so, each install would have its own set of
> directories. Please say more about how to mount the other
> installation and share data. How to mount things in another volume
> group?
>
>
> Good advice so far, but to add a bit: all LVM groups will be seen at
> boot, and /dev will know about them. See man lvm2 and also here:
> https://wiki.debian.org/LVM  for complete information about the
> commands you have available. There are also numerous tutorials on the
> Net which show basic usage of the simpler commands. It's worth having a
> look when you have some spare time, as one day you'll need to know some
> of this and won't have any spare time.
>
>
> Reading the wiki reveals "Grub and ? LiLo
> are not compatible with LVM, so /boot should be outside the storage disk
> managed by LVM." Here's what I have:
>
> Attempts to boot normally do not work. But using Super Grub2 on a bootable
> cd and selecting "Boot manually" and picking the hd1 entry brings up the
> jessie system that the installer reports as successfully installed on sda3.
> Using fdisk to take a look:
>
> dan@debian8:~$ sudo fdisk /dev/sda
> Command (m for help): p
> Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
> Units: sectors of 1 * 512 = 512 bytes
> Sector size (logical/physical): 512 bytes / 512 bytes
> I/O size (minimum/optimal): 512 bytes / 512 bytes
> Disklabel type: gpt
> Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24
>
> Device StartEndSectors   Size Type
> /dev/sda1   2048 411647 409600   200M BIOS boot
> /dev/sda2 411648   16783359   16371712   7.8G Linux swap
> /dev/sda3   16783360  151001087  13421772864G Linux LVM
> /dev/sda4  151001088  285218815  13421772864G Linux LVM
> /dev/sda5  285218816  419436543  13421772864G Linux LVM
> /dev/sda6  419436544  553654271  13421772864G Linux LVM
> /dev/sda7  553654272 1953525134 1399870863 667.5G Linux filesystem
>
> Is there a problem here?
>

​I have been using Supergrub for ages now.   If you use Gentoo as I do you
appreciate the irritation of continuous babysitting of grub2 every time
you reconfigure the kernel (e.g. after setting up the long winded nonsense
required to make a firewall work properly) then grub tends to go awol
and requires TLC, massage therapy and soft music to start working
properly.

Meanwhile you will feel so frustrated that you want to chew on barbed wire
etc.

But supergrub is cool.  It is like a bloodhound.  You just can't fool it.  ​

The more you use it the dumber you feel using grub2 until you have finally
sorted the gripes and things have calmed down and stabilised - usually a
month later.

Remember the scene from 2001 space odyssey where the monkey throws the
bones up in the air.

​Do the same with grub2, then use supergrub until things work and then
de-evolve again and vegetate in this area as I do.

Cheers​

​MF​

>
> dan@debian8:~$ df -h
> FilesystemSize  Used Avail Use% Mounted on
> /dev/dm-0 9.1G  3.0G  5.7G  35% /
> udev   10M 0   10M   0% /dev
> tmpfs 775M  9.0M  766M   2% /run
> tmpfs 1.9G   68K  1.9G   1% /dev/shm
> tmpfs 5.0M  4.0K  5.0M   1% /run/lock
> tmpfs 1.9G 0  1.9G   0% /sys/fs/cgroup
> /dev/sda1 992K  142K  851K  15% /boot/efi
> /dev/mapper/debian8--vg-var   8.2G  1.3G  6.4G  17% /var
> /dev/mapper/debian8--vg-home  9.1G  356M  8.3G   5% /home
> /dev/mapper/debian8--vg-tmp   268M  2.1M  247M   1% /tmp
> tmpfs 388M  4.0K  388M   1% /run/user/115
> tmpfs 388M   12K  388M   1% /run/user/1000
>
>
> Doesn't this satisfy the statement that "/boot should be outside the
> storage disk managed by LVM" since it is on sda1?
>
> Look in /etc/fstab for lines beginning /dev/mapper/[volume] which will
> be the volumes mounted in the running installation. The 'mapper' is
> turning LVM volumes into things which look like partitions for many
> purposes.
>
>
> Here is fstab:
> # /etc/fstab: static file system information.
> #
> # Use 'blkid' to print the universally unique identifier for a
> # device; this may be used with UUID= as a more robust way to name devices
> # that works even if disks are added and removed. See fstab(5).
> #
> #
> /dev/mapper/debian8--vg-root /  ext4errors=remount-ro
> 0   1
> # /boot/efi was on /dev/sda1 during installation
> UUID=B07E-1F0B  /boot/efi   vfatumask=0077  0   1
> /dev/mapper/debian8--vg-home /home  ext4defaults
> 0   2
> /dev/mapper/debian8--vg-tmp /tmp

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 01:55 PM, Joe wrote:

On Mon, 13 Nov 2017 11:01:27 -0500
Dan Norton  wrote:



Although I didn't say so, each install would have its own set of
directories. Please say more about how to mount the other
installation and share data. How to mount things in another volume
group?


Good advice so far, but to add a bit: all LVM groups will be seen at
boot, and /dev will know about them. See man lvm2 and also here:

https://wiki.debian.org/LVM  for complete information about the
commands you have available. There are also numerous tutorials on the
Net which show basic usage of the simpler commands. It's worth having a
look when you have some spare time, as one day you'll need to know some
of this and won't have any spare time.


Reading the wiki reveals "Grub and ? LiLo 
are not compatible with LVM, so /bootshould be outside the storage disk 
managed by LVM." Here's what I have:


Attempts to boot normally do not work. But using Super Grub2 on a 
bootable cd and selecting "Boot manually" and picking the hd1 entry 
brings up the jessie system that the installer reports as successfully 
installed on sda3. Using fdisk to take a look:


dan@debian8:~$ sudo fdisk /dev/sda
Command (m for help): p
Disk /dev/sda: 931.5 GiB, 1000204886016 bytes, 1953525168 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: A615A904-0620-459F-BF44-5E53E54FDF24

Device Start    End    Sectors   Size Type
/dev/sda1   2048 411647 409600   200M BIOS boot
/dev/sda2 411648   16783359   16371712   7.8G Linux swap
/dev/sda3   16783360  151001087  134217728    64G Linux LVM
/dev/sda4  151001088  285218815  134217728    64G Linux LVM
/dev/sda5  285218816  419436543  134217728    64G Linux LVM
/dev/sda6  419436544  553654271  134217728    64G Linux LVM
/dev/sda7  553654272 1953525134 1399870863 667.5G Linux filesystem

Is there a problem here?

