[gentoo-user] duplicate gentoo system - errors

[gentoo-user] duplicate gentoo system - errors

[thelma]

OK, I used Gparted (Bootable usb) to copy partition from:
Western Digital driver, usually:
/dev/sda1 etc

to M.2 SSD
/dev/ nvme0n1p1 etc

I can boot M.2 drive, but the x-server doesn't work (even though I use
same graphical card).  Network is not working (easy fix, new driver
needs to be compiled IN) but there is a bigger problem.

Duplicating was easy, but when I try to recompile a kernel I get an error:

make menuconfig
HOSTCC script/kconfig/mconf.o
<build-in> : internal compiler error: Illegal instruction

Even if I try to run: emerge --info  I get:
Illegal instruction

In addition my fstab doesn't look correct (but it works)
/dev/sda1    /boot    ext2

It should be something like:
/dev/nvme0n1p1   /boot   ext2

-- 
Thelma

Re: [gentoo-user] duplicate gentoo system - errors

[Michael]

[-- Attachment #1: Type: text/plain, Size: 1393 bytes --]

On Sunday, 22 November 2020 23:39:44 GMT thelma@sys-concept.com wrote:
> OK, I used Gparted (Bootable usb) to copy partition from:
> Western Digital driver, usually:
> /dev/sda1 etc
> 
> to M.2 SSD
> /dev/ nvme0n1p1 etc
> 
> I can boot M.2 drive, but the x-server doesn't work (even though I use
> same graphical card).  Network is not working (easy fix, new driver
> needs to be compiled IN) but there is a bigger problem.

If you have cloned each partition from the old to the new disk, then the new 
disk should work exactly as the old disk does.  I mean, it should have the 
same kernel, the same / filesystem, the same modules, etc.

Since the new disk is an nvme drive, you will need additional drivers - should 
these not be available in the old kernel.


> Duplicating was easy, but when I try to recompile a kernel I get an error:
> 
> make menuconfig
> HOSTCC script/kconfig/mconf.o
> <build-in> : internal compiler error: Illegal instruction
> 
> Even if I try to run: emerge --info  I get:
> Illegal instruction
> 
> In addition my fstab doesn't look correct (but it works)
> /dev/sda1    /boot    ext2
> 
> It should be something like:
> /dev/nvme0n1p1   /boot   ext2

Do you have both disks connected to the MoBo when you're trying to boot from 
the new disk?

Have you changed the UUIDs on the new partitions?

Have you installed the boot manager on the new disk (if using MBR)?

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[thelma]

On 11/22/2020 05:25 PM, Michael wrote:
> On Sunday, 22 November 2020 23:39:44 GMT thelma@sys-concept.com wrote:
>> OK, I used Gparted (Bootable usb) to copy partition from:
>> Western Digital driver, usually:
>> /dev/sda1 etc
>>
>> to M.2 SSD
>> /dev/ nvme0n1p1 etc
>>
>> I can boot M.2 drive, but the x-server doesn't work (even though I use
>> same graphical card).  Network is not working (easy fix, new driver
>> needs to be compiled IN) but there is a bigger problem.
> 
> If you have cloned each partition from the old to the new disk, then the new 
> disk should work exactly as the old disk does.  I mean, it should have the 
> same kernel, the same / filesystem, the same modules, etc.
> 
> Since the new disk is an nvme drive, you will need additional drivers - should 
> these not be available in the old kernel.
> 
> 
>> Duplicating was easy, but when I try to recompile a kernel I get an error:
>>
>> make menuconfig
>> HOSTCC script/kconfig/mconf.o
>> <build-in> : internal compiler error: Illegal instruction
>>
>> Even if I try to run: emerge --info  I get:
>> Illegal instruction
>>
>> In addition my fstab doesn't look correct (but it works)
>> /dev/sda1    /boot    ext2
>>
>> It should be something like:
>> /dev/nvme0n1p1   /boot   ext2
> 
> Do you have both disks connected to the MoBo when you're trying to boot from 
> the new disk?

Yes, they are both connected

> Have you changed the UUIDs on the new partitions?

Never used UUID in fstab. Do I just run: blkid|grep UUID
and copy it to fstab.

> Have you installed the boot manager on the new disk (if using MBR)?

I just copied the whole MBR to a new disk and it worked, the system
boots, but nothing can be compiled.

Re: [gentoo-user] duplicate gentoo system - errors

[Jack]

On 2020.11.22 20:09, thelma@sys-concept.com wrote:
> On 11/22/2020 05:25 PM, Michael wrote:
> > On Sunday, 22 November 2020 23:39:44 GMT thelma@sys-concept.com  
> wrote:
> >> OK, I used Gparted (Bootable usb) to copy partition from:
> >> Western Digital driver, usually:
> >> /dev/sda1 etc
> >>
> >> to M.2 SSD
> >> /dev/ nvme0n1p1 etc
> >>
> >> I can boot M.2 drive, but the x-server doesn't work (even though I  
> use
> >> same graphical card).  Network is not working (easy fix, new driver
> >> needs to be compiled IN) but there is a bigger problem.
> >
> > If you have cloned each partition from the old to the new disk,  
> then the new
> > disk should work exactly as the old disk does.  I mean, it should  
> have the
> > same kernel, the same / filesystem, the same modules, etc.
> >
> > Since the new disk is an nvme drive, you will need additional  
> drivers - should
> > these not be available in the old kernel.
> >
> >
> >> Duplicating was easy, but when I try to recompile a kernel I get  
> an error:
> >>
> >> make menuconfig
> >> HOSTCC script/kconfig/mconf.o
> >> <build-in> : internal compiler error: Illegal instruction
> >>
> >> Even if I try to run: emerge --info  I get:
> >> Illegal instruction
> >>
> >> In addition my fstab doesn't look correct (but it works)
> >> /dev/sda1    /boot    ext2
> >>
> >> It should be something like:
> >> /dev/nvme0n1p1   /boot   ext2
> >
> > Do you have both disks connected to the MoBo when you're trying to  
> boot from
> > the new disk?
> 
> Yes, they are both connected
> 
> > Have you changed the UUIDs on the new partitions?
> 
> Never used UUID in fstab. Do I just run: blkid|grep UUID
> and copy it to fstab.
> 
> > Have you installed the boot manager on the new disk (if using MBR)?
> 
> I just copied the whole MBR to a new disk and it worked, the system
> boots, but nothing can be compiled.
I would confirm that you are really booted from the new disk and not  
the old one.  It is possible that the MBR from the new disk was used to  
boot, but if /etc/fstab says /boot is mounted from /dev/sda1 then that  
does seem wrong.  I almost always put an empty file in the root of each  
partition named for the disk/partition just so I can be sure what's  
actually mounted.  Is /etc/fstab identical on both disks?  What does  
fstab say about where / is mounted from?

Re: [gentoo-user] duplicate gentoo system - errors

[thelma]

On 11/22/2020 06:16 PM, Jack wrote:
> On 2020.11.22 20:09, thelma@sys-concept.com wrote:
>> On 11/22/2020 05:25 PM, Michael wrote:
>> > On Sunday, 22 November 2020 23:39:44 GMT thelma@sys-concept.com wrote:
>> >> OK, I used Gparted (Bootable usb) to copy partition from:
>> >> Western Digital driver, usually:
>> >> /dev/sda1 etc
>> >>
>> >> to M.2 SSD
>> >> /dev/ nvme0n1p1 etc
>> >>
>> >> I can boot M.2 drive, but the x-server doesn't work (even though I use
>> >> same graphical card).  Network is not working (easy fix, new driver
>> >> needs to be compiled IN) but there is a bigger problem.
>> >
>> > If you have cloned each partition from the old to the new disk, then
>> the new
>> > disk should work exactly as the old disk does.  I mean, it should
>> have the
>> > same kernel, the same / filesystem, the same modules, etc.
>> >
>> > Since the new disk is an nvme drive, you will need additional
>> drivers - should
>> > these not be available in the old kernel.
>> >
>> >
>> >> Duplicating was easy, but when I try to recompile a kernel I get an
>> error:
>> >>
>> >> make menuconfig
>> >> HOSTCC script/kconfig/mconf.o
>> >> <build-in> : internal compiler error: Illegal instruction
>> >>
>> >> Even if I try to run: emerge --info  I get:
>> >> Illegal instruction
>> >>
>> >> In addition my fstab doesn't look correct (but it works)
>> >> /dev/sda1    /boot    ext2
>> >>
>> >> It should be something like:
>> >> /dev/nvme0n1p1   /boot   ext2
>> >
>> > Do you have both disks connected to the MoBo when you're trying to
>> boot from
>> > the new disk?
>>
>> Yes, they are both connected
>>
>> > Have you changed the UUIDs on the new partitions?
>>
>> Never used UUID in fstab. Do I just run: blkid|grep UUID
>> and copy it to fstab.
>>
>> > Have you installed the boot manager on the new disk (if using MBR)?
>>
>> I just copied the whole MBR to a new disk and it worked, the system
>> boots, but nothing can be compiled.
> I would confirm that you are really booted from the new disk and not the
> old one.  It is possible that the MBR from the new disk was used to
> boot, but if /etc/fstab says /boot is mounted from /dev/sda1 then that
> does seem wrong.  I almost always put an empty file in the root of each
> partition named for the disk/partition just so I can be sure what's
> actually mounted.  Is /etc/fstab identical on both disks?  What does
> fstab say about where / is mounted from?

You are absolutely correct.  I was booting the whole time the Western
Digital (old drive). :-/  My mistake, once I removed the WD drive the
new M.2 SSD doesn't even boot.
I think the easiest way would be to re-install the Getnoo from scratch
and dig out the old programs I need from "attic".  Mixing/moving SSD
(sda) and M2.2 (nvme0n1) and transferring partitions might not be as easy.

Re: [gentoo-user] duplicate gentoo system - errors

[Rich Freeman]

On Sun, Nov 22, 2020 at 6:39 PM <thelma@sys-concept.com> wrote:
>
> make menuconfig
> HOSTCC script/kconfig/mconf.o
> <build-in> : internal compiler error: Illegal instruction
>
> Even if I try to run: emerge --info  I get:
> Illegal instruction
>

Is this running on the same CPU, or are you migrating to a different
system?  If the CPU changes then you need to make sure that everything
on your system was built with a -march compatible with your new
system.

When you plan on migrating between systems it is a good idea to set
-march very conservatively, such as -march=x86-64.  It is safe to set
-mtune to whatever you want, but -march produces code that only runs
on that particular CPU.  Of course if the new CPU supports EVERY
instruction that the previous CPU has then you're fine, but you'd be
surprised how many oddball instructions aren't on various CPUs.

-- 
Rich

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1263 bytes --]

On Sun, 22 Nov 2020 18:27:53 -0700, thelma@sys-concept.com wrote:

> > I would confirm that you are really booted from the new disk and not
> > the old one.  It is possible that the MBR from the new disk was used
> > to boot, but if /etc/fstab says /boot is mounted from /dev/sda1 then
> > that does seem wrong.  I almost always put an empty file in the root
> > of each partition named for the disk/partition just so I can be sure
> > what's actually mounted.  Is /etc/fstab identical on both disks?
> > What does fstab say about where / is mounted from?  
> 
> You are absolutely correct.  I was booting the whole time the Western
> Digital (old drive). :-/  My mistake, once I removed the WD drive the
> new M.2 SSD doesn't even boot.

