Re: Further to my installation error

[Pascal Hambourg]

Gene Heskett a écrit :
> On Friday 25 March 2016 15:48:56 Pascal Hambourg wrote:
> 
>> Le 20/03/2016 19:58, Gene Heskett a écrit :
>>> On Sunday 20 March 2016 14:39:41 Pascal Hambourg wrote:
 Le 20/03/2016 17:56, Gene Heskett a écrit :
> Now, I do note that sdb has a re-allocated sector count of 25, but
> no clue as to how many spares are left
 You can estimate it from the initial and current normalized values
 of the attribute.

 Remaining = raw value * norm. value / (init. norm. value - norm.
 value)
(...)
> ID# ATTRIBUTE_NAME  FLAG VALUE WORST THRESH TYPE  UPDATED  
> WHEN_FAILED RAW_VALUE
(...)
>   5 Reallocated_Sector_Ct   0x0033   100   100   036Pre-fail  Always  
>  -   25
(...)
> On the drive which does have some "re-allocated sectors", and I read that
> 24 of 36 as getting uncomfortably close to the far end of the bathtub
> curve. I should be acquiring a fresh drive to assure continuity,

I think you're misinterpreting the numbers.

25 is the raw value (RAW_VALUE), the actual reallocated sector count.
36 is the threshold (THRESH) of the normalized value, not of the raw value.
The normalized value (VALUE) is 100.

The normalized value decreases towards 0 as the raw value increases. As
long as the normalized value remains above the threshold, things are
supposed to be fine. However thresholds are somewhat arbitrary.

Here we can see (also from other attributes) that :
- The initial normalized value is 100. This may vary among vendors and
models.
- The threshold is reached when the remaining reserved sector count has
decreased to 36% (threshold / initial value) of the initial count,
whatever it is.
- The current normalized value is still the initial value, so less than
1% of the reserved sector count has be consumed. However we cannot
estimate the total reserved sector count, only that it must be at least
2500 (25 / 1%).

Do you feel more comfortable now ?

Re: Further to my installation error

[Gene Heskett]

On Friday 25 March 2016 15:48:56 Pascal Hambourg wrote:

> Le 20/03/2016 19:58, Gene Heskett a écrit :
> > On Sunday 20 March 2016 14:39:41 Pascal Hambourg wrote:
> >> Le 20/03/2016 17:56, Gene Heskett a écrit :
> >>> Now, I do note that sdb has a re-allocated sector count of 25, but
> >>> no clue as to how many spares are left
> >>
> >> You can estimate it from the initial and current normalized values
> >> of the attribute.
> >>
> >> Remaining = raw value * norm. value / (init. norm. value - norm.
> >> value)
> >
> > But smartctl isn't spitting that data out.
>
> What do you mean ? Doesn't 'smartctl -A' display the normalized and
> raw valued of the attributes ?

It may, on newer drives, but not on these.

> >  It is a 1 terrabyte drive, is there a reserved percentage?
>
> AFAIK there is no standard about the reserved sector count, if this is
> what you want to know.

Thats what I was asking for, a way to tease that out.
ID# ATTRIBUTE_NAME  FLAG VALUE WORST THRESH TYPE  UPDATED  
WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate 0x000f   114   099   006Pre-fail  Always   
-   64516572
  3 Spin_Up_Time0x0003   100   100   000Pre-fail  Always   
-   0
  4 Start_Stop_Count0x0032   100   100   020Old_age   Always   
-   375
  5 Reallocated_Sector_Ct   0x0033   100   100   036Pre-fail  Always   
-   25
  7 Seek_Error_Rate 0x000f   081   060   030Pre-fail  Always   
-   147509778
  9 Power_On_Hours  0x0032   039   039   000Old_age   Always   
-   53804
 10 Spin_Retry_Count0x0013   100   100   097Pre-fail  Always   
-   0
 12 Power_Cycle_Count   0x0032   100   100   020Old_age   Always   
-   380
184 End-to-End_Error0x0032   100   100   099Old_age   Always   
-   0
187 Reported_Uncorrect  0x0032   100   100   000Old_age   Always   
-   0
188 Command_Timeout 0x0032   100   099   000Old_age   Always   
-   4
189 High_Fly_Writes 0x003a   001   001   000Old_age   Always   
-   291
190 Airflow_Temperature_Cel 0x0022   063   057   045Old_age   Always   
-   37 (Min/Max 26/41)
194 Temperature_Celsius 0x0022   037   043   000Old_age   Always   
-   37 (0 18 0 0 0)
195 Hardware_ECC_Recovered  0x001a   025   022   000Old_age   Always   
-   64516572
197 Current_Pending_Sector  0x0012   100   100   000Old_age   Always   
-   0
198 Offline_Uncorrectable   0x0010   100   100   000Old_age   Offline  
-   0
199 UDMA_CRC_Error_Count0x003e   200   200   000Old_age   Always   
-   0
240 Head_Flying_Hours   0x   100   253   000Old_age   Offline  
-   53800 (168 12 0)
241 Total_LBAs_Written  0x   100   253   000Old_age   Offline  
-   3559232973
242 Total_LBAs_Read 0x   100   253   000Old_age   Offline  
-   3031833807

SMART Error Log Version: 1
No Errors Logged

On the drive which does have some "re-allocated sectors", and I read that
24 of 36 as getting uncomfortably close to the far end of the bathtub
curve. I should be acquiring a fresh drive to assure continuity,

Thanks.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Further to my installation error

[Pascal Hambourg]

Le 20/03/2016 19:58, Gene Heskett a écrit :

On Sunday 20 March 2016 14:39:41 Pascal Hambourg wrote:


Le 20/03/2016 17:56, Gene Heskett a écrit :

Now, I do note that sdb has a re-allocated sector count of 25, but
no clue as to how many spares are left


You can estimate it from the initial and current normalized values of
the attribute.

Remaining = raw value * norm. value / (init. norm. value - norm.
value)


But smartctl isn't spitting that data out.


What do you mean ? Doesn't 'smartctl -A' display the normalized and raw 
valued of the attributes ?



 It is a 1 terrabyte drive, is there a reserved percentage?


AFAIK there is no standard about the reserved sector count, if this is 
what you want to know.

Re: Further to my installation error

[Pascal Hambourg]

Adam Wilson a écrit :
> On Sun, 20 Mar 2016 12:51:56 +0100
> Pascal Hambourg  wrote:
>> Adam Wilson a écrit :
> 
>>> USB can still
>>> only be booted from UEFI, but newer d-i means that installation now
>>> proceeds as normal until the point at which UEFI yes/no
>>> (force/leave) selection is reached  
>>
>> I don't remember seeing this option in Jessie's installer when booted
>> in EFI mode. How is it labeled exactly ? At what stage is it
>> proposed ?
(...)
> It is in between the tasksel stage and the GRUB installation stage. A
> prompt appeared telling me that I could either 'Force UEFI' for the
> operating system installation, and have Jessie configured to boot from
> UEFI, or not. It detected the fact that I had booted from UEFI but my
> system was capable of either booting method.

Eventually I was able to trigger the prompt. Thanks for the information.

It actually happens when partman (the partition tool) starts, if an
already installed system which boots from legacy mode is detected.

I think I never met this situation during my previous installations ;
either there was no other system installed, either the other system
booted from EFI mode too. This time by chance there was a previous
Debian installation in legacy mode.

Re: Further to my installation error

[Lisi Reisz]

On Monday 21 March 2016 05:02:12 David Christensen wrote:
> On 03/20/2016 04:00 PM, Lisi Reisz wrote:
> > It had one - on sda.  It wanted the one on sdb as well.  I wanted the one
> > on sdb left alone.  I got what I wanted when I insisted, and the
> > resultant installation was quite happy just using the one disk.  And the
> > swap partition on sdb kept its UUID intact.  So all is well that ends
> > well.
>
> When installing Debian or any other O/S, I've learned the hard way that
> it is best to unplug all the drives except the drive(s) that I'm
> installing onto.  Sometimes, this means additional work after
> installation; but, "better safe than sorry".


When I could see I would just 
have disconnected sdb.  It would then have been left alone.