dan@debian8:~$ df -h
Filesystem    Size  Used Avail Use% Mounted on
/dev/dm-0 9.1G  3.0G  5.7G  35% /
udev   10M 0   10M   0% /dev
tmpfs 775M  9.0M  766M   2% /run
tmpfs 1.9G   68K  1.9G   1% /dev/shm
tmpfs 5.0M  4.0K  5.0M   1% /run/lock
tmpfs 1.9G 0  1.9G   0% /sys/fs/cgroup
/dev/sda1 992K  142K  851K  15% /boot/efi
/dev/mapper/debian8--vg-var   8.2G  1.3G  6.4G  17% /var
/dev/mapper/debian8--vg-home  9.1G  356M  8.3G   5% /home
/dev/mapper/debian8--vg-tmp   268M  2.1M  247M   1% /tmp
tmpfs 388M  4.0K  388M   1% /run/user/115
tmpfs 388M   12K  388M   1% /run/user/1000


Doesn't this satisfy the statement that "/boot should be outside the 
storage disk managed by LVM" since it is on sda1?



Look in /etc/fstab for lines beginning /dev/mapper/[volume] which will
be the volumes mounted in the running installation. The 'mapper' is
turning LVM volumes into things which look like partitions for many
purposes.


Here is fstab:
# /etc/fstab: static file system information.
#
# Use 'blkid' to print the universally unique identifier for a
# device; this may be used with UUID= as a more robust way to name devices
# that works even if disks are added and removed. See fstab(5).
#
#           
/dev/mapper/debian8--vg-root /  ext4 errors=remount-ro 0   1
# /boot/efi was on /dev/sda1 during installation
UUID=B07E-1F0B  /boot/efi   vfat    umask=0077  0   1
/dev/mapper/debian8--vg-home /home  ext4 defaults    0   2
/dev/mapper/debian8--vg-tmp /tmp    ext4 defaults    0   2
/dev/mapper/debian8--vg-var /var    ext4 defaults    0   2
# swap was on /dev/sda2 during installation
UUID=6aa1846f-34dd-424d-b02c-dbd0af037a23 none    swap 
sw  0   0

/dev/sr0    /media/cdrom0   udf,iso9660 user,noauto 0   0


Why won't it boot normally, that is without using the bootable Grub2 cd?

 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/16/2017 02:27 PM, Joe wrote:

On Thu, 16 Nov 2017 13:34:52 -0500
Dan Norton  wrote:


On 11/13/2017 09:52 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-13 21:38 (UTC-0500):
  

How do you set up LVM so that install does not clobber one of the
siblings? During installation, for partitioning, do you pick
"Manual" or "Guided -*" ?

I don't think I've ever used Debian's, or any Debian derivative's,
"guided". I can't remember the last time I didn't pick manual,
after *fully* preparing in advance to my specifications.

Manual will list what partitions and volumes exist, so if you've
fully prepared logical volumes in advance, all that should need to
be done within manual is select mount points and mount options for
those you wish used. If volumes exist but are not formatted, then
formatting would be a third selection using manual.

A better description of what I want is "multiboot", instead of "dual
boot" in this thread. I can see having up to 4 systems installed, all
Debians, on one disk drive. There is an item in [1] which I see that
reads:

    6. Disable LVM support

which is puzzling. Can multiboot be set up within LVM on one disk
drive? Or, is it necessary to disable  LVM, define multiple logical
partitions with fdisk or parted for example, install in these
partitions, and then set up LVM separately on each system?

I see the date of the page is 2015, but to be honest, a lot of this
page is completely obsolete, even more so if you're not using DOS, OS/2
or early Windows. I would not even swear that item 6, LVM, even refers
to the Linux system. There's certainly no problem about mixing multiple
booting with lvm.


Glad to hear it!



A volume group would normally be declared within a single old-style IBM
partition, but doesn't have to be. Think of multiple hard drives, each
containing multiple old-style partitions, then think of multiple lvm
volume groups, each with multiple volumes, roughly overlaid on the
drives and partitions but not physically matching them. A volume group
can consist of multiple complete partitions on different drives, and
the volumes it contains need not bear any relationship to the
underlying partitions.

Multiple OSes will still use one overall bootloader, which may or may
not redirect to other bootloaders. It is common, therefore, to use a
single physical partition for /boot, common to all *nix OSes, but by no
means necessary. /boot can be an LVM volume if you wish, but a booting
problem is slightly easier to deal with if the kernel and grub stuff is
in a standard partition, you don't need to worry about looking up the
lvm commands in addition to the grub set.

My workstation, for example, has a 1GB primary partition for /boot, and
the rest of the drive is a second primary partition which maps exactly
to a volume group. The volume group is then divided into volumes, which
mostly act like old-style partitions, at least as far as mounting goes.
You can have a lot of them.
Apparently there are many ways to do multiboot in LVM. Is it practical 
to, on one fairly large (1TB) disk, [I can hear the snickers when 
someone reads this when 1TB is puny in a year or two] do the following:


1. Enlarge the volume group to span the entire disk (it's currently 40G).
2. Define a number of additional logical volumes, say 3.
3. Install stable in one of the lv's and have install treat it like a 
standard *partition* (logical) and have within it separate root, swap, 
/var, /tmp, and /home?


 - Dan

  


I have two drives set up the same, as I migrated from one to a newer
one. After adjustment, the grub installation of the new drive handles
the boot, and the kernel used by the old system lives in the /boot
partition of the old drive. All of this could have been integrated into
one volume group, but it's two because that's how the system evolved. I
could pull the old drive, which I hardly ever touch now, plonk it into
another PC and boot it with minimal fiddling about.

So you have a common grub, basically from the latest installation
unless you tell the installer otherwise, and before the installation is
finished it should have detected any other bootable OSes anywhere in
the hardware. You've been advised not to share other partitions, even
/home, but several /homes can live in differently-named lvm volumes in
the same volume group. Or not, by your preference. You can, for the
most part, treat volumes as partitions, and not worry about whether
they are primary or logical.

You should probably track down a few tutorials on multiboot and lvm,
read through them to get a feel for what needs doing. It's always
difficult reading a manual for something and working out from that what
steps you need to take to do something. Things like UEFI boot and disc
encryption complicate things, so probably best to avoid them on your
first attempt(s) if the hardware allows. If you're starting with a bare
drive and your BIOS allows old-style IBM partitioning, that's the way
to go 

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/16/2017 02:06 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-16 13:34 (UTC-0500):


A better description of what I want is "multiboot", instead of "dual
boot" in this thread. I can see having up to 4 systems installed, all
Debians, on one disk drive. There is an item in [1] which I see that reads:
    6. Disable LVM support

If you look farther down your cite you'll find "What is LVM?". "Disabling LVM"
is only applicable to the OS/2 type of LVM. It causes the OS/2 kernel to use
native DOS-style FIFO assignment of drive letters.


which is puzzling. Can multiboot be set up within LVM on one disk drive?