Check the settings in your BIOS/firmware to make sure it detects the
drive and is set to boot from it.

> I think the easiest way would be to re-install the Getnoo from scratch

And if it still doesn't work because of a firmware issue, you still have
a non-booting system but with no OS installed either. It's better to try
and diagnose the problem rather than throwing everything away in the hope
that the problem goes with it.


-- 
Neil Bothwick

Contentsoftaglinemaysettleduringshipping.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Michael]

[-- Attachment #1: Type: text/plain, Size: 3064 bytes --]

On Monday, 23 November 2020 01:09:16 GMT thelma@sys-concept.com wrote:
> On 11/22/2020 05:25 PM, Michael wrote:

> > Do you have both disks connected to the MoBo when you're trying to boot
> > from the new disk?
> 
> Yes, they are both connected

In this case the /dev/sda* you see could well be on the old disk.


> > Have you changed the UUIDs on the new partitions?
> 
> Never used UUID in fstab. Do I just run: blkid|grep UUID
> and copy it to fstab.

I warned you about UUIDs.  The block device of /dev/sda* could be pointing at 
a partition either on the old, or the new disk.  In such cases it is a good 
idea to find out what block devices the MoBo identifies and what the kernel.

Go into your BIOS GUI and check if your MoBo sees both disks.

You could select which disk to boot from in the BIOS menu.  Before you change 
this, boot the OS and let's look at what block devices the kernel identifies:

ls -la /dev/disk/by-*

If in the various symlinks listed you will find what disk(s) and partitions 
your kernel identifies.

Alternative commands are blkid, lsblk, udevadm info, hwinfo --block, et al.

As long as the kernel has the appropriate modules and you're not missing any 
necessary firmware, you should find what the nvme disk is identified as and 
corresponding PARTUUIDs and filesystem UUIDs.

When you clone a disk, partition tables, PARTUUIDs, and fs UUIDs, will be 
copied over.  Therefore the kernel will not be able to differentiate reliably 
which is which, old or new.

In addition, sometimes devices are not initialized in the same order.  So when 
probed by the kernel what was identified as /dev/sda last time could be /dev/
sdb now, further adding to confusion when the fs data content is identical.

Gparted has an option to change a UUID to a new random number - I suggest you 
use it.

https://gparted.org/display-doc.php?name=help-manual#gparted-changing-partition-uuid

On the CLI you can use instead uuidgen and then tune2fs.

Once you've done this, it would make sense to mount the / partition on the new 
disk and adjust its fstab accordingly.

In addition you will need to update GRUB so it picks up the new disk /
partition.


> > Have you installed the boot manager on the new disk (if using MBR)?
> 
> I just copied the whole MBR to a new disk and it worked, the system
> boots, but nothing can be compiled.

How did you copy the MBR?  Using dd command?  Unless you cloned a whole disk 
the MBR sector content won't have been copied over.  If the MBR sector has not 
been copied over the new disk will not be able to boot on its own, but you 
could still boot it from the old disk, as long as GRUB on the old disk has 
been updated to pick up the new disk's / partition.  I don't think you 
mentioned it, but in any case I assume you've been using a legacy BIOS with a 
conventional MBR boot loader and GRUB as a boot manager.

As Neil suggested, finding out what's wrong with your current setup and fixing 
it should prove more fruitful than unnecessarily wasting time reinstalling the 
whole OS.  Ask if you get stuck.

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Walter Dnes]

On Sun, Nov 22, 2020 at 04:39:44PM -0700, thelma@sys-concept.com wrote
> 
> Duplicating was easy, but when I try to recompile a kernel I get an error:
> 
> make menuconfig
> HOSTCC script/kconfig/mconf.o
> <build-in> : internal compiler error: Illegal instruction
> 
> Even if I try to run: emerge --info  I get:
> Illegal instruction

  Ouch!  Are the CPUs exactly identical?  If not, then you may get the
"Illegal instruction" error.  This is a "feature" of Gentoo, which is
often user-optimized for a specific CPU.  What's the CPU on the source
machine and what's the CPU on the target machine?  If in doubt,
execute...

gcc -c -Q -march=native --help=target | grep march=

...on each machine.  There are 3 sub-families of "Intel" machines.

1) bog-standard Intel code, e.g. core2, also runs on AMD and Intel Atom
   CPUs at least as new or newer.

2) Intel Atom code runs only on same or newer Intel Atom, because it has
   the MOVBE instruction set the others lack.

3) AMD code runs only on same or newer AMD, because it has the 3DNow!
   instruction set the others lack.

  If you have such a mismatch, your only option may be to rebuild from
scratch.

-- 
Walter Dnes <waltdnes@waltdnes.org>
I don't run "desktop environments"; I run useful applications

Re: [gentoo-user] duplicate gentoo system - errors

[Rich Freeman]

On Mon, Nov 23, 2020 at 9:10 AM Walter Dnes <waltdnes@waltdnes.org> wrote:
>
>
>   Ouch!  Are the CPUs exactly identical?  If not, then you may get the
> "Illegal instruction" error.  This is a "feature" of Gentoo, which is
> often user-optimized for a specific CPU.

This "feature" has nothing to do with Gentoo, but just with the fact
that users can set their CFLAGS to whatever you wish.  If you set a
conservative -march and then set a more aggressive -mtune then your
binaries will run anywhere.  That is how most distros do it.

The default CFLAGS in make.conf are "-O2 -pipe".  If you stick to
those defaults your binaries will run anywhere.

There is nothing wrong with using -march, but you definitely need to
keep in mind that if you ever change CPUs then you're going to need to
rebuild everything.

You could just rebuild @system with generic CFLAGS and then rebuild
everything with -march=native or whatever you prefer once your new
system is running.  Your applications wouldn't run that way, but you
could still boot the OS and the toolchain will work.  If gcc is broken
then you can't rebuild in place.

-- 
Rich

Re: [gentoo-user] duplicate gentoo system - errors

[antlists]

On 23/11/2020 10:37, Michael wrote:
>>> Have you changed the UUIDs on the new partitions?

>> Never used UUID in fstab. Do I just run: blkid|grep UUID
>> and copy it to fstab.

> I warned you about UUIDs.  The block device of /dev/sda* could be pointing at
> a partition either on the old, or the new disk.  In such cases it is a good
> idea to find out what block devices the MoBo identifies and what the kernel.

If you're messing about with disks, partitions, etc, you NEED to have a 
basic understanding of UUIDs. Bear in mind that the UID bit (iirc) 
stands for *UNIQUE* ID.

Linux (as has been said) allocates sda, sdb etc in the order it finds 
partitions, which can be somewhat random. So there is no guarantee, if 
you specify root as sda, something could glitch (or you've stuck an 
eSATA drive on, or or or) and suddenly it's sdb and your system can't 
find root!

So what the initial boot system does is it sets up /dev/.../UUID as 
symlinks to the appropriate sdx. So when boot says "root = UUID", it 
looks at the symlink to find out whether it's sda, sdb, sdc or whatever.

Now if you use dd to copy the old disk to the new, and leave both disks 
connected, anything looking by sda or sdb is going to find it but you 
won't know which disk it's found. At BEST the same applies to anything 
looking by UUID. But it could be a lot worse - anything that relies on 
the UUID being unique (which is what is promised) could do real damage 
to the system if they aren't.

Read up on gdisk - I guess parted, fdisk, etc have the same - but it has 
the option to copy an MBR or GPT and generate new UUIDs for all the 
partitions. You MUST do that if you're leaving both drives in the system.

Cheers,
Wol

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

antlists wrote:
> On 23/11/2020 10:37, Michael wrote:
>>>> Have you changed the UUIDs on the new partitions?
>
>>> Never used UUID in fstab. Do I just run: blkid|grep UUID
>>> and copy it to fstab.
>
>> I warned you about UUIDs.  The block device of /dev/sda* could be
>> pointing at
>> a partition either on the old, or the new disk.  In such cases it is
>> a good
>> idea to find out what block devices the MoBo identifies and what the
>> kernel.
>
> If you're messing about with disks, partitions, etc, you NEED to have
> a basic understanding of UUIDs. Bear in mind that the UID bit (iirc)
> stands for *UNIQUE* ID.
>
> Linux (as has been said) allocates sda, sdb etc in the order it finds
> partitions, which can be somewhat random. So there is no guarantee, if
> you specify root as sda, something could glitch (or you've stuck an
> eSATA drive on, or or or) and suddenly it's sdb and your system can't
> find root!
>
> So what the initial boot system does is it sets up /dev/.../UUID as
> symlinks to the appropriate sdx. So when boot says "root = UUID", it
> looks at the symlink to find out whether it's sda, sdb, sdc or whatever.
>
> Now if you use dd to copy the old disk to the new, and leave both
> disks connected, anything looking by sda or sdb is going to find it
> but you won't know which disk it's found. At BEST the same applies to
> anything looking by UUID. But it could be a lot worse - anything that
> relies on the UUID being unique (which is what is promised) could do
> real damage to the system if they aren't.
>
> Read up on gdisk - I guess parted, fdisk, etc have the same - but it
> has the option to copy an MBR or GPT and generate new UUIDs for all
> the partitions. You MUST do that if you're leaving both drives in the
> system.
>
> Cheers,
> Wol
>
>


This is true.  Even I switched to UUID wherever possible or labels. 
Even external things I use UUID on when possible.  I've been bit by this
once before and it was a hair pulling experience that I wouldn't want to
repeat or wish on one either. 

If UUID is something you don't want to spend time learning right now,
try using labels at least.  Just make sure YOU use unique labels for
each one.  Hint.  home-old, home-new works pretty well at times.  At
least you know it is home and which is old and which is new.  Notes may
help too.  ;-)

Hope that helps.

Dale

:-)  :-) 

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 870 bytes --]

On Mon, 23 Nov 2020 13:51:26 -0600, Dale wrote:

> If UUID is something you don't want to spend time learning right now,
> try using labels at least.  Just make sure YOU use unique labels for
> each one.  Hint.  home-old, home-new works pretty well at times.  At
> least you know it is home and which is old and which is new.  Notes may
> help too.  ;-)

I agree on labels, they are far more readable. But I'm starting to think
that duplicating partitions like this is asking for trouble. I think it
would be better to create the partitions and filesystems you want on the
new disk, then mount both and copy everything over with rsync. That was
you won't get any conflicting UUIDs and you can set filesystem or
partition labels as you see fit.