Re: Further to my installation error

[David Christensen]

On 03/20/2016 04:00 PM, Lisi Reisz wrote:

It had one - on sda.  It wanted the one on sdb as well.  I wanted the one on
sdb left alone.  I got what I wanted when I insisted, and the resultant
installation was quite happy just using the one disk.  And the swap partition
on sdb kept its UUID intact.  So all is well that ends well.


When installing Debian or any other O/S, I've learned the hard way that 
it is best to unplug all the drives except the drive(s) that I'm 
installing onto.  Sometimes, this means additional work after 
installation; but, "better safe than sorry".



David

Re: Further to my installation error

[Lisi Reisz]

On Sunday 20 March 2016 22:00:01 David Wright wrote:
> On Sun 20 Mar 2016 at 11:54:20 (+0300), Adam Wilson wrote:
> > On Fri, 18 Mar 2016 22:21:58 +
> >
> > Lisi Reisz  wrote:
> > > On Friday 18 March 2016 20:49:55 David Wright wrote:
> > > > It's far more likely that you forgot to format the partition, if
> > > > that's indeed what you wanted to do.
> > >
> > > No.  I checked and double checked that the partitions on the disk
> > > which I wanted to use for installation were all marked with the F for
> > > format, and that nothing on the disk it had been told to leave alone
> > > had an F.  It kept wanting to format the spare disk's swap, which I
> > > did not want.
> >
> > Why not? You wanted to carry over the swap created by a previous
> > installation?
>
> Neither grub nor the installer handle partition labels very well.
> If you've labelled your swap partition, letting the installer
> loose on it means you have to restore the label each time.
>
> The d-i does want to format it, and it's fairly insistent that it at
> least be given one to use, but it can be dissuaded from both.

It had one - on sda.  It wanted the one on sdb as well.  I wanted the one on 
sdb left alone.  I got what I wanted when I insisted, and the resultant 
installation was quite happy just using the one disk.  And the swap partition 
on sdb kept its UUID intact.  So all is well that ends well.

Lisi

> The documentation also tells you not to
> # mkswap -L abcd00 /dev/sdz9
> # swapon /dev/sdz9
> after the partitioner has run but before the base installation, but
> I've had no problems doing that in the past. (Remember installing
> linux on a Texas Travelmate with 8MB memory...)
>
> All this may be different for people encrypting swap or using
> containers etc.
>
> Cheers,
> David.

Re: Further to my installation error

[Lisi Reisz]

On Sunday 20 March 2016 22:15:16 David Wright wrote:
> On Sun 20 Mar 2016 at 11:53:43 (-0400), Gene Heskett wrote:
> > The problem with that is that unless you edit fstab before the initial
> > reboot, (difficult to do)
>
> The d-i has a shell available. The file is /target/etc/fstab so
> no difficulty there.
>
> What some people *might* find difficult is fiddling with the home
> directory before the inital reboot because it's created so late
> in the process. However, there is a trick: when you are asked
> whether the clock is on UTC, press GoBack at that point, because
> the home directory has just been created (on /target of course).


Thank you for a very useful tip!

Lisi

Re: Further to my installation error

[David Wright]

On Sun 20 Mar 2016 at 11:53:43 (-0400), Gene Heskett wrote:

> The problem with that is that unless you edit fstab before the initial 
> reboot, (difficult to do)

The d-i has a shell available. The file is /target/etc/fstab so
no difficulty there.

What some people *might* find difficult is fiddling with the home
directory before the inital reboot because it's created so late
in the process. However, there is a trick: when you are asked
whether the clock is on UTC, press GoBack at that point, because
the home directory has just been created (on /target of course).

Cheers,
David.

Re: Further to my installation error

[David Wright]

On Sun 20 Mar 2016 at 11:54:20 (+0300), Adam Wilson wrote:
> On Fri, 18 Mar 2016 22:21:58 +
> Lisi Reisz  wrote:
> 
> > On Friday 18 March 2016 20:49:55 David Wright wrote:
> > > It's far more likely that you forgot to format the partition, if
> > > that's indeed what you wanted to do.  
> > 
> > No.  I checked and double checked that the partitions on the disk
> > which I wanted to use for installation were all marked with the F for
> > format, and that nothing on the disk it had been told to leave alone
> > had an F.  It kept wanting to format the spare disk's swap, which I
> > did not want.
> 
> Why not? You wanted to carry over the swap created by a previous
> installation?

Neither grub nor the installer handle partition labels very well.
If you've labelled your swap partition, letting the installer
loose on it means you have to restore the label each time.

The d-i does want to format it, and it's fairly insistent that it at
least be given one to use, but it can be dissuaded from both.
The documentation also tells you not to
# mkswap -L abcd00 /dev/sdz9
# swapon /dev/sdz9
after the partitioner has run but before the base installation, but
I've had no problems doing that in the past. (Remember installing
linux on a Texas Travelmate with 8MB memory...)

All this may be different for people encrypting swap or using
containers etc.

Cheers,
David.

Re: Further to my installation error

[Lisi Reisz]

On Sunday 20 March 2016 19:15:51 Brian wrote:
> Being under pressure to complete the task set doesn't lend itself to
> considered action and burning a CD can be fraught.

Granted.

> The quality of the 
> disk, the burning and the accuracy of reading the burnt disk are things
> to take into account. I'd have booted the installer from something
> completely different (such as a USB stick) to get a better picture of
> the situation. A repeat performance would have justified alarm bells
> ringing.
>
> Success eventually came; probably because the writing and reading
> processes of the CD-ROM acted correctly. Coming after using GParted was
> likely a coincidence; other changes might have been factors. 

Again, Granted.

> There is 
> nothing to underpin the advice not to use d-i's partitioning methods.
> (As an aside, d-i has fdisk; there was no need to use GParted).

I have always so far used d-i's partitioner.  I wasn't really meaning that as 
permanent standing advice.  But there may have been an odd bug in that 
version of the Jessie d-i.

If you diagnose PEBKAC I shall not disagree.  I shall probably use the d-i 
partitioner next time myself!

Lisi

Re: Further to my installation error

[Brian]

On Sun 20 Mar 2016 at 12:02:14 +, Lisi Reisz wrote:

> On Saturday 19 March 2016 01:27:33 David Wright wrote:
> >
> > At this point, you also mentioned repartitioning, reformatting and
> > wiping a disk thoroughly.
> >
> > In the last post (which I commented on when the second paragraph was
> > quoted in Adam Wilson's post) you said that trying to use CD1 to do a
> > net install made you realise that "if you want to format an existing
> > partition properly, so the stuff on it actually goes, use Gparted not
> > the partitioner in the Jessie installer."
> >
> > Maybe I'm thick, but I can't see why the Debian installer's performance
> > at re-partition/format/wipe would cause the reported download
> > problems, checksum failures, corrupt files and old firmware.

I'd agree with that. Unless this stage is completed satisfactorily it is
highly unlikely the installer will progress to the next stage, getting
the base system.
 
> No, of course not.  But continuing to use the original files, from the 
> original failed installation, because they were still there, might, might it 
> not??  I am open to suggestions.

The original message from the installer was:

  Warning:  file:///cdrom/pool/main/g/gnupg/gnupg_1.4.18-7_amd64.deb was corrupt

The base system was being installed by debootstrap from the CD, not from
any external mirror.

Being under pressure to complete the task set doesn't lend itself to
considered action and burning a CD can be fraught. The quality of the
disk, the burning and the accuracy of reading the burnt disk are things
to take into account. I'd have booted the installer from something
completely different (such as a USB stick) to get a better picture of
the situation. A repeat performance would have justified alarm bells
ringing.

Success eventually came; probably because the writing and reading
processes of the CD-ROM acted correctly. Coming after using GParted was
likely a coincidence; other changes might have been factors. There is
nothing to underpin the advice not to use d-i's partitioning methods.
(As an aside, d-i has fdisk; there was no need to use GParted).

> > Sorry. 
> > I'm no great fan of the d-i partitioner, but I haven't had it fail to
> > do what it promised.
> 
> Previously nor have I.  But, after the first time, it did not appear to be 
> formatting anything, and in the end the disk did not appear to be formatted.

It was impossible for formatting not to have happened. The next stage
would not have run and a big red screen with "No root file system
defined" would have been displayed.