Yes!



OK, great!

Now I'll try to figure out whether each system should have its own 
volume group or its own logical volume within one vg. Would install go 
easier one way or the other?

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Joe composed on 2017-11-16 19:27 (UTC):

> I see the date of the page is 2015,

IIRC, that 2015 update was all about updating and/or replacing broken links.

> Multiple OSes will still use one overall bootloader, which may or may
> not redirect to other bootloaders. It is common, therefore, to use a
> single physical partition for /boot, common to all *nix OSes

Mounting the same single primary partition as /boot to all *nix OSes invites
virtually the same trouble that installing Grub to MBR causes. Absent user
intervention, each OS thinks it owns the space, so at updates times, each stomps
on whatever is there at the time, eventually exhausting its freespace with a
multitude of diverse kernels and initrds. Only one OS, or no OS, should mount
the bootable physical partition to /boot. Others can either have their own
bootloader installed to their own / filesystem, or be used without their own
bootloader.

On most of my PCs, the bootable primary is never mounted to /boot. I manage it
manually myself, providing direct loading of kernels and initrds, as well as
chainloading entries, and configfile entries, the latter two used primarily when
prior kernels are to be booted, the former for most booting.

For someone doing multiboot the first time, it's easiest, if not a requirement,
for the first installation to have the primary mounted to /boot. The subsequent
installations either need to have no bootloader or bootloader only installed to
/; or, the newer can be configured to mount the primary on /boot, and the
original, prior to its next boot, preferably immediately prior to starting the
new installation, reconfigured to not mount the primary on /boot, not manage the
bootloader installed there, and relocate its content to its own / filesystem.

Multiboot setup is complex, one of the reasons why virtualization became
popular. There is no single right way to get where you want to go. For a
beginner it helps a lot to have an extra PC to do the learning, and for doc
access while doing the installations and configurations on the other. :-)
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Joe]

On Thu, 16 Nov 2017 13:34:52 -0500
Dan Norton  wrote:

> On 11/13/2017 09:52 PM, Felix Miata wrote:
> > Dan Norton composed on 2017-11-13 21:38 (UTC-0500):
> >  
> >> How do you set up LVM so that install does not clobber one of the
> >> siblings? During installation, for partitioning, do you pick
> >> "Manual" or "Guided -*" ?  
> > I don't think I've ever used Debian's, or any Debian derivative's,
> > "guided". I can't remember the last time I didn't pick manual,
> > after *fully* preparing in advance to my specifications.
> >
> > Manual will list what partitions and volumes exist, so if you've
> > fully prepared logical volumes in advance, all that should need to
> > be done within manual is select mount points and mount options for
> > those you wish used. If volumes exist but are not formatted, then
> > formatting would be a third selection using manual.  
> 
> A better description of what I want is "multiboot", instead of "dual 
> boot" in this thread. I can see having up to 4 systems installed, all 
> Debians, on one disk drive. There is an item in [1] which I see that
> reads:
> 
>    6. Disable LVM support
> 
> which is puzzling. Can multiboot be set up within LVM on one disk
> drive? Or, is it necessary to disable  LVM, define multiple logical
> partitions with fdisk or parted for example, install in these
> partitions, and then set up LVM separately on each system?

I see the date of the page is 2015, but to be honest, a lot of this
page is completely obsolete, even more so if you're not using DOS, OS/2
or early Windows. I would not even swear that item 6, LVM, even refers
to the Linux system. There's certainly no problem about mixing multiple
booting with lvm.

A volume group would normally be declared within a single old-style IBM
partition, but doesn't have to be. Think of multiple hard drives, each
containing multiple old-style partitions, then think of multiple lvm
volume groups, each with multiple volumes, roughly overlaid on the
drives and partitions but not physically matching them. A volume group
can consist of multiple complete partitions on different drives, and
the volumes it contains need not bear any relationship to the
underlying partitions.

Multiple OSes will still use one overall bootloader, which may or may
not redirect to other bootloaders. It is common, therefore, to use a
single physical partition for /boot, common to all *nix OSes, but by no
means necessary. /boot can be an LVM volume if you wish, but a booting
problem is slightly easier to deal with if the kernel and grub stuff is
in a standard partition, you don't need to worry about looking up the
lvm commands in addition to the grub set.

My workstation, for example, has a 1GB primary partition for /boot, and
the rest of the drive is a second primary partition which maps exactly
to a volume group. The volume group is then divided into volumes, which
mostly act like old-style partitions, at least as far as mounting goes.
You can have a lot of them. 

I have two drives set up the same, as I migrated from one to a newer
one. After adjustment, the grub installation of the new drive handles
the boot, and the kernel used by the old system lives in the /boot
partition of the old drive. All of this could have been integrated into
one volume group, but it's two because that's how the system evolved. I
could pull the old drive, which I hardly ever touch now, plonk it into
another PC and boot it with minimal fiddling about.

So you have a common grub, basically from the latest installation
unless you tell the installer otherwise, and before the installation is
finished it should have detected any other bootable OSes anywhere in
the hardware. You've been advised not to share other partitions, even
/home, but several /homes can live in differently-named lvm volumes in
the same volume group. Or not, by your preference. You can, for the
most part, treat volumes as partitions, and not worry about whether
they are primary or logical.

You should probably track down a few tutorials on multiboot and lvm,
read through them to get a feel for what needs doing. It's always
difficult reading a manual for something and working out from that what
steps you need to take to do something. Things like UEFI boot and disc
encryption complicate things, so probably best to avoid them on your
first attempt(s) if the hardware allows. If you're starting with a bare
drive and your BIOS allows old-style IBM partitioning, that's the way
to go to get the hang of this.

-- 
Joe

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-11-16 13:34 (UTC-0500):

> A better description of what I want is "multiboot", instead of "dual 
> boot" in this thread. I can see having up to 4 systems installed, all 
> Debians, on one disk drive. There is an item in [1] which I see that reads:

>    6. Disable LVM support

If you look farther down your cite you'll find "What is LVM?". "Disabling LVM"
is only applicable to the OS/2 type of LVM. It causes the OS/2 kernel to use
native DOS-style FIFO assignment of drive letters.