-- 
Neil Bothwick

First Law of Laboratory Work:
Hot glass looks exactly the same as cold glass.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Mark Knecht]

[-- Attachment #1: Type: text/plain, Size: 1580 bytes --]

On Mon, Nov 23, 2020 at 1:11 PM Neil Bothwick <neil@digimed.co.uk> wrote:
>
> On Mon, 23 Nov 2020 13:51:26 -0600, Dale wrote:
>
> > If UUID is something you don't want to spend time learning right now,
> > try using labels at least.  Just make sure YOU use unique labels for
> > each one.  Hint.  home-old, home-new works pretty well at times.  At
> > least you know it is home and which is old and which is new.  Notes may
> > help too.  ;-)
>
> I agree on labels, they are far more readable. But I'm starting to think
> that duplicating partitions like this is asking for trouble. I think it
> would be better to create the partitions and filesystems you want on the
> new disk, then mount both and copy everything over with rsync. That was
> you won't get any conflicting UUIDs and you can set filesystem or
> partition labels as you see fit.
>

And correct me if I'm wrong but with rsync if something dies in process
you can usually start it back up and complete the job without starting over
from scratch.

I've avoided this thread until now but my preferred way to accomplish
what the OP set out to do was to do a very minimal install by hand, modify
any make/system specific options as has been discussed earlier, build a
kernel specific to the new machine, and then copy my world file and let
portage do the dirty work. Takes a lot of time but the system has always
been functional with few problems down the road.

After all, if you run Gentoo at all you MUST WANT to build software
from scratch. Why deny the fun of a complete machine rebuild? ;-)

Cheers,
Mark

- Mark

[-- Attachment #2: Type: text/html, Size: 2034 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Matt Connell (Gmail)]

On Mon, 2020-11-23 at 13:25 -0700, Mark Knecht wrote:
> And correct me if I'm wrong but with rsync if something dies in process
> you can usually start it back up and complete the job without starting over
> from scratch.

If you use the --partial flag, yes.  I don't think that is enabled by
default.

Re: [gentoo-user] duplicate gentoo system - errors

[antlists]

On 23/11/2020 20:25, Mark Knecht wrote:
>  > I agree on labels, they are far more readable. But I'm starting to think
>  > that duplicating partitions like this is asking for trouble. I think it
>  > would be better to create the partitions and filesystems you want on the
>  > new disk, then mount both and copy everything over with rsync. That was
>  > you won't get any conflicting UUIDs and you can set filesystem or
>  > partition labels as you see fit.
>  >
> 
> And correct me if I'm wrong but with rsync if something dies in process
> you can usually start it back up and complete the job without starting over
> from scratch.

If you dd the partition (which I'm planning to do), then there's no 
problem of uuids - that's at the MBR GPT level.

The trouble with rsync is if your partition is heavily hard-linked, like 
mine, you need to make sure you get your rsync options right or your 
hard-disk usage explodes, and if you do get them right, your ram usage 
explodes during the copy ... :-)

My problem is I'm going to be dd'ing a near-3TB near-full partition, and 
my wall-time-usage is going to explode ...

Cheers,
Wol

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

Neil Bothwick wrote:
> On Mon, 23 Nov 2020 13:51:26 -0600, Dale wrote:
>
>> If UUID is something you don't want to spend time learning right now,
>> try using labels at least.  Just make sure YOU use unique labels for
>> each one.  Hint.  home-old, home-new works pretty well at times.  At
>> least you know it is home and which is old and which is new.  Notes may
>> help too.  ;-)
> I agree on labels, they are far more readable. But I'm starting to think
> that duplicating partitions like this is asking for trouble. I think it
> would be better to create the partitions and filesystems you want on the
> new disk, then mount both and copy everything over with rsync. That was
> you won't get any conflicting UUIDs and you can set filesystem or
> partition labels as you see fit.
>
>


Way back when I kept outgrowing the disk I had the OS installed on, that
is what I did.  I booted from a CD/DVD.  I created the new partitions
and such.  I then created /mnt/old and /mnt/new.  I mounted the
partitions under the correct one, creating directories for boot, usr,
var etc as needed on the new disk.  Once all was mounted, I used cp -av
/mnt/old* /mnt/new/ to copy things over.  Today, I'd likely use rsync
tho.  If I had to interrupt the process, add the -u option.  That worked
then but not sure about now. 

I'd think UUIDs are always unique.  I guess there is a 1 in a billion
chance of a duplicate but doubtful it would be ran into by many. If
nothing else, labels work as long as unique names are used.  Using sda,
sdb, sdc etc tho is a problem waiting for something bad to happen. 
Heck, nowadays one can even change the order drives are seen in the
BIOS.  I'm not sure about EFI stuff.  I can make what is sda look like
sdb without ever taking the side off my puter case or unhooking
anything.  Using labels or UUIDs eliminate that problem. 

In a way, things are more complicated.  Thing is, if one uses UUIDs, and
only them, it's really pretty simple.  Copy and paste helps tho.  One
wouldn't want a typo.  lol  Of course, if one is careful, labels work
just as well. 

Dale

:-)  :-) 

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 497 bytes --]

On Mon, 23 Nov 2020 20:57:34 +0000, antlists wrote:

> > And correct me if I'm wrong but with rsync if something dies in
> > process you can usually start it back up and complete the job without
> > starting over from scratch.  
> 
> If you dd the partition (which I'm planning to do), then there's no 
> problem of uuids - that's at the MBR GPT level.

PARTUUIDs are are the GPT level, UUIDs are filesystem metadata.


-- 
Neil Bothwick

Stupid user error. Terminate user (Y/n) ?

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 754 bytes --]

On Mon, 23 Nov 2020 14:31:26 -0600, Matt Connell (Gmail) wrote:

> > And correct me if I'm wrong but with rsync if something dies in
> > process you can usually start it back up and complete the job without
> > starting over from scratch.  
> 
> If you use the --partial flag, yes.  I don't think that is enabled by
> default.

The --partial flag applies to individual files, so it may help if the
copy fails partway through copying a very large file, it's more useful
when used with networked drives.

Rsync will carry on where it left of as far as fully copied files are
concerned.


-- 
Neil Bothwick

Programming Language: (n.) a shorthand way of describing a series of bugs
                      to a computer or a programmer.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[William Kenworthy]

On 23/11/20 10:10 pm, Walter Dnes wrote:
> On Sun, Nov 22, 2020 at 04:39:44PM -0700, thelma@sys-concept.com wrote
>> Duplicating was easy, but when I try to recompile a kernel I get an error:
>>
>> make menuconfig
>> HOSTCC script/kconfig/mconf.o
>> <build-in> : internal compiler error: Illegal instruction
>>
>> Even if I try to run: emerge --info  I get:
>> Illegal instruction
>   Ouch!  Are the CPUs exactly identical?  If not, then you may get the
> "Illegal instruction" error.  This is a "feature" of Gentoo, which is
> often user-optimized for a specific CPU.  What's the CPU on the source
> machine and what's the CPU on the target machine?  If in doubt,
> execute...
>
> gcc -c -Q -march=native --help=target | grep march=
>
> ...on each machine.  There are 3 sub-families of "Intel" machines.
>
> 1) bog-standard Intel code, e.g. core2, also runs on AMD and Intel Atom
>    CPUs at least as new or newer.
>
> 2) Intel Atom code runs only on same or newer Intel Atom, because it has
>    the MOVBE instruction set the others lack.
>
> 3) AMD code runs only on same or newer AMD, because it has the 3DNow!
>    instruction set the others lack.
>
>   If you have such a mismatch, your only option may be to rebuild from
> scratch.
>
quick package the toolchain packages from the wanted cpu system, untar
them in the root of the new system so they overwrite the existing
package files, do an emerge -e @system, then follow with another for
world.  Can also use the install CD packages - I have used some quite
different versions when recovering a broken system so it <can> work :).

Have done it a few times, its a bit messier than than just
gcc/glibc/libtool/binutils for the toolchain but does work.  The two
emerge -e shouldn't be strictly necessary, but it seems to get a better
result than just doing world.


BillK

Re: [gentoo-user] duplicate gentoo system - errors

[Adam Carter]

[-- Attachment #1: Type: text/plain, Size: 207 bytes --]

> 3) AMD code runs only on same or newer AMD, because it has the 3DNow!
>    instruction set the others lack.
>

FYI 3dnow and 3dnowext went away some time ago. It's not in any of the
Bulldozer or Zen CPUs.

[-- Attachment #2: Type: text/html, Size: 454 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Peter Humphrey]

[-- Attachment #1: Type: text/plain, Size: 712 bytes --]

On Monday, 23 November 2020 19:02:57 GMT antlists wrote:

> If you're messing about with disks, partitions, etc, you NEED to have a
> basic understanding of UUIDs.

That may be true if you have more than one disk of a given type, but if you have only 
one SATA drive and one NVMe, for instance, there's no chance of their being 
misnumbered at boot.

My workstation has one NVMe drive and two SATAs. They're always detected in the 
same order, so I've no need to render my fstab illegible with UUIDs. I could use labels, 
but why bother? The old system ain't broke, so I've no need to fix it.

Can you imagine an fstab with 22 partitions specified with UUIDs? Doesn't bear 
thinking about.

-- 
Regards,
Peter.


[-- Attachment #2: Type: text/html, Size: 1558 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Michael]

[-- Attachment #1: Type: text/plain, Size: 1355 bytes --]

On Tuesday, 24 November 2020 09:20:52 GMT Peter Humphrey wrote:
> On Monday, 23 November 2020 19:02:57 GMT antlists wrote:
> > If you're messing about with disks, partitions, etc, you NEED to have a
> > basic understanding of UUIDs.
> 
> That may be true if you have more than one disk of a given type, but if you
> have only one SATA drive and one NVMe, for instance, there's no chance of
> their being misnumbered at boot.
> 
> My workstation has one NVMe drive and two SATAs. They're always detected in
> the same order, so I've no need to render my fstab illegible with UUIDs. I
> could use labels, but why bother? The old system ain't broke, so I've no
> need to fix it.

It depends on the bus and disk technology.  I have an ARM driven box with a 
conventional 1TB spinning SATA drive and a USB stick.  You can never tell 
which one will be detected as /dev/sda and which as /dev/sdb.  If you have 
more than one pluggable devices the same identification problem is likely to 
arise.  LABELs and/or UUIDs solve this problem - reliably.


> Can you imagine an fstab with 22 partitions specified with UUIDs? Doesn't
> bear thinking about.

Copying and pasting the output of blkid helps complete the fstab easily and 
commented lines allow me to explain to myself block device location and 
purpose, should I need to revisit it some months/years later.

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Peter Humphrey]

[-- Attachment #1: Type: text/plain, Size: 1601 bytes --]

On Tuesday, 24 November 2020 10:43:25 GMT Michael wrote:
> On Tuesday, 24 November 2020 09:20:52 GMT Peter Humphrey wrote:
> > My workstation has one NVMe drive and two SATAs. They're always detected
> > in the same order, so I've no need to render my fstab illegible with
> > UUIDs. I could use labels, but why bother? The old system ain't broke, so
> > I've no need to fix it.
> 
> It depends on the bus and disk technology.  I have an ARM driven box with a
> conventional 1TB spinning SATA drive and a USB stick.  You can never tell
> which one will be detected as /dev/sda and which as /dev/sdb.  If you have
> more than one pluggable devices the same identification problem is likely to
> arise.  LABELs and/or UUIDs solve this problem - reliably.