> Note appear.  And it was by then very late and I was tired, and am not beyond 
> making mistakes of interpretation at the best of times.  For whatever reason, 
> when I formatted with GParted, something appeared to be happening, and the 
> result appeared to be successful.

Linking the use of GParted with success gives the impression it was the
cause of the success. It probably wasn't.

> > (Not that I understand precisely the term "wipe".) 
> 
> Sorry.  It is not a precise term.  Reformat thoroughly I suppose I meant.  
> 
> I am about to install again on another computer.  Unfortunately, I can't 
> reuse 
> the original CD because I destroyed it when I thought it was faulty.  I would 
> have liked to reuse it to see what happened!

99.9% of used CDs go to landfill after a single use. I've contributed my
share. It is unknown for a USB stick to end up in the same place after
one Debian image has been put on it.

A single CD is indeed cheaper and easier to label. It is possibly easier
to store (but not in your pocket), Parting with a USB stick is less easy
than giving away a CD, (unless the giver is the generous type).

Flexibility? About 5GB storage with the cheap DVDs. Compared with.

Rewritable? Maybe.

Can go through a complete cycle in an automatic washing machine wihout
harm? Needs testing. :)

Re: Further to my installation error

[Gene Heskett]

On Sunday 20 March 2016 14:39:41 Pascal Hambourg wrote:

> Le 20/03/2016 17:56, Gene Heskett a écrit :
> > Now, I do note that sdb has a re-allocated sector count of 25, but
> > no clue as to how many spares are left
>
> You can estimate it from the initial and current normalized values of
> the attribute.
>
> Remaining = raw value * norm. value / (init. norm. value - norm.
> value)

But smartctl isn't spitting that data out.  It is a 1 terrabyte drive, is 
there a reserved percentage?

Thanks.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Further to my installation error

[Pascal Hambourg]

Le 20/03/2016 17:56, Gene Heskett a écrit :

Now, I do note that sdb has a re-allocated sector count of 25, but no
clue as to how many spares are left


You can estimate it from the initial and current normalized values of 
the attribute.


Remaining = raw value * norm. value / (init. norm. value - norm. value)

Re: Further to my installation error

[Lisi Reisz]

On Sunday 20 March 2016 15:53:43 Gene Heskett wrote:
> On Sunday 20 March 2016 07:47:02 Lisi Reisz wrote:
> > On Sunday 20 March 2016 09:34:53 Gene Heskett wrote:
> > > On Sunday 20 March 2016 04:54:20 Adam Wilson wrote:
> > > > On Fri, 18 Mar 2016 22:21:58 +
> > > >
> > > > Lisi Reisz  wrote:
> > > > > On Friday 18 March 2016 20:49:55 David Wright wrote:
> > > > > > It's far more likely that you forgot to format the partition,
> > > > > > if that's indeed what you wanted to do.
> > > > >
> > > > > No.  I checked and double checked that the partitions on the
> > > > > disk which I wanted to use for installation were all marked with
> > > > > the F for format, and that nothing on the disk it had been told
> > > > > to leave alone had an F.  It kept wanting to format the spare
> > > > > disk's swap, which I did not want.
> > > >
> > > > Why not? You wanted to carry over the swap created by a previous
> > > > installation?
> > >
> > > Doing that, leaving a potentially dirty swap for a new install?  No
> > > sensible reason to do so, format that puppy.
> >
> > Am I the only person on this list who has ever wanted to install on
> > one disk and leave another alone for some reason? Surely not!!  I
> > wanted to install on sda and leave sdb alone.  So I told it not to use
> > sdb.  Not to format sdb. Not to touch sdb.  Why?  Because I didn't
> > want sdb touched.  I was not leaving a potentially dirty swap for a
> > new install.  I was telling the new install not to use the second
> > disk.  Sheesh.  When I could see I would just have disconnected sdb.
> > It would then have been left alone.  OK.
> >
> > I regard that as a sensible reason.
> >
> > Lisi
>
> The problem with that is that unless you edit fstab before the initial
> reboot, (difficult to do) to remove the auto-found swap partition on
> sdb, it will be found and mounted. But by then it likely has been
> formatted anyway.

I didn't and it didn't.

Lisi

Re: Further to my installation error

[Gene Heskett]

On Sunday 20 March 2016 09:45:36 Felix Miata wrote:

> Gene Heskett composed on 2016-03-20 05:34 (UTC-0400):
> > Adam Wilson wrote:
> >> On Fri, 18 Mar 2016 22:21:58 + Lisi Reisz wrote:
> >> > No.  I checked and double checked that the partitions on the disk
> >> > which I wanted to use for installation were all marked with the F
> >> > for format, and that nothing on the disk it had been told to
> >> > leave alone had an F.  It kept wanting to format the spare disk's
> >> > swap, which I did not want.
> >>
> >> Why not? You wanted to carry over the swap created by a previous
> >> installation?
> >
> > Doing that, leaving a potentially dirty swap for a new install?  No
> > sensible reason to do so, format that puppy.
>
> In multiboot, re-formatting swap is a problem creator. Mounting is
> generally via UUID. Formatting changes UUID. Missing UUID at boot
> usually means booting into rescue mode to repair fstab. So at best,
> assuming one swap per system and not mounting by device name, one
> installation *might be* benefited to the (probable) detriment of only
> one installation. In all other cases in which a unique swap per
> installation is not used (which never happens here), and swap mounting
> is not by device name (here, swap is often by device name), one
> installation might be benefited, to the detriment of multiple
> installations.

Faced with a similar problem as Lisi's during this install as I had 4 
drives in it then and 3 swaps mounted, I did remove the other drives, so 
my fstab is by UUID for everything I am using. Memory is now faint, but 
ISTR, when I first rebooted after plugging in the drive amanda uses as a 
tape vault, I ran blkid and used the UUID to add that drive to fstab.

When I do a new install, I generally do it on a new drive, but its been 
in the machine long enough to go get any firmware updates it needs and 
reflash them. Every drive I've bought locally or from NewEgg or Tiger 
Direct in the last 5 years has had old, buggy, sometimes slow firmware 
in it, so they all will probably need reflashed for using them for work.

I might add that my drive failure rate since I started doing that has 
fallen off a cliff.

sda says 23,000+ hours
sdb says 58,000+ hours

Now, I do note that sdb has a re-allocated sector count of 25, but no 
clue as to how many spares are left but that does sound like its time to 
retire it.  Both are older 1 terrabyte drives that I planned on 
replacing this summer. The newer crop of drives seems to be maturing, 
but I also expect the newer drives will be 4 kilobyte sectors too. 
Hopefully by then our disk tools will have caught up and can 
transparently use both GPT partitions AND 4Kb sectors.

The current boot drive is a 4Kb sector drive, and I frankly had one hell 
of a time using the installers disk tools, getting it properly aligned 
so its read/write speed was at least usable.

Thanks Felix & everybody that has read this far.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Further to my installation error

[Gene Heskett]

On Sunday 20 March 2016 07:47:02 Lisi Reisz wrote:

> On Sunday 20 March 2016 09:34:53 Gene Heskett wrote:
> > On Sunday 20 March 2016 04:54:20 Adam Wilson wrote:
> > > On Fri, 18 Mar 2016 22:21:58 +
> > >
> > > Lisi Reisz  wrote:
> > > > On Friday 18 March 2016 20:49:55 David Wright wrote:
> > > > > It's far more likely that you forgot to format the partition,
> > > > > if that's indeed what you wanted to do.
> > > >
> > > > No.  I checked and double checked that the partitions on the
> > > > disk which I wanted to use for installation were all marked with
> > > > the F for format, and that nothing on the disk it had been told
> > > > to leave alone had an F.  It kept wanting to format the spare
> > > > disk's swap, which I did not want.
> > >
> > > Why not? You wanted to carry over the swap created by a previous
> > > installation?
> >
> > Doing that, leaving a potentially dirty swap for a new install?  No
> > sensible reason to do so, format that puppy.
>
> Am I the only person on this list who has ever wanted to install on
> one disk and leave another alone for some reason? Surely not!!  I
> wanted to install on sda and leave sdb alone.  So I told it not to use
> sdb.  Not to format sdb. Not to touch sdb.  Why?  Because I didn't
> want sdb touched.  I was not leaving a potentially dirty swap for a
> new install.  I was telling the new install not to use the second
> disk.  Sheesh.  When I could see I would just have disconnected sdb. 
> It would then have been left alone.  OK.
>
> I regard that as a sensible reason.
>
> Lisi

The problem with that is that unless you edit fstab before the initial 
reboot, (difficult to do) to remove the auto-found swap partition on 
sdb, it will be found and mounted. But by then it likely has been 
formatted anyway.