> which is puzzling. Can multiboot be set up within LVM on one disk drive? 

Yes!

> Or, is it necessary to disable  LVM, define multiple logical partitions 
> with fdisk or parted for example, install in these partitions, and then 
> set up LVM separately on each system?

That would defeat most of the value of Linux LVM.

> [1] http://fm.no-ip.com/PC/partitioningindex.html
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 09:52 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-13 21:38 (UTC-0500):


How do you set up LVM so that install does not clobber one of the
siblings? During installation, for partitioning, do you pick "Manual" or
"Guided -*" ?

I don't think I've ever used Debian's, or any Debian derivative's, "guided". I
can't remember the last time I didn't pick manual, after *fully* preparing in
advance to my specifications.

Manual will list what partitions and volumes exist, so if you've fully prepared
logical volumes in advance, all that should need to be done within manual is
select mount points and mount options for those you wish used. If volumes exist
but are not formatted, then formatting would be a third selection using manual.


A better description of what I want is "multiboot", instead of "dual 
boot" in this thread. I can see having up to 4 systems installed, all 
Debians, on one disk drive. There is an item in [1] which I see that reads:


  6. Disable LVM support

which is puzzling. Can multiboot be set up within LVM on one disk drive? 
Or, is it necessary to disable  LVM, define multiple logical partitions 
with fdisk or parted for example, install in these partitions, and then 
set up LVM separately on each system?


[1] http://fm.no-ip.com/PC/partitioningindex.html

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 09:52 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-13 21:38 (UTC-0500):


How do you set up LVM so that install does not clobber one of the
siblings? During installation, for partitioning, do you pick "Manual" or
"Guided -*" ?

I don't think I've ever used Debian's, or any Debian derivative's, "guided". I
can't remember the last time I didn't pick manual, after *fully* preparing in
advance to my specifications.

Manual will list what partitions and volumes exist, so if you've fully prepared
logical volumes in advance, all that should need to be done within manual is
select mount points and mount options for those you wish used. If volumes exist
but are not formatted, then formatting would be a third selection using manual.


OK, I may be able to do that. Thanks for your help.

 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-11-13 21:38 (UTC-0500):

> How do you set up LVM so that install does not clobber one of the 
> siblings? During installation, for partitioning, do you pick "Manual" or 
> "Guided -*" ?

I don't think I've ever used Debian's, or any Debian derivative's, "guided". I
can't remember the last time I didn't pick manual, after *fully* preparing in
advance to my specifications.

Manual will list what partitions and volumes exist, so if you've fully prepared
logical volumes in advance, all that should need to be done within manual is
select mount points and mount options for those you wish used. If volumes exist
but are not formatted, then formatting would be a third selection using manual.
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 07:42 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-13 18:04 (UTC-0500):


Felix Miata wrote:

There's no need for separate home file systems to keep one installation from
causing corruption to another's user settings. Simply do not reuse UIDs[1].
On your first, your Jessie probably has a user dan with UID 1000. On the next,
don't use 1000 for user dan. Pick any other number, e.g. 1200 for user dan in
Stretch. During Stretch installation make the initial user dan8 or dan01 using
Jessie's UID (probably 1000) for dan.

Do you pick a UID during install somehow or do you login as root before
anything else and edit /etc/passwd?

I could have been clearer. "During installation" isn't necessarily the correct
time for user management. I don't remember whether Stretch forces ordinary user
creation at installation time, or leaves it up to the admin to do when he 
chooses.

With installers that force creation of an ordinary user in addition to providing
a root password, or that don't permit UID selection for an ordinary user, create
a throwaway user. On first boot after installation, login as root and use ls -n,
groupadd, useradd, userdel, etc. to delete the throwaway, and manage users as
you wish. Editing /etc/passwd should not be necessary, but is an option
available to the adventurous.


Gee, I had it bracketed :-)
I'll gladly refrain from editing /etc/passwd thank you.

How do you set up LVM so that install does not clobber one of the 
siblings? During installation, for partitioning, do you pick "Manual" or 
"Guided -*" ?


 - Dan



In Jessie, create a new user dan2 or dan09 using the UID you picked for Stretch.
Booted to either, the homedir of the other can optionally be bind mounted as a
subdir of the current dan for more convenient data access without interference
with settings of the other. Either way you boot, you'll be user "dan", but
without risk of settings corruption.
[1] https://en.wikipedia.org/wiki/User_identifier--

"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

  Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-11-13 18:04 (UTC-0500):

> Felix Miata wrote:

>> There's no need for separate home file systems to keep one installation from
>> causing corruption to another's user settings. Simply do not reuse UIDs[1].

>> On your first, your Jessie probably has a user dan with UID 1000. On the 
>> next,
>> don't use 1000 for user dan. Pick any other number, e.g. 1200 for user dan in
>> Stretch. During Stretch installation make the initial user dan8 or dan01 
>> using
>> Jessie's UID (probably 1000) for dan.

> Do you pick a UID during install somehow or do you login as root before 
> anything else and edit /etc/passwd?

I could have been clearer. "During installation" isn't necessarily the correct
time for user management. I don't remember whether Stretch forces ordinary user
creation at installation time, or leaves it up to the admin to do when he 
chooses.

With installers that force creation of an ordinary user in addition to providing
a root password, or that don't permit UID selection for an ordinary user, create
a throwaway user. On first boot after installation, login as root and use ls -n,
groupadd, useradd, userdel, etc. to delete the throwaway, and manage users as
you wish. Editing /etc/passwd should not be necessary, but is an option
available to the adventurous.

>> In Jessie, create a new user dan2 or dan09 using the UID you picked for 
>> Stretch.

>> Booted to either, the homedir of the other can optionally be bind mounted as 
>> a
>> subdir of the current dan for more convenient data access without 
>> interference
>> with settings of the other. Either way you boot, you'll be user "dan", but
>> without risk of settings corruption.

>> [1] https://en.wikipedia.org/wiki/User_identifier-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[David Christensen]

On 11/13/17 15:27, Dan Norton wrote:



On 11/13/2017 02:14 PM, David Christensen wrote:

[...]
Disk space is far cheaper than data loss.  Don't make the mistake of 
trying to save money through partition, LVM, file system, etc., 
gymnastics rather than simply buying another drive and doing it the 
KISS way.