Yes, I have several USB sticks, but specifically because they're transient I expect those to 
have sdX assigned chronologically. I don't boot with them inserted, so I still don't need 
anything more than /dev/sdX in fstab.

> > Can you imagine an fstab with 22 partitions specified with UUIDs? Doesn't
> > bear thinking about.
> 
> Copying and pasting the output of blkid helps complete the fstab easily and
> commented lines allow me to explain to myself block device location and
> purpose, should I need to revisit it some months/years later.

That's still much more complex than my setup, and less legible. To a degree, this is a 
hobby machine, so I create, delete and move partitions more often than many people 
do. I couldn't possibly work with UUIDs; it'd be as bad as trying to read someone else's 
perl code.   :)

-- 
Regards,
Peter.


[-- Attachment #2: Type: text/html, Size: 3533 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 682 bytes --]

On Tue, 24 Nov 2020 09:20:52 +0000, Peter Humphrey wrote:

> My workstation has one NVMe drive and two SATAs. They're always
> detected in the same order, so I've no need to render my fstab
> illegible with UUIDs. I could use labels, but why bother? The old
> system ain't broke, so I've no need to fix it.

But you can fix it in your own time, waiting until it breaks is never
convenient.
 
> Can you imagine an fstab with 22 partitions specified with UUIDs?
> Doesn't bear thinking about.

Can you imagine an fstab with 22 partitions? Doesn't bear thinking about.

;-)


-- 
Neil Bothwick

Anything is possible if you don't know what
you are talking about.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Walter Dnes]

On Tue, Nov 24, 2020 at 07:56:07PM +1100, Adam Carter wrote
> > 3) AMD code runs only on same or newer AMD, because it has the 3DNow!
> >    instruction set the others lack.
> >
> 
> FYI 3dnow and 3dnowext went away some time ago. It's not in any of the
> Bulldozer or Zen CPUs.

  So you're saying that newer AMDs may not run code optimized for older
AMDs ???

-- 
Walter Dnes <waltdnes@waltdnes.org>
I don't run "desktop environments"; I run useful applications

[gentoo-user] Re: duplicate gentoo system - errors

[Grant Edwards]

On 2020-11-24, Neil Bothwick <neil@digimed.co.uk> wrote:
> On Tue, 24 Nov 2020 09:20:52 +0000, Peter Humphrey wrote:
>  
>> Can you imagine an fstab with 22 partitions specified with UUIDs?
>> Doesn't bear thinking about.
>
> Can you imagine an fstab with 22 partitions? Doesn't bear thinking
> about.

Yes. I have one with 12 and often wish it had more.

It comes in handy when you need to test drivers and applications
against different distros, and your customers use 4-5 different
versions of RHEL, 4-5 different versions of Ubuntu, a couple random
versions of SuSe, Mint, Arch, etc.  You could of course, have 12 (or
22) different hard drives that you swap in/out of the machine, but
that's tiresome and expensive.

--
Grant

Re: [gentoo-user] Re: duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 961 bytes --]

On Tue, 24 Nov 2020 15:01:35 -0000 (UTC), Grant Edwards wrote:

> > Can you imagine an fstab with 22 partitions? Doesn't bear thinking
> > about.  
> 
> Yes. I have one with 12 and often wish it had more.
> 
> It comes in handy when you need to test drivers and applications
> against different distros, and your customers use 4-5 different
> versions of RHEL, 4-5 different versions of Ubuntu, a couple random
> versions of SuSe, Mint, Arch, etc.  You could of course, have 12 (or
> 22) different hard drives that you swap in/out of the machine, but
> that's tiresome and expensive.

But actual partitions? Without some sort of volume management? I have 27
btrfs subvolumes on one system, but I wouldn't want to do that with
partitions. And you don't find yourself wishing for more with a volume
manager, just create whatever you need... which is how I reached 27.


-- 
Neil Bothwick

DOS never says "EXCELLENT command or filename"...

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] Re: duplicate gentoo system - errors

[Jack]

On 11/24/20 10:41 AM, Neil Bothwick wrote:
> On Tue, 24 Nov 2020 15:01:35 -0000 (UTC), Grant Edwards wrote:
>
>>> Can you imagine an fstab with 22 partitions? Doesn't bear thinking
>>> about.
>> Yes. I have one with 12 and often wish it had more.
>>
>> It comes in handy when you need to test drivers and applications
>> against different distros, and your customers use 4-5 different
>> versions of RHEL, 4-5 different versions of Ubuntu, a couple random
>> versions of SuSe, Mint, Arch, etc.  You could of course, have 12 (or
>> 22) different hard drives that you swap in/out of the machine, but
>> that's tiresome and expensive.
> But actual partitions? Without some sort of volume management? I have 27
> btrfs subvolumes on one system, but I wouldn't want to do that with
> partitions. And you don't find yourself wishing for more with a volume
> manager, just create whatever you need... which is how I reached 27.
I only have two or three such distros I use for testing, but I have each 
in a VirtualBox machine.  For me, spinning up a VM is easier than a real 
reboot.

[gentoo-user] Re: duplicate gentoo system - errors

[Grant Edwards]

On 2020-11-24, Neil Bothwick <neil@digimed.co.uk> wrote:
> On Tue, 24 Nov 2020 15:01:35 -0000 (UTC), Grant Edwards wrote:
>
>> > Can you imagine an fstab with 22 partitions? Doesn't bear thinking
>> > about.  
>> 
>> Yes. I have one with 12 and often wish it had more.
>> 
>> It comes in handy when you need to test drivers and applications
>> against different distros, and your customers use 4-5 different
>> versions of RHEL, 4-5 different versions of Ubuntu, a couple random
>> versions of SuSe, Mint, Arch, etc.  You could of course, have 12 (or
>> 22) different hard drives that you swap in/out of the machine, but
>> that's tiresome and expensive.
>
> But actual partitions?

Yes. Each with a separate Linux distro installed.

Perhaps you can do that with a volume manager instead of partitions
(using partitions has worked fine for the past 20 years, so I've never
looked into it).

[gentoo-user] Re: duplicate gentoo system - errors

[Grant Edwards]

On 2020-11-24, Jack <ostroffjh@users.sourceforge.net> wrote:

> I only have two or three such distros I use for testing, but I have
> each in a VirtualBox machine. For me, spinning up a VM is easier
> than a real reboot.

I don't trust VMs when testing drivers for PCI cards or applicatoins
that use raw Ethernet.

--
Grant

Re: [gentoo-user] Re: duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 688 bytes --]

On Tue, 24 Nov 2020 16:38:59 -0000 (UTC), Grant Edwards wrote:

> > But actual partitions?  
> 
> Yes. Each with a separate Linux distro installed.
> 
> Perhaps you can do that with a volume manager instead of partitions
> (using partitions has worked fine for the past 20 years, so I've never
> looked into it).

Of course you can, and it makes adding and removing distros much simpler.

I've also done it on this laptop with btrfs subvolumes, a different
subvolume for each distro's root but a shared /home subvolume. All of
them on a dm-crypt device, so only one password needed whichever distro
is used.


-- 
Neil Bothwick

Can vegetarians eat animal crackers?

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[gentoo-user] Re: duplicate gentoo system - errors

[Grant Edwards]

On 2020-11-24, Neil Bothwick <neil@digimed.co.uk> wrote:
> On Tue, 24 Nov 2020 16:38:59 -0000 (UTC), Grant Edwards wrote:
>
>> > But actual partitions?  
>> 
>> Yes. Each with a separate Linux distro installed.
>> 
>> Perhaps you can do that with a volume manager instead of partitions
>> (using partitions has worked fine for the past 20 years, so I've
>> never looked into it).
>
> Of course you can, and it makes adding and removing distros much
> simpler.

Cool, I'll have to read up on using volumes for that. How far back in
time can you go before you get to distros that would have problems?

--
Grant

Re: [gentoo-user] duplicate gentoo system - errors

[thelma]

On 11/23/2020 01:29 AM, Neil Bothwick wrote:
> On Sun, 22 Nov 2020 18:27:53 -0700, thelma@sys-concept.com wrote:
> 
>>> I would confirm that you are really booted from the new disk and not
>>> the old one.  It is possible that the MBR from the new disk was used
>>> to boot, but if /etc/fstab says /boot is mounted from /dev/sda1 then
>>> that does seem wrong.  I almost always put an empty file in the root
>>> of each partition named for the disk/partition just so I can be sure
>>> what's actually mounted.  Is /etc/fstab identical on both disks?
>>> What does fstab say about where / is mounted from?  
>>
>> You are absolutely correct.  I was booting the whole time the Western
>> Digital (old drive). :-/  My mistake, once I removed the WD drive the
>> new M.2 SSD doesn't even boot.
> 
> Check the settings in your BIOS/firmware to make sure it detects the
> drive and is set to boot from it.

I'm back.

Yes, BIOS recognized the system, I have one SATA disk connected to it and it is booting OK. But disconnected it so I don't mess something up.

>> I think the easiest way would be to re-install the Getnoo from scratch
> 
> And if it still doesn't work because of a firmware issue, you still have
> a non-booting system but with no OS installed either. It's better to try
> and diagnose the problem rather than throwing everything away in the hope
> that the problem goes with it.

I booted from Gontoo bootable USB and running: blkid
showing:
blkid
/dev/loop0: TYPE="squashfs"
/dev/nvme0n1p1: UUID="5db43d49-810a-4806-955e-d59c4d35ec23" BLOCK_SIZE="4096" TYPE="ext2" PARTUUID="743a7887-c02e-4855-8cb7-865247682bff"
/dev/nvme0n1p2: UUID="0c23b340-b5c6-437d-bde9-c5539e64677a" BLOCK_SIZE="4096" TYPE="ext4" PARTUUID="696091c3-ed27-4a4a-b371-8fd59e2b7a4d"
/dev/nvme0n1p3: UUID="77e449db-7dca-410d-9e70-50165c6ccbb8" TYPE="swap" PARTUUID="b2871b7b-5bd0-4db3-90a5-50545b129a97"


Disk /dev/nvme0n1: 1.84 TiB, 2000398934016 bytes, 3907029168 sectors
Disk model: Force MP600                             
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: FE896335-0C8D-487E-9391-ED43A85D3292

Device              Start        End    Sectors   Size Type
/dev/nvme0n1p1       2048    1050623    1048576   512M Linux filesystem
/dev/nvme0n1p2    1050624 1971752959 1970702336 939.7G Linux filesystem
/dev/nvme0n1p3 3840102400 3907028991   66926592  31.9G Linux swap

Shouldn't:  /dev/nvme0n1p1  be "Bios Boot" 
Do I need to change it with "fdisk"

When trying to mount /boot
mount /dev/nvme0n1p1 /mnt/gentoo/boot
livecd ~ # ll /mnt/gentoo/boot/
total 16
drwx------ 2 root root 16384 Nov 22 22:26 lost+found

There is nothing there.  
I'm not sure if there is a point of fixing it at this point.  It might take less time to reinstall gentoo. 