I think my point is that its a shrug, or should be.

If you want separate swaps because you are dual booting, that isolation 
can be handled by a root edit session on /etc/fstab of each install.  
And it may sound counterintuitive, but if the install on sda uses /swap 
on /sdb, and the install on sdb used the swap on sda, the machine may be 
noticeably faster when its actually useing swap.  OTOH. I have never 
done it. I just think its possible with no side effects other than a bit 
faster when its using swap when booted to either install.

Unless your machines are really memory starved, swap s/b clean anyway if 
its not been booted long.   I'd also expect that doing, as root, 
swapoff -aRETURN should result in a clean swap being found at the reboot 
you are using to install.  Then it wouldn't matter if it was 
reformatted.  I have never believed in having anything in swap actually 
surviving a reboot anyway. At 12 days & change uptime, I show 58 megs of 
swap in use.  So this 3.16.0 kernel I'm running now is not particularly 
swappy, but it also recognizes the 8Gb of dram in this machine.

OTOH, each machine seems to have a mind of its own. So my advice may be 
for naught.

Linux it seems to me is entirely too eager to outwit the experienced 
user.  But my oar is too small to fit the oarlocks in this ship called 
linux.  Linus may have had a very good reason for making it do what it 
does.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Further to my installation error

[Pascal Hambourg]

Adam Wilson a écrit :
> On Sun, 20 Mar 2016 12:51:56 +0100
> Pascal Hambourg  wrote:
> 
>> Did you try to boot a non-EFI capable USB boot media, such as a Debian
>> live image or a Debian installation image with the EFI partition
>> deleted ?

(I guess that just changing the EFI partition type ID to "00" (empty)
should  do the trick too).

> No. That did not occur to me; I assumed it would then refuse to boot at
> all. I shall try this out and get back to you.

Thanks.

> And I don't like / don't need either GPT, LVM, or logical volumes! Four
> primary partitions has always been enough for me.

Lucky you...

>> I have seen this display problem with the EFI framebuffer driver in
>> Wheezy's kernel on some machines. Installing with a serial console
>> worked.
> 
> How does one do that?

First, the computer must have a serial (RS232 UART) interface. I guess
it must be an internal one which may be quite rare these days, not a
USB-serial adapter. You must connect it to a serial console (with a
straight serial cable) or another computer (with a crossed serial cable)
running a serial console emulator such as putty, picocom... Finally you
must pass some parameters on the linux command line in GRUB. These are
mentionned in the installation manual.

There are a coupe of drawback, though : the installer through the serial
console is available only in black and english, and after the
installation, only the serial console is enabled. At least that was true
in Wheezy, and enabling the virtual consoles tty1-6 in /etc/inittab was
easy. I don't know about Jessie and systemd.

>>> USB can still
>>> only be booted from UEFI, but newer d-i means that installation now
>>> proceeds as normal until the point at which UEFI yes/no
>>> (force/leave) selection is reached  
>>
>> I don't remember seeing this option in Jessie's installer when booted
>> in EFI mode. How is it labeled exactly ? At what stage is it
>> proposed ?
> 
> It isn't something I saw either until my first UEFI installation media
> boot- it may have been triggered by the fact that I was using 'Launch
> CSM' Legacy mode so that I wouldn't have to bother with GPT/UEFI and
> could use something resembling traditional BIOS and MBR.

I don't remember seeing it even in my latest EFI installations.

> It is in between the tasksel stage and the GRUB installation stage. A
> prompt appeared telling me that I could either 'Force UEFI' for the
> operating system installation, and have Jessie configured to boot from
> UEFI, or not. It detected the fact that I had booted from UEFI but my
> system was capable of either booting method.

Thanks, I'll double check that during my next installation.

Re: Further to my installation error

[Adam Wilson]

On Sun, 20 Mar 2016 11:47:02 +
Lisi Reisz  wrote:

> On Sunday 20 March 2016 09:34:53 Gene Heskett wrote:
> > On Sunday 20 March 2016 04:54:20 Adam Wilson wrote:  
> > > On Fri, 18 Mar 2016 22:21:58 +
> > >
> > > Lisi Reisz  wrote:  
> > > > On Friday 18 March 2016 20:49:55 David Wright wrote:  
> > > > > It's far more likely that you forgot to format the partition,
> > > > > if that's indeed what you wanted to do.  
> > > >
> > > > No.  I checked and double checked that the partitions on the
> > > > disk which I wanted to use for installation were all marked
> > > > with the F for format, and that nothing on the disk it had been
> > > > told to leave alone had an F.  It kept wanting to format the
> > > > spare disk's swap, which I did not want.  
> > >
> > > Why not? You wanted to carry over the swap created by a previous
> > > installation?  
> >
> > Doing that, leaving a potentially dirty swap for a new install?  No
> > sensible reason to do so, format that puppy.  
> 
> Am I the only person on this list who has ever wanted to install on
> one disk and leave another alone for some reason? Surely not!!  I
> wanted to install on sda and leave sdb alone.  So I told it not to
> use sdb.  Not to format sdb. Not to touch sdb.  Why?  Because I
> didn't want sdb touched.  I was not leaving a potentially dirty swap
> for a new install.  I was telling the new install not to use the
> second disk.  Sheesh.  When I could see I would just have
> disconnected sdb.  It would then have been left alone.  OK.

What was on sdb?

And even if you didn't want the rest of sdb touched, why didn't you
simply direct the partitioner to format swap (which I assume was on
sdb), but nothing else?


pgpUpTIyNS_kw.pgp
Description: OpenPGP digital signature

Re: Further to my installation error

[Adam Wilson]

On Sun, 20 Mar 2016 12:51:56 +0100
Pascal Hambourg  wrote:

> Adam Wilson a écrit :
> > 
> > There is a potential problem also where modern machines (with UEFI
> > boot) fail to install off USB without UEFI, meaning that if you
> > want a UEFI-free install, you have to use optical media.  
> 
> I have a rather old (~2007) UEFI motherboard which does the opposite :
> its UEFI firmware lacks USB and AHCI drivers so it is impossible to
> boot in UEFI mode from a USB device or from a SATA drive in AHCI mode.
> 
> > This happened to me
> > on an ASUS UX51vz with UEFI- with Debian 7 d-i, inserting an optical
> > disc would provide the option to either boot the disk with UEFI or
> > without it,  
> 
> Was this option displayed in the regular firmware boot menu or in a
> different menu ?

The boot order menu in the UEFI settings.

> > while inserting a USB would only allow UEFI boot from the
> > flash drive, and not legacy booting, even with Launch CSM enabled.  
> 
> Do you mean the the firmware boot menu only displayed one option (EFI
> mode) for the USB device ?

Yes.

> Did you try to boot a non-EFI capable USB boot media, such as a Debian
> live image or a Debian installation image with the EFI partition
> deleted ?

No. That did not occur to me; I assumed it would then refuse to boot at
all. I shall try this out and get back to you.

> > So while using Debian 7, I just had to use optical media to install
> > Debian sans UEFI (which I always do- call it nostalgia, but I like
> > the simplicity of MBR and four primary partitions rather than the
> > unfamiliar layers of GPT/UEFI related cruft).  
> 
> Note that UEFI is not tied to GPT and vice versa. I happily use GPT on
> legacy systems when I need many partitions and LVM is not an option.
> Extended and logical partitions just suck.

And I don't like / don't need either GPT, LVM, or logical volumes! Four
primary partitions has always been enough for me.

> > Booting off USB (forced
> > to use UEFI) would simply result in a black screen after selecting
> > "Install".  
> 
> I have seen this display problem with the EFI framebuffer driver in
> Wheezy's kernel on some machines. Installing with a serial console
> worked.

How does one do that?

> > Debian 8, however, seems to have solved this problem.  
> 
> Indeed, the kernel EFI framebuffer driver in Jessie's driver seems to
> have been improved.