Well I'm much in favor of the KISS principle, it's just that there is 
too much PC equipment lying around here doing nothing. I haul away, 
donate from time to time but it simply accumulates. I hate the thought 
of buying more when there is 900g unused on sda. :-)


So, a 1 TB drive?  Make and model?

Laptop, desktop, server, other?  Make and model?

Any spare drive bays?

Do you have any 16 ~ 128 GB HDD's or SSD's lying around?

David

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 02:14 PM, David Christensen wrote:

[...]
Disk space is far cheaper than data loss.  Don't make the mistake of 
trying to save money through partition, LVM, file system, etc., 
gymnastics rather than simply buying another drive and doing it the 
KISS way.




Well I'm much in favor of the KISS principle, it's just that there is 
too much PC equipment lying around here doing nothing. I haul away, 
donate from time to time but it simply accumulates. I hate the thought 
of buying more when there is 900g unused on sda. :-)


 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 02:08 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-13 10:26 (UTC-0500):


David Christensen wrote:

Dan Norton wrote:

My first Linux install was about one year ago. After some missteps, I
have used Debian 8 in reasonable satisfaction on the desktop during
that year. Now I want to leave 8 in place and do a network install
for Debian 9 on the same disk and switch back and forth at boot time.

I think your best bet is to migrate from Debian 8 to Debian 9:
1.  Image, backup, and/or archive everything.  You will especially
want to get a copy of the /etc tree onto a USB flash drive so you can
see LVM, fstab, etc., configuration settings for mounting the Debian 8
disk under Debian 9.

Please say more about the configuration settings.

All my machines are multiboot. Few have as few as 4 operating system
installations, most have more than 12, and most of those not running RAID use
only one disk.

There's no need for separate home file systems to keep one installation from
causing corruption to another's user settings. Simply do not reuse UIDs[1].

On your first, your Jessie probably has a user dan with UID 1000. On the next,
don't use 1000 for user dan. Pick any other number, e.g. 1200 for user dan in
Stretch. During Stretch installation make the initial user dan8 or dan01 using
Jessie's UID (probably 1000) for dan.


Do you pick a UID during install somehow or do you login as root before 
anything else and edit /etc/passwd?


 - Dan


In Jessie, create a new user dan2 or dan09 using the UID you picked for Stretch.

Booted to either, the homedir of the other can optionally be bind mounted as a
subdir of the current dan for more convenient data access without interference
with settings of the other. Either way you boot, you'll be user "dan", but
without risk of settings corruption.

[1] https://en.wikipedia.org/wiki/User_identifier

Re: Debian 8 and Debian 9 Dual Boot

[Jochen Spieker]

Dan Norton:
> On 11/13/2017 03:53 AM, Jochen Spieker wrote:
>> Dan Norton:
>> 
>>> LVM reports as follows:
>>> 
>>> dan@debian:/$ sudo vgdisplay -C
>>>VG   #PV #LV #SN AttrVSize  VFree
>>>debian-vg   1 5 0wz--n- 976.56g 938.20g
>>
>> You can vgreduce debian-vg easily to make room for your planned
>> debian9-vg.
> 
> So this must mean that VFree above is bound to debian-vg currently. I had
> not recognized the need to vgreduce. Thanks.

Good that you didn't, because I just noticed I was wrong. :) vgreduce is
only for removing physical volumes from a volume group.

It never occured to me before, but I think each physical volume can only
belong to exactly one volume group. That means if you want to use
separate volume groups you would need to pvresize (shrink) the physical
volume. I have never actually done this (only enlarged PVs).

I would probably accept that this is not easily possible with the
current layout and just use one VG. You do not gain much by separate
volume groups anyway.

J.
-- 
If I could travel in time I would show my minidisc to the Romans and
become Caesar until the batteries ran out.
[Agree]   [Disagree]
 


signature.asc
Description: PGP signature

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 02:08 PM, Felix Miata wrote:

Dan Norton composed on 2017-11-13 10:26 (UTC-0500):


David Christensen wrote:

Dan Norton wrote:

My first Linux install was about one year ago. After some missteps, I
have used Debian 8 in reasonable satisfaction on the desktop during
that year. Now I want to leave 8 in place and do a network install
for Debian 9 on the same disk and switch back and forth at boot time.

I think your best bet is to migrate from Debian 8 to Debian 9:
1.  Image, backup, and/or archive everything.  You will especially
want to get a copy of the /etc tree onto a USB flash drive so you can
see LVM, fstab, etc., configuration settings for mounting the Debian 8
disk under Debian 9.

Please say more about the configuration settings.

All my machines are multiboot. Few have as few as 4 operating system
installations, most have more than 12, and most of those not running RAID use
only one disk.


Excellent. You have my undivided attention :-)


There's no need for separate home file systems to keep one installation from
causing corruption to another's user settings. Simply do not reuse UIDs[1].

On your first, your Jessie probably has a user dan with UID 1000. On the next,
don't use 1000 for user dan. Pick any other number, e.g. 1200 for user dan in
Stretch. During Stretch installation make the initial user dan8 or dan01 using
Jessie's UID (probably 1000) for dan.

In Jessie, create a new user dan2 or dan09 using the UID you picked for Stretch.


Interesting.



Booted to either, the homedir of the other can optionally be bind mounted as a
subdir of the current dan for more convenient data access without interference
with settings of the other. Either way you boot, you'll be user "dan", but
without risk of settings corruption.
[1] https://en.wikipedia.org/wiki/User_identifier


I'm going to study this and I may have questions later.

Thanks, Felix

 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[David Christensen]

On 11/13/17 07:26, Dan Norton wrote:


On 11/13/2017 01:10 AM, David Christensen wrote:


1.  Image, backup, and/or archive everything.  You will especially 
want to get a copy of the /etc tree onto a USB flash drive so you can 
see LVM, fstab, etc., configuration settings for mounting the Debian 8 
disk under Debian 9.


Please say more about the configuration settings. I back up /home and 
/var. The git stuff is in /opt and pushed to GitHub. Should the /etc 
tree be backed up also? How would the settings be used from 9?


Every operating system and every software application installed on that 
operating system has configuration settings.  One of the advantages of 
the traditional Unix model is that most OS settings are in plain text 
files that you can change with a text editor.  Similarly so for server 
applications.  (Desktops and their applications are another story, see 
below.)  Good software has manual pages that explain where its 
configuration files reside and the format and meaning of the contents. 
Debian and Debian-ized applications seem to put most of their 
configuration files under /etc.



