Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #247310 > unrolled thread

What happened?

Back to article view | Back to linux.debian.user

Contents

  What happened? Dennis Wicks <wix@mgssub.com> - 2022-04-17 01:20 +0200
    Re: What happened? Nicholas Geovanis <nickgeovanis@gmail.com> - 2022-04-17 04:40 +0200
      Re: What happened? The Wanderer <wanderer@fastmail.fm> - 2022-04-17 05:10 +0200
        Re: What happened? Stefan Monnier <monnier@iro.umontreal.ca> - 2022-04-17 07:30 +0200
          Re: What happened? The Wanderer <wanderer@fastmail.fm> - 2022-04-17 13:30 +0200
          Re: What happened? Dan Ritter <dsr@randomstring.org> - 2022-04-17 13:30 +0200
            Re: What happened? Charles Curley <charlescurley@charlescurley.com> - 2022-04-17 23:30 +0200
              Re: What happened? Dan Ritter <dsr@randomstring.org> - 2022-04-18 01:50 +0200
    Solved (partially) Re: What happened? Dennis Wicks <wix@mgssub.com> - 2022-04-21 20:40 +0200
      Re: Solved (partially) Re: What happened? The Wanderer <wanderer@fastmail.fm> - 2022-04-22 00:20 +0200
        Re: Solved (partially) Re: What happened? Dennis Wicks <wix@mgssub.com> - 2022-04-23 04:50 +0200

#247310 — What happened?

When I first installed Debian 10, I installed Win 10 in a 
virtual machine using KVM/QEMU and everything just worked. I 
could copy/paste between host and vm and access host disks 
in the vm. And zfs file systems worked.

The main problem was that dpkg would kill the system trying 
to setup
linux-image-4.19.0-18-amd64.

I decided to bite the bullet and upgrade to Debian 10.11. 
That process managed  to install linux-image-4.19.0-19-amd64 
and everything was fine. Then somewhere in the process of 
installing additional software apt? killed the system trying 
to setup linux-image-4.19.0-20-amd64. (I found out that I 
could do dpkg --configure for each package individually and 
skip the linux image and headers.)

And, now I can't copy/paste between host and vm, can't 
access host disks from the vm, and the zfs file system 
doesn't work! And then because dpkg killed the system so 
many times the archive file is locked somehow and the 
various apt(x) programs and dpkg won't run, so I can't 
install any tools or different software to try and fix the 
problems!!

ARGHHH!

If anybody has any hints or tips or pointers toward any 
possible solutions I would appreciate hearing them!

TIA,
Dennis

PS; dpkg says;

> E: dpkg was interrupted, you must manually run 'sudo dpkg --configure -a' to correct the problem. 
> W: Could not lock the cache file; this usually means that dpkg or another apt tool is already installing packages.  Opening in read-only mode; any changes you make to the states of packages will NOT be preserved!

[toc] | [next] | [standalone]

#247316

[Multipart message — attachments visible in raw view] — view raw

n Sat, Apr 16, 2022, 6:17 PM Dennis Wicks <wix@mgssub.com> wrote:

> When I first installed Debian 10, I installed Win 10 in a
> virtual machine using KVM/QEMU and everything just worked. I
> could copy/paste between host and vm and access host disks
> in the vm. And zfs file systems worked.
>
> The main problem was that dpkg would kill the system trying
> to setup
> linux-image-4.19.0-18-amd64.
>
> I decided to bite the bullet and upgrade to Debian 10.11.
> That process managed  to install linux-image-4.19.0-19-amd64
> and everything was fine. Then somewhere in the process of
> installing additional software apt? killed the system trying
> to setup linux-image-4.19.0-20-amd64. (I found out that I
> could do dpkg --configure for each package individually and
> skip the linux image and headers.)
>
> And, now I can't copy/paste between host and vm, can't
> access host disks from the vm, and the zfs file system
> doesn't work! And then because dpkg killed the system so
> many times the archive file is locked somehow and the
> various apt(x) programs and dpkg won't run, so I can't
> install any tools or different software to try and fix the
> problems!!
>
> ARGHHH!
>
> If anybody has any hints or tips or pointers toward any
> possible solutions I would appreciate hearing them!
>
> TIA,
> Dennis
>
> PS; dpkg says;
>
> > E: dpkg was interrupted, you must manually run 'sudo dpkg --configure
> -a' to correct the problem.
> > W: Could not lock the cache file; this usually means that dpkg or
> another apt tool is already installing packages.  Opening in read-only
> mode; any changes you make to the states of packages will NOT be preserved!
>

First: See if there are apt processes from previous attempts that are still
running. Maybe uselessly spinning thru CPU cycles, maybe pegging a CPU.
Kill those processes, you'll need to use the "-9" or KILL signal.

For your issue that brought you there: it sounds like maybe a filesystem
filled during the installation? Guessing.

