Re: Problem with crash post-install [ was: Update or reinstall ]

[Polyna-Maude Racicot-Summerside]

Hi,

On 2021-07-29 3:13 p.m., David Wright wrote:

> 
> I realise that we're not all receiving posts from Gunnar, but do we
> actually know what the "update problem" is?
> 
> I recall:
> https://lists.debian.org/debian-user/2021/07/msg01003.html
> Is this the origin of the system under discussion?
> 
> Or is this?
> https://lists.debian.org/debian-user/2021/07/msg00945.html
> 
> Then we have:
> https://lists.debian.org/debian-user/2021/07/msg01167.html
> Do we know what is meant by "crashing"?
> 
> In between, we have snippets on Grub, bash, busybox, keyboards,
> root partition flags, and more; none of it in a coherent thread.
> Quite honestly, I don't know where syncing the filesystem comes in.
> Why not just touch wood.
> 
> Cheers,
> David.
> 
> The sync filesystem comes for a suggestion.
Sync before update and after update.
3 sync in all so I don't remember what was the 3rd one.

But yes, it's quite messy all of this !

And like you mentioned, we don't even know if it's a Debian or what type.

When I read something like this...

On 2021-07-28 3:16 p.m., Gunnar Gervin wrote:
> It is a Toshiba 160 gb hd in a 14 years old Macbook i386 ❤️/x86 32 b
> booting from Bios not uefi. I'll give full report in 1-2 weeks, after
> put in VM in it, faster internet to it to handle VM.
> And built websites with it.
> Geg

*I get really confused...*

And when you post

In all of Gunnar's posts, I think I have only seen one report of actual
output posted, and I have no idea what the origin of that was, viz:

  # UNCONFIGURED FSTAB FOR BASE SYSTEM
  overlay / overlay rw 0 0
  tmpfs /tmp tmpfs nosuid,nodev 0 0
  /dev/sda1 /mnt/sda1 EXT2 nosuid,nodev,nofail,x-gvfs-show 0 0
  /dev/sda3 /mnt/sda3 EXT2
nosuid,nodev,nofail,x-gvfs-show,noauto,x-udisks-auth 0 0
  /dev/sda2 none swap sw,x-udisks-auth,noauto 0 0

*It doesn't help much ! And you ain't to blame...*
-- 
Polyna-Maude R.-Summerside
-Be smart, Be wise, Support opensource development



OpenPGP_signature
Description: OpenPGP digital signature

Re: Problem with crash post-install [ was: Update or reinstall ]

[David Wright]

On Thu 29 Jul 2021 at 18:09:51 (+0200), Marco Möller wrote:
> On 29.07.21 12:52, Polyna-Maude Racicot-Summerside wrote:
> > I'm really asking why do it 3 time.
> > In case it came back to the prompt *before* finishing to do it's job ?
> > (...)
> Misunderstanding: I wanted to suggest to run a "sudo sync" (1) before
> and (2) after the update and (3) before the system shutdown, in order
> to exclude that data wasn't fully written to disk at possible critical
> moments. This in total are three different situations when I suggest
> to run the sync command, and in all this three different situations it
> shall be run, thus running it three times (in total three times, not
> triple repeated in one of the situations).

It doesn't actually matter in the first two cases whether the data has
been written to the device itself, as the filesystem access methods
won't allow you to read from the device, but you will receive the
up to date buffer contents instead.

> By the way, running it as "sudo" is not harming and makes sure that
> everything is written to disk without having to know if the sync
> command would behave different with and without the sudo.

You're presupposing that everyone is using the same superuser access
method. Many people will merely get the response:

$ sudo sync
Sorry, user foo is not allowed to execute '/usr/bin/sync' as root on bar.
$ 

> The point
> is, you want to debug a situation, you want to find out why an update
> is failing so severely and not reported to fail like this elsewhere,
> and sudo will not harm for this purpose while not using sudo could
> have left doubts if really all is written to disk - pragmatically
> focusing on problem solving without maybe producing more problems and
> without loosing time to construct a most elegant version of problem
> solving. Just get the problem solved!
> I am curious to read if the "sudo sync" runs solved the update problem.

I realise that we're not all receiving posts from Gunnar, but do we
actually know what the "update problem" is?

I recall:
https://lists.debian.org/debian-user/2021/07/msg01003.html
Is this the origin of the system under discussion?