Whenever I decide to change such files, I make a copy with '-orig' 
appended, put the original file into a version control system (CVS), 
edit-test-repeat, and check-in to CVS when done.  On a typical 
workstation, I might have a dozen files under /etc checked in.  On a 
server, more.



I once tried putting the entire /etc directory into CVS.  Adding and 
removing software made a lot of changes, as expected.  But, so did 
normal operations.  Locating meaningful changes in became an exercise in 
"find the needle in the hay stack".  And, adding/removing lots of files 
and directories in CVS working directories is labor intensive.  I don't 
do that any more.



For migrations, unless you are an expert already or psychic, you will 
want the old drive in it's entirety.  (I keep my system drives under 16 
GB, to encourage me to take images at opportune moments like this.) 
It's useful to also copy /etc onto a flash drive when your old drive is 
more complex than a partition table and normal file systems (so that you 
will have clues about how to mount it, for starters).



As for how the settings are used, each OS instance and it applications 
determine that.  Migrating settings involves editing configuration files 
and/or overwriting configuration files.  You have to go through each 
subsystem one at a time and see what configuration files it wants, 
format, and content, and then do what it takes to make it happy.



As others have replied, desktop and desktop application configuration 
settings and data in home directories are not agreeable to migration.  I 
don't even try.  I backup Thunderbird mail and export address books on 
the old system, and restore/ import on the new system.  I use the sync 
service built into Firefox to keep it current on all my machines where 
it is installed.  I check in my CVS working directories on the old 
system and check them out on the new system.  My bulk data is on the 
file server.  I manually configure whatever is left -- desktop, Open 
Office preferences, etc..




2.  Put your bulk data onto a separate drive or a file server/ NAS.

3.  If your computer has a spare drive bay, buy a new drive, unplug 
your old drive, install the new drive, install Debian 9 onto the new 
drive, reconnect the old drive and mount read-only under Debian 9, and 
migrate your settings and remaining data from the old drive to the new 
drive. Your BIOS/UEFI should allow you to boot from either drive 
during this process.  So long as you don't damage the Debian 8 disk, 
you can always fall back to Debian 8 if the migration goes badly.


4.  If you computer does not have a spare drive bay, buy a new drive 
and a USB external drive enclosure, put your old drive in the 
enclosure, install the new drive, install Debian 9 onto the new drive, 
plug in the old drive and mount read-only under Debian 9, and migrate 
your settings and remaining data from the old drive to the new drive. 
Again, your BIOS/UEFI should allow you to boot from either drive and 
you can fall back to Debian 8 if necessary.


Sobering advice - good! Makes me re-think some things. A bit inefficient 
WRT disk space.


Disk space is far cheaper than data loss.  Don't make the mistake of 
trying to save money through partition, LVM, file system, etc., 
gymnastics rather than simply buying another drive and doing it the KISS 
way.




You did say "migrate" and not "upgrade" as I first read. Thanks David.


YW.  :-)


David

Re: Debian 8 and Debian 9 Dual Boot

[Felix Miata]

Dan Norton composed on 2017-11-13 10:26 (UTC-0500):

> David Christensen wrote:

>> Dan Norton wrote:

>>> My first Linux install was about one year ago. After some missteps, I 
>>> have used Debian 8 in reasonable satisfaction on the desktop during 
>>> that year. Now I want to leave 8 in place and do a network install 
>>> for Debian 9 on the same disk and switch back and forth at boot time.

>> I think your best bet is to migrate from Debian 8 to Debian 9:

>> 1.  Image, backup, and/or archive everything.  You will especially 
>> want to get a copy of the /etc tree onto a USB flash drive so you can 
>> see LVM, fstab, etc., configuration settings for mounting the Debian 8 
>> disk under Debian 9.

> Please say more about the configuration settings.

All my machines are multiboot. Few have as few as 4 operating system
installations, most have more than 12, and most of those not running RAID use
only one disk.

There's no need for separate home file systems to keep one installation from
causing corruption to another's user settings. Simply do not reuse UIDs[1].

On your first, your Jessie probably has a user dan with UID 1000. On the next,
don't use 1000 for user dan. Pick any other number, e.g. 1200 for user dan in
Stretch. During Stretch installation make the initial user dan8 or dan01 using
Jessie's UID (probably 1000) for dan.

In Jessie, create a new user dan2 or dan09 using the UID you picked for Stretch.

Booted to either, the homedir of the other can optionally be bind mounted as a
subdir of the current dan for more convenient data access without interference
with settings of the other. Either way you boot, you'll be user "dan", but
without risk of settings corruption.

[1] https://en.wikipedia.org/wiki/User_identifier
-- 
"Wisdom is supreme; therefore get wisdom. Whatever else you
get, get wisdom." Proverbs 4:7 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: Debian 8 and Debian 9 Dual Boot

[Joe]

On Mon, 13 Nov 2017 11:01:27 -0500
Dan Norton  wrote:


> Although I didn't say so, each install would have its own set of 
> directories. Please say more about how to mount the other
> installation and share data. How to mount things in another volume
> group?
> 

Good advice so far, but to add a bit: all LVM groups will be seen at
boot, and /dev will know about them. See man lvm2 and also here:

https://wiki.debian.org/LVM  for complete information about the
commands you have available. There are also numerous tutorials on the
Net which show basic usage of the simpler commands. It's worth having a
look when you have some spare time, as one day you'll need to know some
of this and won't have any spare time.

Look in /etc/fstab for lines beginning /dev/mapper/[volume] which will
be the volumes mounted in the running installation. The 'mapper' is
turning LVM volumes into things which look like partitions for many
purposes.

To mount the volumes in the non-running installation, you need to add
similar-looking lines to fstab, and don't forget to create directories
somewhere quiet, like /mnt, to mount them on. You will need to add an
option to the entries that the running mounts do not have: if you
don't, systemd will try to mount them on boot, and if it fails, will
abort the boot. No problem if the extra mounts are on a drive that is
always there, but if stored on a second drive which is later removed,
it will kill the machine.

So you want either the option noauto, which tells systemd to ignore the
volume and you will always mount it manually (just mount 
in a terminal will do that, the fstab entry will have all the options
needed), or x-systemd.automount which will also be ignored during boot
but will be automounted when you try to access it. I mount my network
shares with that option, for the same reason (losing one will only put
a huge timeout into boot, but won't actually abort it) and also so that
mounting them will not slow down booting.

Here are a couple of spare LVM volumes on my workstation, one of which I
use only occasionally and always mount manually. You might notice that
the automount point is called oldroot, guess what that is...