Re: [gentoo-user] Re: duplicate gentoo system - errors

[antlists]

On 24/11/2020 16:51, Grant Edwards wrote:
> On 2020-11-24, Neil Bothwick <neil@digimed.co.uk> wrote:
>> On Tue, 24 Nov 2020 16:38:59 -0000 (UTC), Grant Edwards wrote:
>>
>>>> But actual partitions?
>>>
>>> Yes. Each with a separate Linux distro installed.
>>>
>>> Perhaps you can do that with a volume manager instead of partitions
>>> (using partitions has worked fine for the past 20 years, so I've
>>> never looked into it).
>>
>> Of course you can, and it makes adding and removing distros much
>> simpler.
> 
> Cool, I'll have to read up on using volumes for that. How far back in
> time can you go before you get to distros that would have problems?
> 
How old is LVM? It's been around for ages, I think. The other nice 
advantage of LVM is that if you've got a big archive disk, you can 
migrate the partitions to that disk and move them off of your live 
system (but you can migrate them back if you need them).

Personally, I wouldn't use dm-crypt, but then I'm not particularly into 
crypto.

Read up on it, but I would be surprised if it didn't support stuff right 
back to last century.

Cheers,
Wol

Re: [gentoo-user] duplicate gentoo system - errors

[Michael]

[-- Attachment #1: Type: text/plain, Size: 5168 bytes --]

On Tuesday, 24 November 2020 17:23:41 GMT thelma@sys-concept.com wrote:
> On 11/23/2020 01:29 AM, Neil Bothwick wrote:
> > On Sun, 22 Nov 2020 18:27:53 -0700, thelma@sys-concept.com wrote:
> >>> I would confirm that you are really booted from the new disk and not
> >>> the old one.  It is possible that the MBR from the new disk was used
> >>> to boot, but if /etc/fstab says /boot is mounted from /dev/sda1 then
> >>> that does seem wrong.  I almost always put an empty file in the root
> >>> of each partition named for the disk/partition just so I can be sure
> >>> what's actually mounted.  Is /etc/fstab identical on both disks?
> >>> What does fstab say about where / is mounted from?
> >> 
> >> You are absolutely correct.  I was booting the whole time the Western
> >> Digital (old drive). :-/  My mistake, once I removed the WD drive the
> >> new M.2 SSD doesn't even boot.
> > 
> > Check the settings in your BIOS/firmware to make sure it detects the
> > drive and is set to boot from it.
> 
> I'm back.
> 
> Yes, BIOS recognized the system, I have one SATA disk connected to it and it
> is booting OK. But disconnected it so I don't mess something up.

You'll need to reconnect the old SATA in a minute - see below.


> >> I think the easiest way would be to re-install the Getnoo from scratch
> > 
> > And if it still doesn't work because of a firmware issue, you still have
> > a non-booting system but with no OS installed either. It's better to try
> > and diagnose the problem rather than throwing everything away in the hope
> > that the problem goes with it.
> 
> I booted from Gontoo bootable USB and running: blkid
> showing:
> blkid
> /dev/loop0: TYPE="squashfs"
> /dev/nvme0n1p1: UUID="5db43d49-810a-4806-955e-d59c4d35ec23"
> BLOCK_SIZE="4096" TYPE="ext2"
> PARTUUID="743a7887-c02e-4855-8cb7-865247682bff" /dev/nvme0n1p2:
> UUID="0c23b340-b5c6-437d-bde9-c5539e64677a" BLOCK_SIZE="4096" TYPE="ext4"
> PARTUUID="696091c3-ed27-4a4a-b371-8fd59e2b7a4d" /dev/nvme0n1p3:
> UUID="77e449db-7dca-410d-9e70-50165c6ccbb8" TYPE="swap"
> PARTUUID="b2871b7b-5bd0-4db3-90a5-50545b129a97"
> 
> 
> Disk /dev/nvme0n1: 1.84 TiB, 2000398934016 bytes, 3907029168 sectors
> Disk model: Force MP600
> 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: FE896335-0C8D-487E-9391-ED43A85D3292
> 
> Device              Start        End    Sectors   Size Type
> /dev/nvme0n1p1       2048    1050623    1048576   512M Linux filesystem
> /dev/nvme0n1p2    1050624 1971752959 1970702336 939.7G Linux filesystem
> /dev/nvme0n1p3 3840102400 3907028991   66926592  31.9G Linux swap
> 
> Shouldn't:  /dev/nvme0n1p1  be "Bios Boot"
> Do I need to change it with "fdisk"

The /dev/nvme0n1p1 partition type should be whatever the /dev/sda1 of the old 
disk was.  If you are using a GPT disk on a Legacy BIOS MoBo, with a Hybrid 
MBR configuration, then yes, it should be a 'BIOS Boot partition' to contain 
the GRUB Stage 2 code.

However, if /dev/nvme0n1p1 is just a conventional /boot partition with an ext2 
fs as you show above, containing your /boot/grub/ directory files, then 'Linux 
filesystem' is the correct partition type.


> When trying to mount /boot
> mount /dev/nvme0n1p1 /mnt/gentoo/boot
> livecd ~ # ll /mnt/gentoo/boot/
> total 16
> drwx------ 2 root root 16384 Nov 22 22:26 lost+found
> 
> There is nothing there.

So the contents of /dev/sda1 were not copied over.

Were the contents of /dev/sda2 copied over to dev/nvme0n1p2 ?


> I'm not sure if there is a point of fixing it at this point.  It might take
> less time to reinstall gentoo.

It will take much less time to use clonezilla to clone the whole disk as I 
originally suggested, change the UUIDs of the partitions, update GRUB on the 
new disk, then update GRUB on the old disk.

Alternatively, it will also take less time to use Gparted to create and format 
the partitions you want, then use rsync/tar/cp to copy over the filesystem 
contents from the old to the new partitions, then update GRUB.

If /dev/nvme0n1p1 is the only partition which needs fixing and since this is a 
small partition in size, it won't take long to reconnect /dev/sda to the MoBo, 
boot a LiveUSB and run:

dd if=/dev/sda1 of=/dev/nvme0n1p1

If you have not copied over the MBR boot loader in sector 0, or have not 
installed afresh GRUB on the new disk, then you will also need to copy over 
the MBR sector:

dd if=/dev/sda of=/dev/nvme0n1 bs=446 count=1

Or, if you want to also copy over the MBR partition table as it was on the old 
disk, change the block size to 512 bytes:

dd if=/dev/sda of=/dev/nvme0n1 bs=512 count=1

Any of the above should take significantly less time than reinstalling gentoo, 
then fishing old ebuilds of your required applications from the attic, 
fetching corresponding source files, only to discover they won't compile on 
the new installation because of whatever changes may have taken place in your 
installed toolchain.

Anyway, once you get the knack of this, you'll be able to back up or clone 
disks/partitions in the future for all sorts of reasons.

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Rich Freeman]

On Tue, Nov 24, 2020 at 9:49 AM Walter Dnes <waltdnes@waltdnes.org> wrote:
>
> On Tue, Nov 24, 2020 at 07:56:07PM +1100, Adam Carter wrote
> > > 3) AMD code runs only on same or newer AMD, because it has the 3DNow!
> > >    instruction set the others lack.
> > >
> >
> > FYI 3dnow and 3dnowext went away some time ago. It's not in any of the
> > Bulldozer or Zen CPUs.
>
>   So you're saying that newer AMDs may not run code optimized for older
> AMDs ???

It depends on what you mean by "optimized."  If you mean using -march,
then yes, newer AMD OR Intel CPUs may not run software built for older
ones.  They get rid of instructions as well as adding them, and some
lines have instructions not present in others.

If you use -mtune then the code will work on any amd64 processor - AMD or Intel.

If you use -march then you generally need to rebuild everything
without it (or at least @system) before you go switching CPUs.

-- 
Rich

Re: [gentoo-user] Re: duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 457 bytes --]

On Tue, 24 Nov 2020 18:37:43 +0000, antlists wrote:

> Personally, I wouldn't use dm-crypt, but then I'm not particularly into 
> crypto.

It's a laptop that could be lost or stolen, not using some form of
encryption would be insane. I don't use dm-crypt on desktop systems as
long as the environment is secure.


-- 
Neil Bothwick

"We can't solve problems by using the same kind of thinking we used when
we created them." (Albert Einstein)

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Peter Humphrey]

[-- Attachment #1: Type: text/plain, Size: 1094 bytes --]

On Tuesday, 24 November 2020 14:18:58 GMT Neil Bothwick wrote:
> On Tue, 24 Nov 2020 09:20:52 +0000, Peter Humphrey wrote:
> > My workstation has one NVMe drive and two SATAs. They're always
> > detected in the same order, so I've no need to render my fstab
> > illegible with UUIDs. I could use labels, but why bother? The old
> > system ain't broke, so I've no need to fix it.
> 
> But you can fix it in your own time, waiting until it breaks is never
> convenient.

There's nothing to fix, as I said. I'm happy to stick with the /dev/sdX syntax for as long as 
it remains valid. Occam's Razor applies: "don't complicate beyond need."

> > Can you imagine an fstab with 22 partitions specified with UUIDs?
> > Doesn't bear thinking about.
> 
> Can you imagine an fstab with 22 partitions? Doesn't bear thinking about.

The NVMe drive, the main one, has 18; I could merge some of those and delete a couple 
that aren't used any more. The packages and distfiles directories don't need separate 
partitions, for example. I suppose it's a bit like Topsy, who "just growed."

-- 
Regards,
Peter.


[-- Attachment #2: Type: text/html, Size: 2633 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

Peter Humphrey wrote:
>
> On Tuesday, 24 November 2020 14:18:58 GMT Neil Bothwick wrote:
>
> > On Tue, 24 Nov 2020 09:20:52 +0000, Peter Humphrey wrote:
>
> > > My workstation has one NVMe drive and two SATAs. They're always
>
> > > detected in the same order, so I've no need to render my fstab
>
> > > illegible with UUIDs. I could use labels, but why bother? The old
>
> > > system ain't broke, so I've no need to fix it.
>
> >
>
> > But you can fix it in your own time, waiting until it breaks is never
>
> > convenient.
>
>
> There's nothing to fix, as I said. I'm happy to stick with the
> /dev/sdX syntax for as long as it remains valid. Occam's Razor
> applies: "don't complicate beyond need."
>
>
> > > Can you imagine an fstab with 22 partitions specified with UUIDs?
>
> > > Doesn't bear thinking about.
>
> >
>
> > Can you imagine an fstab with 22 partitions? Doesn't bear thinking
> about.
>
>
> The NVMe drive, the main one, has 18; I could merge some of those and
> delete a couple that aren't used any more. The packages and distfiles
> directories don't need separate partitions, for example. I suppose
> it's a bit like Topsy, who "just growed."
>
>
> -- 
>
> Regards,
>
> Peter.
>
>

I didn't think I needed to "fix" it either until it hit me and caused
confusion.  Eventually I figured out it was mounting the wrong thing but
it was a head scratcher for a while.  I was about to start over when I
noticed it was mounting the wrong partitions.  I can't recall what
changed the order but suddenly sda and sdb switched.  Believe me, when I
got booted, I started setting it up in a way that can't happen again. 