I wouldn't know anything about that, but from my field experience UEFI
works with far less hassle in Jessie as opposed to Wheezy.

> > USB can still
> > only be booted from UEFI, but newer d-i means that installation now
> > proceeds as normal until the point at which UEFI yes/no
> > (force/leave) selection is reached  
> 
> I don't remember seeing this option in Jessie's installer when booted
> in EFI mode. How is it labeled exactly ? At what stage is it
> proposed ?

It isn't something I saw either until my first UEFI installation media
boot- it may have been triggered by the fact that I was using 'Launch
CSM' Legacy mode so that I wouldn't have to bother with GPT/UEFI and
could use something resembling traditional BIOS and MBR.

It is in between the tasksel stage and the GRUB installation stage. A
prompt appeared telling me that I could either 'Force UEFI' for the
operating system installation, and have Jessie configured to boot from
UEFI, or not. It detected the fact that I had booted from UEFI but my
system was capable of either booting method.


pgpg_GgEyTTW7.pgp
Description: OpenPGP digital signature

Re: Further to my installation error

[Felix Miata]

Gene Heskett composed on 2016-03-20 05:34 (UTC-0400):


Adam Wilson wrote:



On Fri, 18 Mar 2016 22:21:58 + Lisi Reisz wrote:



> No.  I checked and double checked that the partitions on the disk
> which I wanted to use for installation were all marked with the F
> for format, and that nothing on the disk it had been told to leave
> alone had an F.  It kept wanting to format the spare disk's swap,
> which I did not want.



Why not? You wanted to carry over the swap created by a previous
installation?



Doing that, leaving a potentially dirty swap for a new install?  No
sensible reason to do so, format that puppy.


In multiboot, re-formatting swap is a problem creator. Mounting is generally 
via UUID. Formatting changes UUID. Missing UUID at boot usually means booting 
into rescue mode to repair fstab. So at best, assuming one swap per system 
and not mounting by device name, one installation *might be* benefited to the 
(probable) detriment of only one installation. In all other cases in which a 
unique swap per installation is not used (which never happens here), and swap 
mounting is not by device name (here, swap is often by device name), one 
installation might be benefited, to the detriment of multiple installations.

--
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

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

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

Re: Further to my installation error

[Lisi Reisz]

On Wednesday 16 March 2016 20:49:28 Brian wrote:
> But the vast majortiy of users are not using the ancient machines you
> and I have. The OP is at liberty to indicate whether her machine falls
> into this class. A round shiny disc could be her only solution to
> booting a Debian image.

In this case no.   But I have had several failures recently with USB sticks, 
including but not limited to one where the machine couldn't boot from most 
USB media.

And as Felix says, CDs are cheaper, easier to label, easier to store, easier 
to give away and generally more flexible IMHO.  But I also prefer 
non-electronic stoves.  Not that I can get one. :-(.

This one wasn't an ancient machine.  But I handle more ancient machines than 
modern ones.  I call  3 1/2 years old new, to establish what we mean by new 
and old.

Lisi

Re: Further to my installation error

[Lisi Reisz]

On Saturday 19 March 2016 01:27:33 David Wright wrote:
> On Fri 18 Mar 2016 at 22:21:58 (+), Lisi Reisz wrote:
> > On Friday 18 March 2016 20:49:55 David Wright wrote:
> > > It's far more likely that you forgot to format the partition, if
> > > that's indeed what you wanted to do.
> >
> > No.  I checked and double checked that the partitions on the disk which I
> > wanted to use for installation were all marked with the F for format, and
> > that nothing on the disk it had been told to leave alone had an F.  It
> > kept wanting to format the spare disk's swap, which I did not want.  So
> > whilst I certainly do not think that it did format properly, it was
> > clearly, explicitly and repeatedly (all three times) asked to do so.
> >
> > Since it was clearly important that the partitions should be reformatted,
> > it would have been remarkably stupid of me to have omitted to do so.  On
> > what basis are you accusing me of having forgotten?  That this particular
> > problem has not hit you?  No, it has not hit me before.
>
> Um, the quotation "It's far...to do." above was written in response
> to two paragraphs by Adam Wilson "Is it possible...if you will.", and
> had nothing to do with you or your problem/situation at all. The "you"
> in "you forgot" is Adam. So I won't comment further on what you've just
> written above.
>
> So why did I write "I can't see why Lisi came to that conclusion from
> what she reported here." Because I read the OP which AFAICT talked
> about corrupt files, checksums, and old firmware. Your next post said
> the crisis was over. Your next post (new thread) mentioned a download
> problem. The next post said you had the same problem as before.
>
> At this point, you also mentioned repartitioning, reformatting and
> wiping a disk thoroughly.
>
> In the last post (which I commented on when the second paragraph was
> quoted in Adam Wilson's post) you said that trying to use CD1 to do a
> net install made you realise that "if you want to format an existing
> partition properly, so the stuff on it actually goes, use Gparted not
> the partitioner in the Jessie installer."
>
> Maybe I'm thick, but I can't see why the Debian installer's performance
> at re-partition/format/wipe would cause the reported download
> problems, checksum failures, corrupt files and old firmware.

No, of course not.  But continuing to use the original files, from the 
original failed installation, because they were still there, might, might it 
not??  I am open to suggestions.

> Sorry. 
> I'm no great fan of the d-i partitioner, but I haven't had it fail to
> do what it promised.

Previously nor have I.  But, after the first time, it did not appear to be 
formatting anything, and in the end the disk did not appear to be formatted.  
Note appear.  And it was by then very late and I was tired, and am not beyond 
making mistakes of interpretation at the best of times.  For whatever reason, 
when I formatted with GParted, something appeared to be happening, and the 
result appeared to be successful.

> (Not that I understand precisely the term "wipe".) 

Sorry.  It is not a precise term.  Reformat thoroughly I suppose I meant.  

I am about to install again on another computer.  Unfortunately, I can't reuse 
the original CD because I destroyed it when I thought it was faulty.  I would 
have liked to reuse it to see what happened!

Lisi

Re: Further to my installation error

[Pascal Hambourg]

Adam Wilson a écrit :
> 
> There is a potential problem also where modern machines (with UEFI
> boot) fail to install off USB without UEFI, meaning that if you want a
> UEFI-free install, you have to use optical media.

I have a rather old (~2007) UEFI motherboard which does the opposite :
its UEFI firmware lacks USB and AHCI drivers so it is impossible to boot
in UEFI mode from a USB device or from a SATA drive in AHCI mode.

> This happened to me
> on an ASUS UX51vz with UEFI- with Debian 7 d-i, inserting an optical
> disc would provide the option to either boot the disk with UEFI or
> without it,

Was this option displayed in the regular firmware boot menu or in a
different menu ?

> while inserting a USB would only allow UEFI boot from the
> flash drive, and not legacy booting, even with Launch CSM enabled.

Do you mean the the firmware boot menu only displayed one option (EFI
mode) for the USB device ?

Did you try to boot a non-EFI capable USB boot media, such as a Debian
live image or a Debian installation image with the EFI partition
deleted ?

> So while using Debian 7, I just had to use optical media to install
> Debian sans UEFI (which I always do- call it nostalgia, but I like the
> simplicity of MBR and four primary partitions rather than the
> unfamiliar layers of GPT/UEFI related cruft).

Note that UEFI is not tied to GPT and vice versa. I happily use GPT on
legacy systems when I need many partitions and LVM is not an option.
Extended and logical partitions just suck.

> Booting off USB (forced
> to use UEFI) would simply result in a black screen after selecting
> "Install".

I have seen this display problem with the EFI framebuffer driver in
Wheezy's kernel on some machines. Installing with a serial console worked.

> Debian 8, however, seems to have solved this problem.

Indeed, the kernel EFI framebuffer driver in Jessie's driver seems to
have been improved.

> USB can still
> only be booted from UEFI, but newer d-i means that installation now
> proceeds as normal until the point at which UEFI yes/no (force/leave)
> selection is reached

I don't remember seeing this option in Jessie's installer when booted in
EFI mode. How is it labeled exactly ? At what stage is it proposed ?