/dev/mapper/newerlvm-spare  /mnt/spare  ext4  rw,user,noauto  0  0
/dev/mapper/newlvm-root  /mnt/oldroot  ext4  rw,user,x-systemd.automount
0 0

If you do mount the non-running system root, which is convenient for
many purposes, do be especially careful where you are in the directory
tree. The directories will look very similar, and it is easy to
misconfigure the wrong system. Years ago, I copied a system to a second
drive, then went into it to adjust the /etc/fstab to the new values it
needed. I managed to edit the wrong fstab, and neither system would
boot... fortunately, that was an easy one to fix.

So yes, it is safer only to mount the data of the other installation,
at least once you have the new one working properly and have duplicated
what you want from the old one.

-- 
Joe

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 11:07 AM, Roberto C. Sánchez wrote:

On Mon, Nov 13, 2017 at 11:01:27AM -0500, Dan Norton wrote:

Although I didn't say so, each install would have its own set of
directories. Please say more about how to mount the other installation and
share data. How to mount things in another volume group?


That would be as simple as creating a logical volume, call it "data,"
and then configuring it to mount in each installation with a line in
/etc/fstab:

/dev/vg00/data /data .
Ah. OK. If each installation is in its own volume group, then vg00 above 
would be replaced with the actual vg.


Since you are talking about installing more than one version of Debian
and switching back and forth between them I assume that you are the only
user of the machine.

Yes.

You would need to make sure that the assigned user
ID for your username matches in both installations and then just make
sure that /data has the appropriate permissions for your user account to
be able to write to it.

OK.


This is generally preferrable to directly mounting directories of the
other installation just to share data, as the dedicated data sharing
directory approach reduces the likelihood that a mistake will cause
problems in the other installation.

Great. If there's a mistake to be made ...

Thanks, Roberto.

 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[Roberto C . Sánchez]

On Mon, Nov 13, 2017 at 11:01:27AM -0500, Dan Norton wrote:
> 
> Although I didn't say so, each install would have its own set of
> directories. Please say more about how to mount the other installation and
> share data. How to mount things in another volume group?
> 
That would be as simple as creating a logical volume, call it "data,"
and then configuring it to mount in each installation with a line in
/etc/fstab:

/dev/vg00/data /data .

Since you are talking about installing more than one version of Debian
and switching back and forth between them I assume that you are the only
user of the machine.  You would need to make sure that the assigned user
ID for your username matches in both installations and then just make
sure that /data has the appropriate permissions for your user account to
be able to write to it.

This is generally preferrable to directly mounting directories of the
other installation just to share data, as the dedicated data sharing
directory approach reduces the likelihood that a mistake will cause
problems in the other installation.

Regards,

-Roberto

-- 
Roberto C. Sánchez

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 06:56 AM, Roberto C. Sánchez wrote:

On Sun, Nov 12, 2017 at 10:27:36PM -0500, Dan Norton wrote:

My first Linux install was about one year ago. After some missteps, I have
used Debian 8 in reasonable satisfaction on the desktop during that year.
Now I want to leave 8 in place and do a network install for Debian 9 on the
same disk and switch back and forth at boot time.


If you use a desktop environment (e.g., GNOME or KDE) and want to share
the /home directory between both installations, you might encounter some
problems.  In particular, the desktop environments are known for not
behaving well when a newer version updates all the user-specific files
and then an older version tries to access them.


Xfce is what I use and it would surely encounter the same kinds of problems.



For example, the first time you boot to Debian 9 and log in (assume
GNOME for moment), files under ~/.config, ~/.gnome*, and possibly other
places will be modified by the new version of GNOME in Debian 9.  There
is no guarantee that when you go back to Debian 8 that the version of
GNOME there will cope will at all with the changed files under ~/.config
and ~/.gnome*.  In fact, it is quite likely that it will encounter some
sort of significant problem.

Your safest bet if you really intend to dual-boot in this way is to have
separate /home directories for each install and then use something like
/data to mount into both installations in order to share documents and
data between sessions.


Although I didn't say so, each install would have its own set of 
directories. Please say more about how to mount the other installation 
and share data. How to mount things in another volume group?


Thanks, Roberto.

 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 03:53 AM, Jochen Spieker wrote:

Dan Norton:
henny|i

My first Linux install was about one year ago. After some missteps, I have
used Debian 8 in reasonable satisfaction on the desktop during that year.
Now I want to leave 8 in place and do a network install for Debian 9 on the
same disk and switch back and forth at boot time.

Your disk setup allows to do this comparably easy, but I would think
twice whether that is actually what I want to do. You end up with two
systems which you have to maintain and, most importantly, you cannot
properly share your /home. Most programs that save their settings to
$HOME will automatically upgrade their configuration files on first
start with a new version and after that you have to assume that the
older version cannot read it anymore. I realize that this is part of
your disk space calculation but I want to stress that the result is
probably not something that you will want to use for an extended period
of time.


Indeed, I had thought to have 9 with a complete set of partitions as if 
8 didn't exist, but I did not say so.





LVM reports as follows:

dan@debian:/$ sudo vgdisplay -C
   VG   #PV #LV #SN AttrVSize  VFree
   debian-vg   1 5 0wz--n- 976.56g 938.20g

You can vgreduce debian-vg easily to make room for your planned
debian9-vg.


So this must mean that VFree above is bound to debian-vg currently. I 
had not recognized the need to vgreduce. Thanks.





So there is plenty of disk space for the two Debians and more besides. The
question is how to prepare to install 9? My guess is to define another
volume group called debian9-vg perhaps but how will this be recognized
during network install?

Yes, it will. It should also be possible to vgreduce debian.vg and
vgcreate debian9-vg inside the installer, but I would be more
comfortable doing that from the running system, using the regular tools.


For sure. Trying to do things within the installer has been the source 
of a lot of my problems.




During installation, you just have to make sure that all LVs in
debian-vg are marked as "Do not use".


Good. This is the kind of information I need.




I've clobbered stuff before during installs and I'm
gun shy. Maybe there is a better way. Any thoughts on this will be
appreciated.

It happens to everybody. The only things that help are a good plan and
concentration on the task at hand. :)


+1 and thank you, Jochen.

 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[Dan Norton]

On 11/13/2017 01:10 AM, David Christensen wrote:

On 11/12/17 19:27, Dan Norton wrote:
My first Linux install was about one year ago. After some missteps, I 
have used Debian 8 in reasonable satisfaction on the desktop during 
that year. Now I want to leave 8 in place and do a network install 
for Debian 9 on the same disk and switch back and forth at boot time.