I might add, the more partitions you have, the more likely this is to
bite you at some point.  You already have a complicated system with
chainloading bootloaders and such so Occam left the building long ago. 
Do you really need for a hard drive to be recognized differently and
create problems?  At the very least, labels would be a much better
option.  Labels like ubuntu-home, ubuntu-usr, or redhat-root, or
redhat-usr.  Those explain what they are and makes them unique.  If you
have more than one version, include part of a version if needed. 

You may recall my hatred of the init thingys.  I still hate them.  I use
them because I want the best chance of my system booting and without it,
that could fail.  It may boot 100 times just fine but then one day, it
breaks and won't boot anymore without a init thingy.  At that point, I
get to sit here, most likely with no way to get help, and figure out how
to fix it.  To me, it's much better to just go ahead and set up using
the thing and not having to worry about that day hitting me.  It seems
bad things always happen at the worst moment too. 

If I can start using a init thingy, using labels should be a easy
thing.  A walk in the park as the saying goes.  ;-)

Just my thoughts and opinions. 

Dale

:-)  :-) 

Re: [gentoo-user] duplicate gentoo system - errors

[Rich Freeman]

On Wed, Nov 25, 2020 at 5:54 AM Peter Humphrey <peter@prh.myzen.co.uk> wrote:
>
> > > Can you imagine an fstab with 22 partitions specified with UUIDs?
>
> > Can you imagine an fstab with 22 partitions?
>
> The NVMe drive, the main one, has 18;

So, if all the partitions are on one drive and that is the only drive
you have, there aren't many issues with using raw kernel device names
to identify them.  It isn't like a partition is just going to
disappear.

Once you have multiple disks, then UUIDs or labels become more
important, especially with a large number.  If you had a dozen disks
with dozens of partitions and tried to use kernel device names, then
anytime a device failed or was enumerated differently you'd have stuff
mounted all over the place.

That said, something like lvm is a good solution in almost all cases
(or something semi-equivalent like zfs/btrfs/etc which have similar
functionality built-in).  If I had that many partitions I'd hate to
deal with wanting to resize one, and with lvm that is pretty trivial.
You don't need to use UUIDs with lvm - they're basically equivalent to
labels.

Now, one area I would use UUIDs is with mdadm if you're not putting
lvm on top.  I've seen mdadm arrays get renumbered and that is a mess
if you're directly mounting them without labels or UUIDs.

-- 
Rich

Re: [gentoo-user] duplicate gentoo system - errors

[Wols Lists]

On 25/11/20 13:31, Rich Freeman wrote:
> Now, one area I would use UUIDs is with mdadm if you're not putting
> lvm on top.  I've seen mdadm arrays get renumbered and that is a mess
> if you're directly mounting them without labels or UUIDs.

Or if you do it properly you don't need UUIDs :-)

mdadm always USED to number its arrays starting with 0. Now it counts
down from 127.

Worse, if you use numbers, mdadm just changes them as it sees fit. I
created my array as md0, next thing I know it's md127.

It is recommended to use names, so I call it by what it is, so I have
things like /dev/md/gentoo, /dev/md/home etc.

I'm moving to lvm, and will use the same tactic.

Cheers,
Wol

Re: [gentoo-user] duplicate gentoo system - errors

[Peter Humphrey]

[-- Attachment #1: Type: text/plain, Size: 3713 bytes --]

On Wednesday, 25 November 2020 13:06:49 GMT Dale wrote:
> Peter Humphrey wrote:
> > On Tuesday, 24 November 2020 14:18:58 GMT Neil Bothwick wrote:
> > > On Tue, 24 Nov 2020 09:20:52 +0000, Peter Humphrey wrote:
> > > > My workstation has one NVMe drive and two SATAs. They're always
> > > > 
> > > > detected in the same order, so I've no need to render my fstab
> > > > 
> > > > illegible with UUIDs. I could use labels, but why bother? The old
> > > > 
> > > > system ain't broke, so I've no need to fix it.
> > > 
> > > But you can fix it in your own time, waiting until it breaks is never
> > > 
> > > convenient.
> > 
> > There's nothing to fix, as I said. I'm happy to stick with the
> > /dev/sdX syntax for as long as it remains valid. Occam's Razor
> > applies: "don't complicate beyond need."
> > 
> > > > Can you imagine an fstab with 22 partitions specified with UUIDs?
> > > > 
> > > > Doesn't bear thinking about.
> > > 
> > > Can you imagine an fstab with 22 partitions? Doesn't bear thinking
> > 
> > about.
> > 
> > 
> > The NVMe drive, the main one, has 18; I could merge some of those and
> > delete a couple that aren't used any more. The packages and distfiles
> > directories don't need separate partitions, for example. I suppose
> > it's a bit like Topsy, who "just growed."
> > 
> > 
> > 
> > Regards,
> > 
> > Peter.
> 
> I didn't think I needed to "fix" it either until it hit me and caused
> confusion.  Eventually I figured out it was mounting the wrong thing but
> it was a head scratcher for a while.  I was about to start over when I
> noticed it was mounting the wrong partitions.  I can't recall what
> changed the order but suddenly sda and sdb switched.  Believe me, when I
> got booted, I started setting it up in a way that can't happen again. 
> 
> I might add, the more partitions you have, the more likely this is to
> bite you at some point.  You already have a complicated system with
> chainloading bootloaders and such so Occam left the building long ago. 

Actually I haven't any of those things. Grub, in particular, will never have a place in my 
home. I jusr have EFI boot images in the UEFI BIOS. Simple. I do take the point, though.

> Do you really need for a hard drive to be recognized differently and
> create problems?  At the very least, labels would be a much better
> option.  Labels like ubuntu-home, ubuntu-usr, or redhat-root, or
> redhat-usr.  Those explain what they are and makes them unique.  If you
> have more than one version, include part of a version if needed. 

I'll think about that the very first time I get sda and sdb reversed. Honest. :)

> You may recall my hatred of the init thingys.  I still hate them.

I do. Me too. I still don't use one.

> I use them because I want the best chance of my system booting and without
> it, that could fail.  It may boot 100 times just fine but then one day, it
> breaks and won't boot anymore without a init thingy.  At that point, I
> get to sit here, most likely with no way to get help, and figure out how
> to fix it.  To me, it's much better to just go ahead and set up using
> the thing and not having to worry about that day hitting me.  It seems
> bad things always happen at the worst moment too. 
> 
> If I can start using a init thingy, using labels should be a easy
> thing.  A walk in the park as the saying goes.  ;-)

Quite so. As I said, I haven't needed to yet, but I'll think about it in due course.  :)

You may remember my asking why you need a separate /usr partition. You wouldn't 
need an init thingy if you merged it into the root partition. I have /var separate for 
simplicity of backup and recovery, and to contain log-file runaways, but not /usr.

-- 
Regards,
Peter.


[-- Attachment #2: Type: text/html, Size: 9848 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Peter Humphrey]

On Wednesday, 25 November 2020 13:31:27 GMT Rich Freeman wrote:
> On Wed, Nov 25, 2020 at 5:54 AM Peter Humphrey <peter@prh.myzen.co.uk> 
wrote:
> > > > Can you imagine an fstab with 22 partitions specified with UUIDs?
> > > 
> > > Can you imagine an fstab with 22 partitions?
> > 
> > The NVMe drive, the main one, has 18;
> 
> So, if all the partitions are on one drive and that is the only drive
> you have, there aren't many issues with using raw kernel device names
> to identify them.  It isn't like a partition is just going to
> disappear.
> 
> Once you have multiple disks, then UUIDs or labels become more
> important, especially with a large number.  If you had a dozen disks
> with dozens of partitions and tried to use kernel device names, then
> anytime a device failed or was enumerated differently you'd have stuff
> mounted all over the place.

Oh yes, of course, I can see that. I'm only saying that a simple system only 
needs simple setup.

> That said, something like lvm is a good solution in almost all cases
> (or something semi-equivalent like zfs/btrfs/etc which have similar
> functionality built-in).  If I had that many partitions I'd hate to
> deal with wanting to resize one, and with lvm that is pretty trivial.
> You don't need to use UUIDs with lvm - they're basically equivalent to
> labels.

My old system had two 1TB SSDs, and I used lvm on them. It was a lot of extra 
complication, so I didn't take that approach on this box. (I still have to 
have mdadm and friends installed though; the grains aren't fine enough to split 
them out. USE in make.conf includes "-dmraid -device-mapper -lvm" but it's 
ineffective.)

> Now, one area I would use UUIDs is with mdadm if you're not putting
> lvm on top.  I've seen mdadm arrays get renumbered and that is a mess
> if you're directly mounting them without labels or UUIDs.

<Shudder.>

-- 
Regards,
Peter.

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

Peter Humphrey wrote:
>
> On Wednesday, 25 November 2020 13:06:49 GMT Dale wrote:
>
> > Peter Humphrey wrote:
>
> > > On Tuesday, 24 November 2020 14:18:58 GMT Neil Bothwick wrote:
>
> > > > On Tue, 24 Nov 2020 09:20:52 +0000, Peter Humphrey wrote:
>
> > > > > My workstation has one NVMe drive and two SATAs. They're always
>
> > > > >
>
> > > > > detected in the same order, so I've no need to render my fstab
>
> > > > >
>
> > > > > illegible with UUIDs. I could use labels, but why bother? The old
>
> > > > >
>
> > > > > system ain't broke, so I've no need to fix it.
>
> > > >
>
> > > > But you can fix it in your own time, waiting until it breaks is
> never
>
> > > >
>
> > > > convenient.
>
> > >
>
> > > There's nothing to fix, as I said. I'm happy to stick with the
>
> > > /dev/sdX syntax for as long as it remains valid. Occam's Razor
>
> > > applies: "don't complicate beyond need."
>
> > >
>
> > > > > Can you imagine an fstab with 22 partitions specified with UUIDs?
>
> > > > >
>
> > > > > Doesn't bear thinking about.
>
> > > >
>
> > > > Can you imagine an fstab with 22 partitions? Doesn't bear thinking
>
> > >
>
> > > about.
>
> > >
>
> > >
>
> > > The NVMe drive, the main one, has 18; I could merge some of those and
>
> > > delete a couple that aren't used any more. The packages and distfiles
>
> > > directories don't need separate partitions, for example. I suppose
>
> > > it's a bit like Topsy, who "just growed."
>
> > >
>
> > >
>
> > >
>
> > > Regards,
>
> > >
>
> > > Peter.
>
> >
>
> > I didn't think I needed to "fix" it either until it hit me and caused
>
> > confusion.  Eventually I figured out it was mounting the wrong thing but
>
> > it was a head scratcher for a while.  I was about to start over when I
>
> > noticed it was mounting the wrong partitions.  I can't recall what
>
> > changed the order but suddenly sda and sdb switched.  Believe me, when I
>
> > got booted, I started setting it up in a way that can't happen again. 
>
> >
>
> > I might add, the more partitions you have, the more likely this is to
>
> > bite you at some point.  You already have a complicated system with
>
> > chainloading bootloaders and such so Occam left the building long ago. 
>
>
> Actually I haven't any of those things. Grub, in particular, will
> never have a place in my home. I jusr have EFI boot images in the UEFI
> BIOS. Simple. I do take the point, though.
>
>
> > Do you really need for a hard drive to be recognized differently and
>
> > create problems?  At the very least, labels would be a much better
>
> > option.  Labels like ubuntu-home, ubuntu-usr, or redhat-root, or
>
> > redhat-usr.  Those explain what they are and makes them unique.  If you
>
> > have more than one version, include part of a version if needed. 
>
>
> I'll think about that the very first time I get sda and sdb reversed.
> Honest. :)
>
>
> > You may recall my hatred of the init thingys.  I still hate them.
>
>
> I do. Me too. I still don't use one.
>
>
> > I use them because I want the best chance of my system booting and
> without
>
> > it, that could fail.  It may boot 100 times just fine but then one
> day, it
>
> > breaks and won't boot anymore without a init thingy.  At that point, I
>
> > get to sit here, most likely with no way to get help, and figure out how
>
> > to fix it.  To me, it's much better to just go ahead and set up using
>
> > the thing and not having to worry about that day hitting me.  It seems
>
> > bad things always happen at the worst moment too. 
>
> >
>
> > If I can start using a init thingy, using labels should be a easy
>
> > thing.  A walk in the park as the saying goes.  ;-)
>
>
> Quite so. As I said, I haven't needed to yet, but I'll think about it
> in due course.  :)
>
>
> You may remember my asking why you need a separate /usr partition. You
> wouldn't need an init thingy if you merged it into the root partition.
> I have /var separate for simplicity of backup and recovery, and to
> contain log-file runaways, but not /usr.
>
>
> -- 
>
> Regards,
>
> Peter.
>
>