Or is this?
https://lists.debian.org/debian-user/2021/07/msg00945.html

Then we have:
https://lists.debian.org/debian-user/2021/07/msg01167.html
Do we know what is meant by "crashing"?

In between, we have snippets on Grub, bash, busybox, keyboards,
root partition flags, and more; none of it in a coherent thread.
Quite honestly, I don't know where syncing the filesystem comes in.
Why not just touch wood.

Cheers,
David.

Re: Problem with crash post-install [ was: Update or reinstall ]

[Marco Möller]

On 29.07.21 12:52, Polyna-Maude Racicot-Summerside wrote:

Hi,

On 2021-07-29 6:48 a.m., to...@tuxteam.de wrote:

On Thu, Jul 29, 2021 at 12:29:16PM +0200, Marco Möller wrote:


On 29.07.21 10:25, Polyna-Maude Racicot-Summerside wrote:

Hi,
(...)
Gunnar  ( Gunnar Gervin  ) is having problem with
his laptop.
The system crashes post-update.

He need to do clean install every time.
And this is probably time consuming for him.

If other user has some hints...



[...] three times run a "sudo sync" [...]


While I've read about issuing sync *twice* with the explanation
that sysadmins are a supersticious bunch [1], the sudo is most
definitely unnecessary.

I think the running twice is also a bit of a too much extra.

I'm really asking why do it 3 time.
In case it came back to the prompt *before* finishing to do it's job ?
(...)
Misunderstanding: I wanted to suggest to run a "sudo sync" (1) before 
and (2) after the update and (3) before the system shutdown, in order to 
exclude that data wasn't fully written to disk at possible critical 
moments. This in total are three different situations when I suggest to 
run the sync command, and in all this three different situations it 
shall be run, thus running it three times (in total three times, not 
triple repeated in one of the situations).
By the way, running it as "sudo" is not harming and makes sure that 
everything is written to disk without having to know if the sync command 
would behave different with and without the sudo. The point is, you want 
to debug a situation, you want to find out why an update is failing so 
severely and not reported to fail like this elsewhere, and sudo will not 
harm for this purpose while not using sudo could have left doubts if 
really all is written to disk - pragmatically focusing on problem 
solving without maybe producing more problems and without loosing time 
to construct a most elegant version of problem solving. Just get the 
problem solved!

I am curious to read if the "sudo sync" runs solved the update problem.
Good Luck! Marco.

Re: Problem with crash post-install [ was: Update or reinstall ]

[songbird]

Dan Ritter wrote:
...
> There are situations in which it can take much longer than a few
> seconds. They require a large amount of data to write and a slow
> storage mechanism.
>
> Specific example: recursive copying of music files to my USB-2
> attached music player, which has an actual write speed of about
> 20 MB per second. The cp -r or rsync of a few gigabytes can
> return well before the data is actually written.

  i have a pretty large io buffer so when i copy some big files
and then sync it may take as long as 30 or more seconds even 
with a decent SSD.


  songbird

Re: Problem with crash post-install [ was: Update or reinstall ]

[Dan Ritter]

Greg Wooledge wrote: 
>  which says:
> 
>   On Linux, sync is only guaranteed to schedule the dirty blocks for
>   writing; it can actually take a short time before all the blocks are
>   finally written. The reboot(8) and halt(8) commands take this into
>   account by sleeping for a few seconds after calling sync(2).

There are situations in which it can take much longer than a few
seconds. They require a large amount of data to write and a slow
storage mechanism.

Specific example: recursive copying of music files to my USB-2
attached music player, which has an actual write speed of about
20 MB per second. The cp -r or rsync of a few gigabytes can
return well before the data is actually written.

-dsr-

Re: Problem with crash post-install [ was: Update or reinstall ]

[Greg Wooledge]

On Thu, Jul 29, 2021 at 06:52:37AM -0400, Polyna-Maude Racicot-Summerside wrote:
> As far as I know, sync doesn't work in the background !

On many systems, it does.  I can't even find a clear statement in the
bullseye sync(1) man page of whether it does or not.

The bullseye sync(1) comes from GNU coreutils 8.32.