I think your best bet is to migrate from Debian 8 to Debian 9:

1.  Image, backup, and/or archive everything.  You will especially 
want to get a copy of the /etc tree onto a USB flash drive so you can 
see LVM, fstab, etc., configuration settings for mounting the Debian 8 
disk under Debian 9.


Please say more about the configuration settings. I back up /home and 
/var. The git stuff is in /opt and pushed to GitHub. Should the /etc 
tree be backed up also? How would the settings be used from 9?




2.  Put your bulk data onto a separate drive or a file server/ NAS.

3.  If your computer has a spare drive bay, buy a new drive, unplug 
your old drive, install the new drive, install Debian 9 onto the new 
drive, reconnect the old drive and mount read-only under Debian 9, and 
migrate your settings and remaining data from the old drive to the new 
drive. Your BIOS/UEFI should allow you to boot from either drive 
during this process.  So long as you don't damage the Debian 8 disk, 
you can always fall back to Debian 8 if the migration goes badly.


4.  If you computer does not have a spare drive bay, buy a new drive 
and a USB external drive enclosure, put your old drive in the 
enclosure, install the new drive, install Debian 9 onto the new drive, 
plug in the old drive and mount read-only under Debian 9, and migrate 
your settings and remaining data from the old drive to the new drive.  
Again, your BIOS/UEFI should allow you to boot from either drive and 
you can fall back to Debian 8 if necessary.


Sobering advice - good! Makes me re-think some things. A bit inefficient 
WRT disk space.

You did say "migrate" and not "upgrade" as I first read. Thanks David.

 - Dan

Re: Debian 8 and Debian 9 Dual Boot

[Roberto C . Sánchez]

On Sun, Nov 12, 2017 at 10:27:36PM -0500, Dan Norton wrote:
> My first Linux install was about one year ago. After some missteps, I have
> used Debian 8 in reasonable satisfaction on the desktop during that year.
> Now I want to leave 8 in place and do a network install for Debian 9 on the
> same disk and switch back and forth at boot time.
> 
If you use a desktop environment (e.g., GNOME or KDE) and want to share
the /home directory between both installations, you might encounter some
problems.  In particular, the desktop environments are known for not
behaving well when a newer version updates all the user-specific files
and then an older version tries to access them.

For example, the first time you boot to Debian 9 and log in (assume
GNOME for moment), files under ~/.config, ~/.gnome*, and possibly other
places will be modified by the new version of GNOME in Debian 9.  There
is no guarantee that when you go back to Debian 8 that the version of
GNOME there will cope will at all with the changed files under ~/.config
and ~/.gnome*.  In fact, it is quite likely that it will encounter some
sort of significant problem.

Your safest bet if you really intend to dual-boot in this way is to have
separate /home directories for each install and then use something like
/data to mount into both installations in order to share documents and
data between sessions.

Regards,

-Roberto

-- 
Roberto C. Sánchez

Re: Debian 8 and Debian 9 Dual Boot

[Jochen Spieker]

Dan Norton:
henny|i
> My first Linux install was about one year ago. After some missteps, I have
> used Debian 8 in reasonable satisfaction on the desktop during that year.
> Now I want to leave 8 in place and do a network install for Debian 9 on the
> same disk and switch back and forth at boot time.

Your disk setup allows to do this comparably easy, but I would think
twice whether that is actually what I want to do. You end up with two
systems which you have to maintain and, most importantly, you cannot
properly share your /home. Most programs that save their settings to
$HOME will automatically upgrade their configuration files on first
start with a new version and after that you have to assume that the
older version cannot read it anymore. I realize that this is part of
your disk space calculation but I want to stress that the result is
probably not something that you will want to use for an extended period
of time.

> LVM reports as follows:
> 
> dan@debian:/$ sudo vgdisplay -C
>   VG   #PV #LV #SN AttrVSize  VFree
>   debian-vg   1 5 0wz--n- 976.56g 938.20g

You can vgreduce debian-vg easily to make room for your planned
debian9-vg.

> So there is plenty of disk space for the two Debians and more besides. The
> question is how to prepare to install 9? My guess is to define another
> volume group called debian9-vg perhaps but how will this be recognized
> during network install?

Yes, it will. It should also be possible to vgreduce debian.vg and
vgcreate debian9-vg inside the installer, but I would be more
comfortable doing that from the running system, using the regular tools.

During installation, you just have to make sure that all LVs in
debian-vg are marked as "Do not use".

> I've clobbered stuff before during installs and I'm
> gun shy. Maybe there is a better way. Any thoughts on this will be
> appreciated.

It happens to everybody. The only things that help are a good plan and
concentration on the task at hand. :)

J.
-- 
I like my Toyota RAV4 because of the commanding view of the traffic
jams.
[Agree]   [Disagree]
 


signature.asc
Description: PGP signature

Re: Debian 8 and Debian 9 Dual Boot

[David Christensen]

On 11/12/17 19:27, Dan Norton wrote:
My first Linux install was about one year ago. After some missteps, I 
have used Debian 8 in reasonable satisfaction on the desktop during that 
year. Now I want to leave 8 in place and do a network install for Debian 
9 on the same disk and switch back and forth at boot time.


I think your best bet is to migrate from Debian 8 to Debian 9:

1.  Image, backup, and/or archive everything.  You will especially want 
to get a copy of the /etc tree onto a USB flash drive so you can see 
LVM, fstab, etc., configuration settings for mounting the Debian 8 disk 
under Debian 9.


2.  Put your bulk data onto a separate drive or a file server/ NAS.

3.  If your computer has a spare drive bay, buy a new drive, unplug your 
old drive, install the new drive, install Debian 9 onto the new drive, 
reconnect the old drive and mount read-only under Debian 9, and migrate 
your settings and remaining data from the old drive to the new drive. 
Your BIOS/UEFI should allow you to boot from either drive during this 
process.  So long as you don't damage the Debian 8 disk, you can always 
fall back to Debian 8 if the migration goes badly.


4.  If you computer does not have a spare drive bay, buy a new drive and 
a USB external drive enclosure, put your old drive in the enclosure, 
install the new drive, install Debian 9 onto the new drive, plug in the 
old drive and mount read-only under Debian 9, and migrate your settings 
and remaining data from the old drive to the new drive.  Again, your 
BIOS/UEFI should allow you to boot from either drive and you can fall 
back to Debian 8 if necessary.



David