Since I did this install a few years ago, I've had to grow /usr at least
twice.  I also had to grow /var when I moved the portage tree and such
to /var.  I have /boot on a plain ext2 partition, root is also on a
plain ext4 partition.  Everything else, /home, /usr, /var etc is on
LVM.  I have a very good reason for having a separate /usr.  Generally,
the only thing a person would need to expand as software grows is /usr. 
That is where most things go except for things needed to boot.  I've
also expanded /home many times.  Using LVM for that makes it very easy. 
If I were to build a new system and do a fresh install today, I'd like
to have everything possible on LVM, including the root partition if it's
doable.  If Grub supports /boot on LVM, I'd do that too. 

If I hadn't had a separate /usr, I would have had to move things around
to grow /usr.  I've done that in the past and got very tired of doing it
the hard way.  With LVM, it's just a few commands and is done while in
use even.  I don't even have to logout, reboot or anything.  That's a
very good reason for having /usr separate from /.  I might add, I
installed a couple CAD type programs to do circuit boards with and they
are huge.  If I ever remove those, I can likely shrink /usr a fair
amount if needed.

I can't think of a reason not to use labels, at the very least, in most
situations.  The only one I can think of, a laptop that has only one
hard drive.  Sort of hard to install two hard drives on a laptop.  A
external one can be done but never seen one with two spots for internal
hard drives.  Do they make those???

Of course, one can do things the hard way.  I'm just glad I listened
when Alan, Neil and several others who recommended LVM to me several
years ago.  It made my life a lot easier.  I might add, I think that is
what got me away from sda* and such and using labels and UUIDs.  I'm not
real big on change at times but when people who have been there explains
a better way, I do give it some serious thought.  Most of the time, I
change and I might add, I'm all the better for it.  If something goes
wrong with how your drives are seen, you got problems.  Given I use
labels and UUIDs, I won't have to worry about any of that. 

Just my thoughts.  Trying to help.  It's not like I don't ever find
myself on the receiving end of that.  :-D 

Dale

:-)  :-) 

Re: [gentoo-user] duplicate gentoo system - errors

[Rich Freeman]

On Wed, Nov 25, 2020 at 8:55 AM Wols Lists <antlists@youngman.org.uk> wrote:
>
> On 25/11/20 13:31, Rich Freeman wrote:
> > Now, one area I would use UUIDs is with mdadm if you're not putting
> > lvm on top.  I've seen mdadm arrays get renumbered and that is a mess
> > if you're directly mounting them without labels or UUIDs.
>
> It is recommended to use names, so I call it by what it is, so I have
> things like /dev/md/gentoo, /dev/md/home etc.

Is that supported with the original metadata format?  I suspect that
was a big constraint since at the time my bootloader didn't support
anything newer.

This was many years ago.  I haven't touched mdadm recently in favor of
zfs and lizardfs.

-- 
Rich

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 795 bytes --]

On Wed, 25 Nov 2020 09:13:07 -0600, Dale wrote:

> I have /boot on a plain ext2 partition, root is also on a
> plain ext4 partition.  Everything else, /home, /usr, /var etc is on
> LVM.

> If I hadn't had a separate /usr, I would have had to move things around
> to grow /usr.  I've done that in the past and got very tired of doing it
> the hard way.  With LVM, it's just a few commands and is done while in
> use even.  I don't even have to logout, reboot or anything.  That's a
> very good reason for having /usr separate from /. 

I'd say it's more a very good reason to put / on LVM too. I used to use a
separate /usr but found no real benefit so I now leave it as part of /.


-- 
Neil Bothwick

A journey of a thousand miles begins with a cash advance from Mom.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Peter Humphrey]

On Wednesday, 25 November 2020 15:13:07 GMT Dale wrote:

> Just my thoughts.  Trying to help.  It's not like I don't ever find
> myself on the receiving end of that.  :-D 

Indeed. Thanks Dale.

-- 
Regards,
Peter.

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

Neil Bothwick wrote:
> On Wed, 25 Nov 2020 09:13:07 -0600, Dale wrote:
>
>> I have /boot on a plain ext2 partition, root is also on a
>> plain ext4 partition.  Everything else, /home, /usr, /var etc is on
>> LVM.
>> If I hadn't had a separate /usr, I would have had to move things around
>> to grow /usr.  I've done that in the past and got very tired of doing it
>> the hard way.  With LVM, it's just a few commands and is done while in
>> use even.  I don't even have to logout, reboot or anything.  That's a
>> very good reason for having /usr separate from /. 
> I'd say it's more a very good reason to put / on LVM too. I used to use a
> separate /usr but found no real benefit so I now leave it as part of /.
>
>


Well, if / is on LVM and /usr needs room, one can just grow / which
would increase /usr to, if it is on / and not separate.  At the time, I
wasn't comfortable putting / or /boot on LVM.  I'm not sure it was
doable then.  I think it required more of the init thingy than I knew
how to deal with.  It sounds like it may be a lot easier now.  Come to
think of it, I think I was on the old grub back then.  Speaking of, can
I get rid of one of these or are both required?  If I can remove one,
which one?  I'm on the new grub and have been for a while.  I think I
uninstalled the old grub a long time ago. 


root@fireball / # du -shc /boot/grub*
34M     /boot/grub
6.9M    /boot/grub2
41M     total
root@fireball / #


If I can get rid of the plain grub, that would free up some space.  The
grub2 directory isn't as big but still wouldn't hurt. 

Dale

:-)  :-) 

Re: [gentoo-user] duplicate gentoo system - errors

[antlists]

On 25/11/2020 15:13, Dale wrote:
> I can't think of a reason not to use labels, at the very least, in most
> situations.  The only one I can think of, a laptop that has only one
> hard drive.  Sort of hard to install two hard drives on a laptop.  A
> external one can be done but never seen one with two spots for internal
> hard drives.  Do they make those???

I'm writing this on one of those right now ...

Windows on the first drive, and SUSE and gentoo on the second, except I 
can't get SUSE to realise I want the boot files on the first drive, so 
of course EFI can't find it to boot it.

(SUSE would put the boot files in the right place if I did an "expert 
partition" jobbie, but I don't want to do that seeing as I've never 
played with EFI before.)

Cheers,
Wol

Re: [gentoo-user] duplicate gentoo system - errors

[antlists]

On 25/11/2020 15:17, Rich Freeman wrote:
> On Wed, Nov 25, 2020 at 8:55 AM Wols Lists <antlists@youngman.org.uk> wrote:
>>
>> On 25/11/20 13:31, Rich Freeman wrote:
>>> Now, one area I would use UUIDs is with mdadm if you're not putting
>>> lvm on top.  I've seen mdadm arrays get renumbered and that is a mess
>>> if you're directly mounting them without labels or UUIDs.
>>
>> It is recommended to use names, so I call it by what it is, so I have
>> things like /dev/md/gentoo, /dev/md/home etc.
> 
> Is that supported with the original metadata format?  I suspect that
> was a big constraint since at the time my bootloader didn't support
> anything newer.
> 
Which format is this? version zero (i.e. just mdadm.conf), or 0.9 which 
is the kernel auto-assembly version. Either way, they're obsolete and 
bit-rotting.

I guess you do need version 1 (the difference between 1.0, 1.1 and 1.2 
is the location of the superblock, not the layout).

Cheers,
Wol

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

antlists wrote:
> On 25/11/2020 15:13, Dale wrote:
>> I can't think of a reason not to use labels, at the very least, in most
>> situations.  The only one I can think of, a laptop that has only one
>> hard drive.  Sort of hard to install two hard drives on a laptop.  A
>> external one can be done but never seen one with two spots for internal
>> hard drives.  Do they make those???
>
> I'm writing this on one of those right now ...
>
> Windows on the first drive, and SUSE and gentoo on the second, except
> I can't get SUSE to realise I want the boot files on the first drive,
> so of course EFI can't find it to boot it.
>
> (SUSE would put the boot files in the right place if I did an "expert
> partition" jobbie, but I don't want to do that seeing as I've never
> played with EFI before.)
>
> Cheers,
> Wol
>
>


First I've heard of a laptop having space for two hard drives.  I need
to make a note of that.  Now one has reason to use labels on laptops
too.  o_O 

Dale

:-)  :-) 

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 765 bytes --]

On Wed, 25 Nov 2020 10:16:28 -0600, Dale wrote:

> I think I was on the old grub back then.  Speaking of, can
> I get rid of one of these or are both required?  If I can remove one,
> which one?  I'm on the new grub and have been for a while.  I think I
> uninstalled the old grub a long time ago. 
> 
> 
> root@fireball / # du -shc /boot/grub*
> 34M     /boot/grub
> 6.9M    /boot/grub2
> 41M     total
> root@fireball / #
> 
> 
> If I can get rid of the plain grub, that would free up some space.  The
> grub2 directory isn't as big but still wouldn't hurt.

GRUB2 uses /boot/grub here, I suspect /boot/grub2 might be the surplus
one, but check the timestamps.


-- 
Neil Bothwick

Echo > Speak: "Whale oil beef hooked"

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 381 bytes --]

On Wed, 25 Nov 2020 11:55:56 -0600, Dale wrote:

> First I've heard of a laptop having space for two hard drives.  I need
> to make a note of that.  Now one has reason to use labels on laptops
> too.  o_O 

You already have. what if you boot with a flash drive connected and it is
recognised first?