Re: Further to my installation error

[Lisi Reisz]

On Sunday 20 March 2016 09:34:53 Gene Heskett wrote:
> On Sunday 20 March 2016 04:54:20 Adam Wilson wrote:
> > On Fri, 18 Mar 2016 22:21:58 +
> >
> > Lisi Reisz  wrote:
> > > On Friday 18 March 2016 20:49:55 David Wright wrote:
> > > > It's far more likely that you forgot to format the partition, if
> > > > that's indeed what you wanted to do.
> > >
> > > No.  I checked and double checked that the partitions on the disk
> > > which I wanted to use for installation were all marked with the F
> > > for format, and that nothing on the disk it had been told to leave
> > > alone had an F.  It kept wanting to format the spare disk's swap,
> > > which I did not want.
> >
> > Why not? You wanted to carry over the swap created by a previous
> > installation?
>
> Doing that, leaving a potentially dirty swap for a new install?  No
> sensible reason to do so, format that puppy.

Am I the only person on this list who has ever wanted to install on one disk 
and leave another alone for some reason? Surely not!!  I wanted to install on 
sda and leave sdb alone.  So I told it not to use sdb.  Not to format sdb.  
Not to touch sdb.  Why?  Because I didn't want sdb touched.  I was not 
leaving a potentially dirty swap for a new install.  I was telling the new 
install not to use the second disk.  Sheesh.  When I could see I would just 
have disconnected sdb.  It would then have been left alone.  OK.

I regard that as a sensible reason.

Lisi

Re: Further to my installation error

[Gene Heskett]

On Sunday 20 March 2016 04:54:20 Adam Wilson wrote:

> On Fri, 18 Mar 2016 22:21:58 +
>
> Lisi Reisz  wrote:
> > On Friday 18 March 2016 20:49:55 David Wright wrote:
> > > It's far more likely that you forgot to format the partition, if
> > > that's indeed what you wanted to do.
> >
> > No.  I checked and double checked that the partitions on the disk
> > which I wanted to use for installation were all marked with the F
> > for format, and that nothing on the disk it had been told to leave
> > alone had an F.  It kept wanting to format the spare disk's swap,
> > which I did not want.
>
> Why not? You wanted to carry over the swap created by a previous
> installation?

Doing that, leaving a potentially dirty swap for a new install?  No 
sensible reason to do so, format that puppy.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page 

Re: Further to my installation error

[Adam Wilson]

On Fri, 18 Mar 2016 22:21:58 +
Lisi Reisz  wrote:

> On Friday 18 March 2016 20:49:55 David Wright wrote:
> > It's far more likely that you forgot to format the partition, if
> > that's indeed what you wanted to do.  
> 
> No.  I checked and double checked that the partitions on the disk
> which I wanted to use for installation were all marked with the F for
> format, and that nothing on the disk it had been told to leave alone
> had an F.  It kept wanting to format the spare disk's swap, which I
> did not want.

Why not? You wanted to carry over the swap created by a previous
installation?


pgprvNFxLwZls.pgp
Description: OpenPGP digital signature

Re: Further to my installation error

[Adam Wilson]

On Fri, 18 Mar 2016 15:49:55 -0500
David Wright  wrote:

> On Wed 16 Mar 2016 at 14:49:44 (+0300), Adam Wilson wrote:
> > On Tue, 15 Mar 2016 22:41:14 + Lisi Reisz 
> > wrote:
> >   
> > > So - if you want to format an existing partition properly, so the
> > > stuff on it actually goes, use Gparted not the partitioner in the
> > > Jessie installer.  Sad.  
> 
> I can't see why Lisi came to that conclusion from what she reported
> here.
> 
> > Is it possible for the d-i partitioning utility to sometimes make
> > mistakes when creating a new FS (I have only observed this with
> > ext4, if it even happened at all and was not a product of my
> > imagination) and somehow leave files from the old FS in the new
> > over-riding FS if the two are in the same partition?
> > 
> > Sometimes after over-writing an old ext4 partition with a new ext4
> > partition of the same size and location, mysterious files have
> > appeared in the new FS which should not exist, but existed in the
> > previous instance; "ghost files" if you will.  
> 
> It's far more likely that you forgot to format the partition, if
> that's indeed what you wanted to do.
> 
> "over-writing an old ext4 partition with a new ext4 partition of the
> same size and location" is exactly what one does if the partition
> table gets mangled. People who do that will then cross their fingers
> and hope that *all* the pre-existing files reappear. They're not
> "ghost" files; they're the actual files which should be untouched by
> repartitioning.
> 
> Some people really can't get on with the d-i partitioner/formatter.
> At least in its curses mode, it's not the easiest interface. (I've
> never tried the graphical version.)

The graphical version is literally a GTK clone of the curses-based
interface. It responds to exactly the same commands- if one were to
close one's eyes and carry out a series of curses interface operations
with keypresses then exactly the same result would be achieved.

I actually really like the partitioner in d-i. It is a nice mix between
user-friendliness (you can just select "guided" if you don't know what
you're doing, and there are nice lists of options for everything) and
usability for more advanced users.

It provides a unified interface for setting mount options, filesystems,
partitioning, etc. rather than the fdisk + mkfs approach.


pgpgWktcB_5In.pgp
Description: OpenPGP digital signature

Re: Further to my installation error

[Adam Wilson]

On Tue, 15 Mar 2016 16:50:43 -0400 doug 
wrote:

> While you have GParted running, why not make the partitions you want 
> using that, and format them to ext4. Then when you go to install
> the new system, just tell it to use existing partitions.

Provided you aren't using GPT, can you use fdisk too? Just create the
desired partitions, set the partition numbers for the intended file
systems (83, most likely) and then let the installer's mkfs do the rest.

Re: Further to my installation error

[Adam Wilson]

On Wed, 16 Mar 2016 20:49:28 + Brian  wrote:

> On Wed 16 Mar 2016 at 16:12:20 -0400, Felix Miata wrote:
> 
> > Brian composed on 2016-03-16 19:21 (UTC):
> > 
> > >On Tue 15 Mar 2016 at 22:41:14 +, Lisi Reisz wrote:
> > 
> > >>Pity it took me until it tried to use CD1 to do a net-install to
> > >>realise that!!  And I have now run out of writable CDs. :-(  I
> > >>had just enough for this shenanigans!
> > 
> > >Why any Debian user should use an antiquated technology to install
> > >is beyond me. USB sticks are two a penny. Isohybrid images rule;
> > >OK!
> > 
> > USB sticks, being of a non-uniform variety of sizes, shapes, speed,
> > and reliability, are a pain to library. Inferior amount of space on
> > which to write on them contributes to the library problem. Pricing
> > of USB sticks on a per device basis remains much higher than OM,
> > making creation of a single device for single purpose generally
> > much more expensive than OM. There still exist working puters that
> > cannot boot USB. I have several.
> 
> So do I. There are ways round it.
> 
> But the vast majortiy of users are not using the ancient machines you
> and I have. The OP is at liberty to indicate whether her machine falls
> into this class. A round shiny disc could be her only solution to
> booting a Debian image.

There is a potential problem also where modern machines (with UEFI
boot) fail to install off USB without UEFI, meaning that if you want a
UEFI-free install, you have to use optical media. This happened to me
on an ASUS UX51vz with UEFI- with Debian 7 d-i, inserting an optical
disc would provide the option to either boot the disk with UEFI or
without it, while inserting a USB would only allow UEFI boot from the
flash drive, and not legacy booting, even with Launch CSM enabled.

So while using Debian 7, I just had to use optical media to install
Debian sans UEFI (which I always do- call it nostalgia, but I like the
simplicity of MBR and four primary partitions rather than the
unfamiliar layers of GPT/UEFI related cruft). Booting off USB (forced
to use UEFI) would simply result in a black screen after selecting
"Install".

Debian 8, however, seems to have solved this problem. USB can still
only be booted from UEFI, but newer d-i means that installation now
proceeds as normal until the point at which UEFI yes/no (force/leave)
selection is reached- so with Debian 7 I had to use CDs/DVDs, but with
8 I am now free to use USB (my preferred way to carry out
installations).

That machine is also unfortunate in the sense that it has a wifi card
(Centrino Advanced-N 6235) which requires non-free firmware in iwlwifi.
So in my latest Debian installation on that machine, I used a full CD-1
image, burned to USB, booted via UEFI (which I was forced to do), and
then did a UEFI-free installation using the options within the menu
(after having turned off the Great Satan that is Restricted Boot), but
declined to set up network interfaces via the installer (since the
firmware in Debian is non-free).