Googling turns up a different sync(8) page,
 which says:

  On Linux, sync is only guaranteed to schedule the dirty blocks for
  writing; it can actually take a short time before all the blocks are
  finally written. The reboot(8) and halt(8) commands take this into
  account by sleeping for a few seconds after calling sync(2).

That man page claims to be from fileutils-4.0.

On bullseye, sync(2) says:

   According to the standard specification  (e.g.,  POSIX.1-2001),  sync()
   schedules the writes, but may return before the actual writing is done.
   However Linux waits for I/O completions, and thus  sync()  or  syncfs()
   provide the same guarantees as fsync() called on every file in the sys‐
   tem or filesystem respectively.

So, even on Linux systems, it would appear the behavior has varied
over time.  On non-Linux systems, I would expect a high probability
that the sync(8) or sync(1) command returns immediately, without
waiting for the synchronization to complete.

Re: Problem with crash post-install [ was: Update or reinstall ]

[Polyna-Maude Racicot-Summerside]

Hi,

On 2021-07-29 6:48 a.m., to...@tuxteam.de wrote:
> On Thu, Jul 29, 2021 at 12:29:16PM +0200, Marco Möller wrote:
>>
>> On 29.07.21 10:25, Polyna-Maude Racicot-Summerside wrote:
>>> Hi,
>>> (...)
>>> Gunnar  ( Gunnar Gervin  ) is having problem with
>>> his laptop.
>>> The system crashes post-update.
>>>
>>> He need to do clean install every time.
>>> And this is probably time consuming for him.
>>>
>>> If other user has some hints...
> 
>> [...] three times run a "sudo sync" [...]
> 
> While I've read about issuing sync *twice* with the explanation
> that sysadmins are a supersticious bunch [1], the sudo is most
> definitely unnecessary.
I think the running twice is also a bit of a too much extra.

I'm really asking why do it 3 time.
In case it came back to the prompt *before* finishing to do it's job ?

As far as I know, sync doesn't work in the background !
> 
> The important part is actually /waiting/ for sync to "come back".
> Then you can assume that pending writes are committed.
> 
That's a fact
> Cheers
> [1] A dash of superstition might even be beneficial in that
>kind of job.
>  - t
> 
This reassure me that I ain't the only one who find it a bit frivolous
to run sync 3 time in a row. This is not like making love, where there's
much pleasure to do so once after another.

-- 
Polyna-Maude R.-Summerside
-Be smart, Be wise, Support opensource development



OpenPGP_signature
Description: OpenPGP digital signature

Re: Problem with crash post-install [ was: Update or reinstall ]

[tomas]

On Thu, Jul 29, 2021 at 12:29:16PM +0200, Marco Möller wrote:
> 
> On 29.07.21 10:25, Polyna-Maude Racicot-Summerside wrote:
> >Hi,
> >(...)
> >Gunnar  ( Gunnar Gervin  ) is having problem with
> >his laptop.
> >The system crashes post-update.
> >
> >He need to do clean install every time.
> >And this is probably time consuming for him.
> >
> >If other user has some hints...

> [...] three times run a "sudo sync" [...]

While I've read about issuing sync *twice* with the explanation
that sysadmins are a supersticious bunch [1], the sudo is most
definitely unnecessary.

The important part is actually /waiting/ for sync to "come back".
Then you can assume that pending writes are committed.

Cheers
[1] A dash of superstition might even be beneficial in that
   kind of job.
 - t


signature.asc
Description: Digital signature

Re: Problem with crash post-install [ was: Update or reinstall ]

[Polyna-Maude Racicot-Summerside]

Hi,

On 2021-07-29 6:29 a.m., Marco Möller wrote:
> 
> On 29.07.21 10:25, Polyna-Maude Racicot-Summerside wrote:
>> Hi,
>> (...)
>> Gunnar  ( Gunnar Gervin  ) is having problem with
>> his laptop.
>> The system crashes post-update.
>>
>> He need to do clean install every time.
>> And this is probably time consuming for him.
>>
>> If other user has some hints...
> 
> Without having all the mail thread available and not being able to see
> what suggestions and tests have already been done, my first idea and
> suggestion would be to care to three times run a "sudo sync" (1) before
> and (2) after the update and (3) before the system shutdown, in order to
> exclude that data wasn't fully written to disk at possible critical
> moments. If this would help, then investigations could accordingly
> continue to concentrate on related configurations.
> Good Luck! Marco.
> 
There's no other mail thread related to this...
Only the original user (Gunnar) who said to me that I shall do clean
install on updates because his system crashes if he tries to updates.