-- 
Neil Bothwick

I am McCoy of Borg. He's assimilated, Jim!

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

Neil Bothwick wrote:
> On Wed, 25 Nov 2020 10:16:28 -0600, Dale wrote:
>
>> I think I was on the old grub back then.  Speaking of, can
>> I get rid of one of these or are both required?  If I can remove one,
>> which one?  I'm on the new grub and have been for a while.  I think I
>> uninstalled the old grub a long time ago. 
>>
>>
>> root@fireball / # du -shc /boot/grub*
>> 34M     /boot/grub
>> 6.9M    /boot/grub2
>> 41M     total
>> root@fireball / #
>>
>>
>> If I can get rid of the plain grub, that would free up some space.  The
>> grub2 directory isn't as big but still wouldn't hurt.
> GRUB2 uses /boot/grub here, I suspect /boot/grub2 might be the surplus
> one, but check the timestamps.
>
>


Well, grub2 shows the latest change.  Plain grub shows older changes. 
Most things in plain grub shows a date of April 2019.  Things in grub2
are 2013 except for grub.cfg which shows June 2020.  That is likely
about the time I rebuilt my last kernel, or somewhere close to that
anyway.  Sort of confusing. 

Just wondering if leaving that alone may be best.  ;-) 

Dale

:-)  :-) 

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

Neil Bothwick wrote:
> On Wed, 25 Nov 2020 11:55:56 -0600, Dale wrote:
>
>> First I've heard of a laptop having space for two hard drives.  I need
>> to make a note of that.  Now one has reason to use labels on laptops
>> too.  o_O 
> You already have. what if you boot with a flash drive connected and it is
> recognised first?
>
>

I wasn't counting a external device.  I was just referring to internal
hard drives.  I don't even put flash drives in the same category as a
hard drive either, even tho they can get large.  I guess SSDs are so
small they can fit two in a laptop nowadays.  I guess if a laptop is big
enough, one could fit a couple regular drives too.  Given displays are
so big now, it makes for more room.  I've just never seen or heard of
one before. 

I wouldn't mind having one of those in a way.  OS on one drive, SSD most
likely, and /home on its own drive.  Almost desktop-ish.  LOL 

Dale

:-)  :-) 

P. S.  No reply from our old Alan yet.  It hasn't bounced yet either. 

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 855 bytes --]

On Wed, 25 Nov 2020 13:30:46 -0600, Dale wrote:

> >> If I can get rid of the plain grub, that would free up some space.
> >> The grub2 directory isn't as big but still wouldn't hurt.  
> > GRUB2 uses /boot/grub here, I suspect /boot/grub2 might be the surplus
> > one, but check the timestamps.
> >
> >  
> 
> 
> Well, grub2 shows the latest change.  Plain grub shows older changes. 
> Most things in plain grub shows a date of April 2019.  Things in grub2
> are 2013 except for grub.cfg which shows June 2020.  That is likely
> about the time I rebuilt my last kernel, or somewhere close to that
> anyway.  Sort of confusing. 
> 
> Just wondering if leaving that alone may be best.  ;-) 

Rename one of the directories and see if you can still boot :)


-- 
Neil Bothwick

I'm moving to theory.  Everything works there!

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[gentoo-user] Re: duplicate gentoo system - errors

[Grant Edwards]

On 2020-11-25, Neil Bothwick <neil@digimed.co.uk> wrote:
> On Wed, 25 Nov 2020 13:30:46 -0600, Dale wrote:
>
>> >> If I can get rid of the plain grub, that would free up some space.
>> >> The grub2 directory isn't as big but still wouldn't hurt.  
>> > GRUB2 uses /boot/grub here, I suspect /boot/grub2 might be the surplus
>> > one, but check the timestamps.
>> 
>> Well, grub2 shows the latest change.  Plain grub shows older changes. 
>> Most things in plain grub shows a date of April 2019.  Things in grub2
>> are 2013 except for grub.cfg which shows June 2020.  That is likely
>> about the time I rebuilt my last kernel, or somewhere close to that
>> anyway.  Sort of confusing. 
>> 
>> Just wondering if leaving that alone may be best.  ;-) 
>
> Rename one of the directories and see if you can still boot :)

That may not be a valid test. If grub is using a blocklist to locate
secondary files, renaming the directory that contains those files
won't bother grub at all. Even rm'ing the files and directory might
not cause problems until the disk blocks of interest get reused by new
files.

--
Grant

Re: [gentoo-user] Re: duplicate gentoo system - errors

[Dale]

Grant Edwards wrote:
> On 2020-11-25, Neil Bothwick <neil@digimed.co.uk> wrote:
>> On Wed, 25 Nov 2020 13:30:46 -0600, Dale wrote:
>>
>>>>> If I can get rid of the plain grub, that would free up some space.
>>>>> The grub2 directory isn't as big but still wouldn't hurt.  
>>>> GRUB2 uses /boot/grub here, I suspect /boot/grub2 might be the surplus
>>>> one, but check the timestamps.
>>> Well, grub2 shows the latest change.  Plain grub shows older changes. 
>>> Most things in plain grub shows a date of April 2019.  Things in grub2
>>> are 2013 except for grub.cfg which shows June 2020.  That is likely
>>> about the time I rebuilt my last kernel, or somewhere close to that
>>> anyway.  Sort of confusing. 
>>>
>>> Just wondering if leaving that alone may be best.  ;-) 
>> Rename one of the directories and see if you can still boot :)
> That may not be a valid test. If grub is using a blocklist to locate
> secondary files, renaming the directory that contains those files
> won't bother grub at all. Even rm'ing the files and directory might
> not cause problems until the disk blocks of interest get reused by new
> files.
>
> --
> Grant

Great.  I don't know if I can remove them and boot normal or not.  :/  I
guess I'll leave them for now.  Maybe I can create a chroot and see what
it looks like after I install grub2???  I got equery to list the files
grub installs but it doesn't show anything being put in /boot.  I
wonder, if I remove all but grub.cfg, would installing it again, like
after a kernel upgrade, would install what is needed again??? 

Dale

:-)  :-) 

Re: [gentoo-user] duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 780 bytes --]

On Wed, 25 Nov 2020 13:37:48 -0600, Dale wrote:

> >> First I've heard of a laptop having space for two hard drives.  I
> >> need to make a note of that.  Now one has reason to use labels on
> >> laptops too.  o_O   
> > You already have. what if you boot with a flash drive connected and
> > it is recognised first?
> 
> I wasn't counting a external device.  I was just referring to internal
> hard drives.  I don't even put flash drives in the same category as a
> hard drive either, even tho they can get large.

Large in capacity but physically small. It's easy to reboot and not
realise you have a flash drive connected, which could possibly mess up
your drive naming.


-- 
Neil Bothwick

Men who have playful kittens shouldn't sleep in the nude.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] Re: duplicate gentoo system - errors

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 561 bytes --]

On Wed, 25 Nov 2020 19:43:04 -0000 (UTC), Grant Edwards wrote:

> > Rename one of the directories and see if you can still boot :)  
> 
> That may not be a valid test. If grub is using a blocklist to locate
> secondary files, renaming the directory that contains those files
> won't bother grub at all. Even rm'ing the files and directory might
> not cause problems until the disk blocks of interest get reused by new
> files.

Good point. I'm so glad I rarely have to use GRUB these days.


-- 
Neil Bothwick

Don't judge a book by its movie.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] duplicate gentoo system - errors

[antlists]

On 25/11/2020 22:59, Neil Bothwick wrote:
> On Wed, 25 Nov 2020 13:37:48 -0600, Dale wrote:
> 
>>>> First I've heard of a laptop having space for two hard drives.  I
>>>> need to make a note of that.  Now one has reason to use labels on
>>>> laptops too.  o_O
>>> You already have. what if you boot with a flash drive connected and
>>> it is recognised first?
>>
>> I wasn't counting a external device.  I was just referring to internal
>> hard drives.  I don't even put flash drives in the same category as a
>> hard drive either, even tho they can get large.
> 
> Large in capacity but physically small. It's easy to reboot and not
> realise you have a flash drive connected, which could possibly mess up
> your drive naming.
> 
Yup. I've forgotten which system it was - possibly this one - but I've 
got a system which will refuse to boot if I forget I've got a USB stick 
in it ...

Cheers,
Wol

Re: [gentoo-user] duplicate gentoo system - errors

[Dale]

antlists wrote:
> On 25/11/2020 22:59, Neil Bothwick wrote:
>> On Wed, 25 Nov 2020 13:37:48 -0600, Dale wrote:
>>
>>>>> First I've heard of a laptop having space for two hard drives.  I
>>>>> need to make a note of that.  Now one has reason to use labels on
>>>>> laptops too.  o_O
>>>> You already have. what if you boot with a flash drive connected and
>>>> it is recognised first?
>>>
>>> I wasn't counting a external device.  I was just referring to internal
>>> hard drives.  I don't even put flash drives in the same category as a
>>> hard drive either, even tho they can get large.
>>
>> Large in capacity but physically small. It's easy to reboot and not
>> realise you have a flash drive connected, which could possibly mess up
>> your drive naming.
>>
> Yup. I've forgotten which system it was - possibly this one - but I've
> got a system which will refuse to boot if I forget I've got a USB
> stick in it ...
>
> Cheers,
> Wol
>
>


I actually have mine set to look for a DVD first, USB stick second and
then for my 1st hard drive.  Thing is, I know not to have those things
there when I boot.  Of course, I don't boot/reboot very often.  When I
do, it's usually a power failure.  I guess because I've never did that,
except when I need to boot that way, I never thought a person would have
one plugged in that way. 

Of course, I'm a bit weird.  Don't bother arguing with me on that.  ROFL

Dale

:-)  :-) 

[gentoo-user] Re: duplicate gentoo system - errors

[Grant Edwards]

On 2020-11-25, Neil Bothwick <neil@digimed.co.uk> wrote:
> On Wed, 25 Nov 2020 19:43:04 -0000 (UTC), Grant Edwards wrote:
>
>> > Rename one of the directories and see if you can still boot :)  
>> 
>> That may not be a valid test. If grub is using a blocklist to locate
>> secondary files, renaming the directory that contains those files
>> won't bother grub at all. Even rm'ing the files and directory might
>> not cause problems until the disk blocks of interest get reused by new
>> files.
>
> Good point. I'm so glad I rarely have to use GRUB these days.

What I don't know anymore is in which situations grub2 uses a
blocklist and for what files. If you install grub2 in a partition
(instead of MBR), I'm pretty sure it does. There are file locking
mechanisms in some filesystems that you can use to make sure the
blocks for the critical file(s) don't change. I've got some notes
somewhere on that...

IIRC, legacy grub used blocklists in most installation cases.

There might be situations with grub2 where it doesn't have to use a
blocklist to find files within the linux /boot/grub directory.
Perhaps an MBR installation with plenty of empty space between the MBR
and first partition? Or when /boot is a FAT partition?

--
Grant