I have a USB wifi adapter based on AR9271 which I then plugged in. I
compiled the free firmware (open-atheros-firmware) on a separate
machine (which generated over 2GB worth of cruft) and copied across the
firmware. Et voila! A working Debian 8 UX51vz with free wifi.

That firmware could prove in handy later.

Re: Further to my installation error

[Brian]

On Wed 16 Mar 2016 at 16:12:20 -0400, Felix Miata wrote:

> Brian composed on 2016-03-16 19:21 (UTC):
> 
> >On Tue 15 Mar 2016 at 22:41:14 +, Lisi Reisz wrote:
> 
> >>Pity it took me until it tried to use CD1 to do a net-install to realise
> >>that!!  And I have now run out of writable CDs. :-(  I had just enough for
> >>this shenanigans!
> 
> >Why any Debian user should use an antiquated technology to install is
> >beyond me. USB sticks are two a penny. Isohybrid images rule; OK!
> 
> USB sticks, being of a non-uniform variety of sizes, shapes, speed, and
> reliability, are a pain to library. Inferior amount of space on which to
> write on them contributes to the library problem. Pricing of USB sticks on a
> per device basis remains much higher than OM, making creation of a single
> device for single purpose generally much more expensive than OM. There still
> exist working puters that cannot boot USB. I have several.

So do I. There are ways round it.

But the vast majortiy of users are not using the ancient machines you
and I have. The OP is at liberty to indicate whether her machine falls
into this class. A round shiny disc could be her only solution to
booting a Debian image.

> Not that I routinely burn OM to install Linux. For that I usually don't burn
> anything, instead installing by loading an installation kernel and initrd
> with an already installed bootloader. HTTP installation means up-to-date at
> the outset.

This makes installing Debian straightforward for everyone? Compared with
just putting an image on a USB stick?

Re: Further to my installation error

[Adam Wilson]

On Tue, 15 Mar 2016 22:41:14 + Lisi Reisz 
wrote:

> So - if you want to format an existing partition properly, so the
> stuff on it actually goes, use Gparted not the partitioner in the
> Jessie installer.  Sad.

Is it possible for the d-i partitioning utility to sometimes make
mistakes when creating a new FS (I have only observed this with ext4,
if it even happened at all and was not a product of my imagination) and
somehow leave files from the old FS in the new over-riding FS if the
two are in the same partition?

Sometimes after over-writing an old ext4 partition with a new ext4
partition of the same size and location, mysterious files have appeared
in the new FS which should not exist, but existed in the previous
instance; "ghost files" if you will.

Re: Further to my installation error

[Brian]

On Tue 15 Mar 2016 at 22:41:14 +, Lisi Reisz wrote:

> On Tuesday 15 March 2016 20:22:23 Lisi Reisz wrote:
> > So I am trying to use Gparted on Knoppix to wipe the disk thoroughly and
> > I'll try in order (with a good Gparted wipe in between) Net install 8.02
> 
> Worked a treat.  No further problems.  OS and DE fully installed and am 
> installing extras like LibreOffice etc., and configuring it for the user.  It 
> will be fully operational before I go to bed, as I had promised.
> 
> So - if you want to format an existing partition properly, so the stuff on it 
> actually goes, use Gparted not the partitioner in the Jessie installer.  Sad.

The installer on a netinst image is identical to the one on CD-1, the 
only difference being that GNOME packages are got from CD-1 rather than
downloading them from the internet. This is well after the partitioning
stage. Nevertheless, strange.

> Pity it took me until it tried to use CD1 to do a net-install to realise 
> that!!  And I have now run out of writable CDs. :-(  I had just enough for 
> this shenanigans!

Why any Debian user should use an antiquated technology to install is
beyond me. USB sticks are two a penny. Isohybrid images rule; OK!

Re: Further to my installation error

[Lisi Reisz]

On Saturday 19 March 2016 01:27:33 David Wright wrote:
> I'm also sorry if placing my comment to you where I did caused you or
> other people to think my later paragraphs were also directed at you.
> I assumed that my use of the standard ">" and "> >" conventions would
> make the referents clear, and they obviously didn't.

Mea culpa.  I'll answer the rest later, but an apology is called for here.  It 
should have been clear.  If it was not, the fault is mine not David's.

Lisi

Re: Further to my installation error

[David Wright]

On Wed 16 Mar 2016 at 14:49:44 (+0300), Adam Wilson wrote:
> On Tue, 15 Mar 2016 22:41:14 + Lisi Reisz 
> wrote:
> 
> > So - if you want to format an existing partition properly, so the
> > stuff on it actually goes, use Gparted not the partitioner in the
> > Jessie installer.  Sad.

I can't see why Lisi came to that conclusion from what she reported
here.

> Is it possible for the d-i partitioning utility to sometimes make
> mistakes when creating a new FS (I have only observed this with ext4,
> if it even happened at all and was not a product of my imagination) and
> somehow leave files from the old FS in the new over-riding FS if the
> two are in the same partition?
> 
> Sometimes after over-writing an old ext4 partition with a new ext4
> partition of the same size and location, mysterious files have appeared
> in the new FS which should not exist, but existed in the previous
> instance; "ghost files" if you will.

It's far more likely that you forgot to format the partition, if
that's indeed what you wanted to do.

"over-writing an old ext4 partition with a new ext4 partition of the
same size and location" is exactly what one does if the partition table
gets mangled. People who do that will then cross their fingers and
hope that *all* the pre-existing files reappear. They're not "ghost"
files; they're the actual files which should be untouched by
repartitioning.

Some people really can't get on with the d-i partitioner/formatter.
At least in its curses mode, it's not the easiest interface. (I've
never tried the graphical version.) And I admit that I prefer using
familiar tools to set up my disks before installing, but that's as
much about making sure that everything I want is well backed-up first.
I don't like gparted either. Give me cfdisk for MSDOS-style and gdisk
(ie GPT fdisk) for GPT-style.

Cheers,
David.

Re: Further to my installation error

[Felix Miata]

Brian composed on 2016-03-16 19:21 (UTC):


On Tue 15 Mar 2016 at 22:41:14 +, Lisi Reisz wrote:



Pity it took me until it tried to use CD1 to do a net-install to realise
that!!  And I have now run out of writable CDs. :-(  I had just enough for
this shenanigans!



Why any Debian user should use an antiquated technology to install is
beyond me. USB sticks are two a penny. Isohybrid images rule; OK!


USB sticks, being of a non-uniform variety of sizes, shapes, speed, and 
reliability, are a pain to library. Inferior amount of space on which to 
write on them contributes to the library problem. Pricing of USB sticks on a 
per device basis remains much higher than OM, making creation of a single 
device for single purpose generally much more expensive than OM. There still 
exist working puters that cannot boot USB. I have several.


Not that I routinely burn OM to install Linux. For that I usually don't burn 
anything, instead installing by loading an installation kernel and initrd 
with an already installed bootloader. HTTP installation means up-to-date at 
the outset.

--
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

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

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

Re: Further to my installation error

[David Wright]

On Fri 18 Mar 2016 at 22:21:58 (+), Lisi Reisz wrote:
> On Friday 18 March 2016 20:49:55 David Wright wrote:
> > It's far more likely that you forgot to format the partition, if
> > that's indeed what you wanted to do.
> 
> No.  I checked and double checked that the partitions on the disk which I 
> wanted to use for installation were all marked with the F for format, and 
> that nothing on the disk it had been told to leave alone had an F.  It kept 
> wanting to format the spare disk's swap, which I did not want.  So whilst I 
> certainly do not think that it did format properly, it was clearly, 
> explicitly and repeatedly (all three times) asked to do so.
> 
> Since it was clearly important that the partitions should be reformatted, it 
> would have been remarkably stupid of me to have omitted to do so.  On what 
> basis are you accusing me of having forgotten?  That this particular problem 
> has not hit you?  No, it has not hit me before.

Um, the quotation "It's far...to do." above was written in response
to two paragraphs by Adam Wilson "Is it possible...if you will.", and
had nothing to do with you or your problem/situation at all. The "you"
in "you forgot" is Adam. So I won't comment further on what you've just
written above.