So I told him that this was painstaking and time consuming, that backups
are indeed a good thing but that you shouldn't need to reinstall every
time you update.

So I've suggested him that he post the system configuration onto the
mailing system so we may help him out (fellow user on the list).

Now considering the way he described his computer installation, I
believe it may be much more complex to solve whatever problem it is than
trying to go take a trip to the planet Mars, no, wait, Jupiter would be
closer to the efforts needed.

In his description he mentions waiting for a better Internet connection
so his virtual machine will run faster or something alike that I don't
quite understand.

So maybe he's running under a VM.
Also, said to be running a Macbook with a BIOS and that is totally
strange to myself...

So the best thing to do (I believe) would be...

I'd start by asking him to use hwinfo... and send a dump.

Steps :
Install package hwinfo
Run hwinfo --all --log laptop.log
Copy the content of laptop.log into your message.
You can use gedit or geany to open the file.

-- 
Polyna-Maude R.-Summerside
-Be smart, Be wise, Support opensource development

cc: Gunnar Gervin



OpenPGP_signature
Description: OpenPGP digital signature

Re: Problem with crash post-install [ was: Update or reinstall ]

[Marco Möller]

On 29.07.21 10:25, Polyna-Maude Racicot-Summerside wrote:

Hi,
(...)
Gunnar  ( Gunnar Gervin  ) is having problem with
his laptop.
The system crashes post-update.

He need to do clean install every time.
And this is probably time consuming for him.

If other user has some hints...


Without having all the mail thread available and not being able to see 
what suggestions and tests have already been done, my first idea and 
suggestion would be to care to three times run a "sudo sync" (1) before 
and (2) after the update and (3) before the system shutdown, in order to 
exclude that data wasn't fully written to disk at possible critical 
moments. If this would help, then investigations could accordingly 
continue to concentrate on related configurations.

Good Luck! Marco.

Problem with crash post-install [ was: Update or reinstall ]

[Polyna-Maude Racicot-Summerside]

Hi,

On 2021-07-29 3:25 a.m., Gunnar Gervin wrote:
> The *link* is your request of it.
> Geg
> 
> On Thu, 29 Jul 2021, 00:25 Polyna-Maude Racicot-Summerside,
> mailto:deb...@polynamaude.com>> wrote:
> 
> Hi,
> 
> On 2021-07-28 3:16 p.m., Gunnar Gervin wrote:
> > It is a Toshiba 160 gb hd in a 14 years old Macbook i386 ❤️/x86 32 b
> > booting from Bios not uefi. I'll give full report in 1-2 weeks, after
> > put in VM in it, faster internet to it to handle VM.
> > And built websites with it.
> > Geg
> 
Gunnar  ( Gunnar Gervin  ) is having problem with
his laptop.
The system crashes post-update.

He need to do clean install every time.
And this is probably time consuming for him.

If other user has some hints...

> What's the link between installing a VM and the description of your
> laptop ?
> 
> Also, Macbook in x86 booting from BIOS ?
> 
> Macbook use either UEFI for the Intel ones or OpenBoot for the older
> ones.
> 
> No such thing as a BIOS on a MacBook.
> 
> Sorry but your message is everything except easy to understand.
> 
> If you are installing Linux in a VM on your laptop then you should also
> describe what type of VM you are using and the specifications. Your
> Friend Linux should have taught you this. Also, he should teach you that
> a faster internet won't make your VM faster.
> 
> Maybe you could either answer a message or include in this one, what
> reason are you writing this ?
> 
> As it may sound obvious for yourself, it's not necessarily for others.
> Plus it's not much of use for someone who'd like into to the archive
> looking for answer.
> 
> If I do get a idea of why you are writing this here, it's not the case
> for other people around.
> 
> Same apply for your message regarding Pine laptop.
> 
> Because now those two message are like a bottle in the ocean, not linked
> to anything.
> 
> -- 
> Polyna-Maude R.-Summerside
> -Be smart, Be wise, Support opensource development
> 

-- 
Polyna-Maude R.-Summerside
-Be smart, Be wise, Support opensource development



OpenPGP_signature
Description: OpenPGP digital signature