So why did I write "I can't see why Lisi came to that conclusion from
what she reported here." Because I read the OP which AFAICT talked
about corrupt files, checksums, and old firmware. Your next post said
the crisis was over. Your next post (new thread) mentioned a download
problem. The next post said you had the same problem as before.

At this point, you also mentioned repartitioning, reformatting and
wiping a disk thoroughly.

In the last post (which I commented on when the second paragraph was
quoted in Adam Wilson's post) you said that trying to use CD1 to do a
net install made you realise that "if you want to format an existing
partition properly, so the stuff on it actually goes, use Gparted not
the partitioner in the Jessie installer."

Maybe I'm thick, but I can't see why the Debian installer's performance
at re-partition/format/wipe would cause the reported download
problems, checksum failures, corrupt files and old firmware. Sorry.
I'm no great fan of the d-i partitioner, but I haven't had it fail to
do what it promised. (Not that I understand precisely the term "wipe".)

I'm also sorry if placing my comment to you where I did caused you or
other people to think my later paragraphs were also directed at you.
I assumed that my use of the standard ">" and "> >" conventions would
make the referents clear, and they obviously didn't. I just tried to
avoid making two posts where one would suffice. Failed again. This
post makes two.

Cheers,
David.

Re: Further to my installation error

[Felix Miata]

Brian composed on 2016-03-16 20:49 (UTC):


On Wed 16 Mar 2016 at 16:12:20 -0400, Felix Miata wrote:



Brian composed on 2016-03-16 19:21 (UTC):



>Why any Debian user should use an antiquated technology to install is
>beyond me. USB sticks are two a penny. Isohybrid images rule; OK!



USB sticks, being of a non-uniform variety of sizes, shapes, speed, and
reliability, are a pain to library. Inferior amount of space on which to
write on them contributes to the library problem. Pricing of USB sticks on a
per device basis remains much higher than OM, making creation of a single
device for single purpose generally much more expensive than OM. There still
exist working puters that cannot boot USB. I have several.



So do I. There are ways round it.


Around which do you refer to with "it"?


A round shiny disc could be her only solution to booting a Debian image.


Of course.


Not that I routinely burn OM to install Linux. For that I usually don't burn
anything, instead installing by loading an installation kernel and initrd
with an already installed bootloader. HTTP installation means up-to-date at
the outset.



This makes installing Debian straightforward for everyone?


I wasn't suggesting anything about straightforward, just providing some 
context based on how things go around here. My installations are far more 
often pre-releases, so discs burned from isos would infrequently get repeated 
use. A lot of time would be wasted downloading isos full of 
never-to-be-installed packages instead of downloading needed packages that 
might be replaced on the mirrors even before the package completes downloading.



Compared with just putting an image on a USB stick?


With the result that the installation is already fully up-to-date when the 
installation process exits?


Lots of ways to skin these Linux cats.
--
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

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

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

Re: Further to my installation error

[Lisi Reisz]

On Friday 18 March 2016 20:49:55 David Wright wrote:
> It's far more likely that you forgot to format the partition, if
> that's indeed what you wanted to do.

No.  I checked and double checked that the partitions on the disk which I 
wanted to use for installation were all marked with the F for format, and 
that nothing on the disk it had been told to leave alone had an F.  It kept 
wanting to format the spare disk's swap, which I did not want.  So whilst I 
certainly do not think that it did format properly, it was clearly, 
explicitly and repeatedly (all three times) asked to do so.

Since it was clearly important that the partitions should be reformatted, it 
would have been remarkably stupid of me to have omitted to do so.  On what 
basis are you accusing me of having forgotten?  That this particular problem 
has not hit you?  No, it has not hit me before.

Lisi

Re: Further to my installation error

[David Christensen]

On 03/15/2016 03:41 PM, Lisi Reisz wrote:

On Tuesday 15 March 2016 20:22:23 Lisi Reisz wrote:

So I am trying to use Gparted on Knoppix to wipe the disk thoroughly and
I'll try in order (with a good Gparted wipe in between) Net install 8.02


Worked a treat.  No further problems.  OS and DE fully installed and am
installing extras like LibreOffice etc., and configuring it for the user.  It
will be fully operational before I go to bed, as I had promised.

So - if you want to format an existing partition properly, so the stuff on it
actually goes, use Gparted not the partitioner in the Jessie installer.  Sad.

Pity it took me until it tried to use CD1 to do a net-install to realise
that!!  And I have now run out of writable CDs. :-(  I had just enough for
this shenanigans!


It is also possible that the Debian mirrors were having problems and 
were fixed on your last attempted install.



Be sure to take an image of the system drive before you put the machine 
into production.  Take another one in a few days once everything is settled.



David

Re: Further to my installation error

[Lisi Reisz]

On Tuesday 15 March 2016 20:22:23 Lisi Reisz wrote:
> So I am trying to use Gparted on Knoppix to wipe the disk thoroughly and
> I'll try in order (with a good Gparted wipe in between) Net install 8.02

Worked a treat.  No further problems.  OS and DE fully installed and am 
installing extras like LibreOffice etc., and configuring it for the user.  It 
will be fully operational before I go to bed, as I had promised.

So - if you want to format an existing partition properly, so the stuff on it 
actually goes, use Gparted not the partitioner in the Jessie installer.  Sad.

Pity it took me until it tried to use CD1 to do a net-install to realise 
that!!  And I have now run out of writable CDs. :-(  I had just enough for 
this shenanigans!

Lisi

Re: Further to my installation error

[doug]

On 03/15/2016 04:22 PM, Lisi Reisz wrote:
So I am trying to use Gparted on Knoppix to wipe the disk thoroughly 
and I'll try in order (with a good Gparted wipe in between) Net 
install 8.02, CD1 8.03, net install 8.0.0 for a minimum install. Then, 
if I am still remotely, vaguely sane, I'll report back. Lisi 
While you have GParted running, why not make the partitions you want 
using that, and format them to ext4. Then when you go to install
the new system, just tell it to use existing partitions. If it asks to 
format them, you can let it--it won't hurt. That may take one possible 
problem

out of the process.

--doug

Re: Further to my installation error

[Lisi Reisz]

On Tuesday 15 March 2016 19:37:24 Andrew M.A. Cater wrote:
> On Tue, Mar 15, 2016 at 07:06:17PM +, Lisi Reisz wrote:
> > I tried another two and the second one the error message changed to
> > "Couldn't download package".
> >
> > Is there a problem on the Debian site?  Or with whatever mirror it is
> > helping itself to??
> >
> > Lisi
>
> What's the model of the new machine?
>
> I'm not sure that there's a problem with all the Debian mirrors: retry the
> downloads - instead of using ftp.uk.debian.org change it, perhaps to
> ftp.de.debian.org to check.
>
> get.debian.org, grab a CD image - myself, I quite like the CD #1 which
> gives you KDE but each to her own.
>
> Quicker, because smaller, is to download another netinstall.
>
> All the very best, as ever,
>
> AndyC

Thanks, Andy.  In the meantime I had downloaded the CD1, burnt it and tried 
with that - and had the same problem, and still trying to do a net install, 
although on every occasion I had told it to re-partition and format.

So I am trying to use Gparted on Knoppix to wipe the disk thoroughly and I'll 
try in order (with a good Gparted wipe in between) Net install 8.02, CD1 
8.03, net install 8.0.0 for a minimum install.  Then, if I am still remotely, 
vaguely sane, I'll report back.

Lisi

Re: Further to my installation error

[Andrew M.A. Cater]

On Tue, Mar 15, 2016 at 07:06:17PM +, Lisi Reisz wrote:
> I tried another two and the second one the error message changed to "Couldn't 
> download package".
> 
> Is there a problem on the Debian site?  Or with whatever mirror it is helping 
> itself to??
> 
> Lisi

What's the model of the new machine?

I'm not sure that there's a problem with all the Debian mirrors: retry the 
downloads - instead of using ftp.uk.debian.org change
it, perhaps to ftp.de.debian.org to check.

get.debian.org, grab a CD image - myself, I quite like the CD #1 which gives 
you KDE but each to her own.

Quicker, because smaller, is to download another netinstall.

All the very best, as ever,

AndyC