[toc] | [prev] | [next] | [standalone]

#247318

[Multipart message — attachments visible in raw view] — view raw

On 2022-04-16 at 22:32, Nicholas Geovanis wrote:

> On Sat, Apr 16, 2022, 6:17 PM Dennis Wicks <wix@mgssub.com> wrote:
> 
>> When I first installed Debian 10, I installed Win 10 in a
>> virtual machine using KVM/QEMU and everything just worked. I
>> could copy/paste between host and vm and access host disks
>> in the vm. And zfs file systems worked.
>>
>> The main problem was that dpkg would kill the system trying
>> to setup
>> linux-image-4.19.0-18-amd64.
>>
>> I decided to bite the bullet and upgrade to Debian 10.11.
>> That process managed  to install linux-image-4.19.0-19-amd64
>> and everything was fine. Then somewhere in the process of
>> installing additional software apt? killed the system trying
>> to setup linux-image-4.19.0-20-amd64. (I found out that I
>> could do dpkg --configure for each package individually and
>> skip the linux image and headers.)
>>
>> And, now I can't copy/paste between host and vm, can't
>> access host disks from the vm, and the zfs file system
>> doesn't work! And then because dpkg killed the system so
>> many times the archive file is locked somehow and the
>> various apt(x) programs and dpkg won't run, so I can't
>> install any tools or different software to try and fix the
>> problems!!
>>
>> ARGHHH!
>>
>> If anybody has any hints or tips or pointers toward any
>> possible solutions I would appreciate hearing them!
>>
>> TIA,
>> Dennis
>>
>> PS; dpkg says;
>>
>> > E: dpkg was interrupted, you must manually run 'sudo dpkg --configure
>> -a' to correct the problem.
>> > W: Could not lock the cache file; this usually means that dpkg or
>> another apt tool is already installing packages.  Opening in read-only
>> mode; any changes you make to the states of packages will NOT be preserved!
> 
> First: See if there are apt processes from previous attempts that are still
> running. Maybe uselessly spinning thru CPU cycles, maybe pegging a CPU.
> Kill those processes, you'll need to use the "-9" or KILL signal.

Also, check that the relevant partition (probably /var, if you're using
separate partitions for such things) isn't full. I'd also suggest
checking under /var for lock files that look relevant, but I just
checked on *my* computer right now and there are several
relevant-looking lock files even though neither dpkg nor apt is running,
so I don't think messing with those would be safe.

> For your issue that brought you there: it sounds like maybe a filesystem
> filled during the installation? Guessing.

My guess is that the upgrade in question was carried out while the VM
was running, and when the upgrade tried to unload one or more relevant
kernel modules, the kernel refused to let that happen because the module
was in use.

I've seen the reverse - the module gets unloaded, and then the VM
hypervisor program dies hard, taking the guest VM down with it - happen
before, with VirtualBox, back when I was still running that under Linux.

-- 
   The Wanderer

The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man.         -- George Bernard Shaw

[toc] | [prev] | [next] | [standalone]

#247319

> My guess is that the upgrade in question was carried out while the VM
> was running, and when the upgrade tried to unload one or more relevant
> kernel modules, the kernel refused to let that happen because the module
> was in use.

Hmm... I can't remember ever seeing an upgrade remove kernel modules.
When/why would that happen?


        Stefan

[toc] | [prev] | [next] | [standalone]

#247322

[Multipart message — attachments visible in raw view] — view raw

On 2022-04-17 at 01:26, Stefan Monnier wrote:

>> My guess is that the upgrade in question was carried out while the
>> VM was running, and when the upgrade tried to unload one or more
>> relevant kernel modules, the kernel refused to let that happen
>> because the module was in use.
> 
> Hmm... I can't remember ever seeing an upgrade remove kernel
> modules. When/why would that happen?

I don't remember the specifics in detail, but I know it involves DKMS at
some stage.

I wouldn't necessarily expect it to happen even with the "rebuild module
on package update" model used with DKMS, but as I said, I know I've
*seen* it happen - with VirtualBox, quite some years ago now.

-- 
   The Wanderer

The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man.         -- George Bernard Shaw

[toc] | [prev] | [next] | [standalone]

#247323

Stefan Monnier wrote: 
> > My guess is that the upgrade in question was carried out while the VM
> > was running, and when the upgrade tried to unload one or more relevant
> > kernel modules, the kernel refused to let that happen because the module
> > was in use.
> 
> Hmm... I can't remember ever seeing an upgrade remove kernel modules.
> When/why would that happen?

Doing this while running is odd. However --

ZFS is supplied via DKMS, which needs to have the proper version
of kernel headers installed to build it. There are configuration
choices which can be made so that:

- with kernel v1, headers v1 are installed and DKMS builds ZFS

- an apt upgrade occurs, installing kernel v2

- with kernel v2, headers v2 are not installed and DKMS cannot
build ZFS. After reboot, ZFS pools cannot be found.

In which case installing v2 of the headers will allow DKMS to
rebuild ZFS, and pools will either be automatically found or can
be found with 'zfs import'.

-dsr-

[toc] | [prev] | [next] | [standalone]

#247331

On Sun, 17 Apr 2022 07:06:44 -0400
Dan Ritter <dsr@randomstring.org> wrote:

> Doing this while running is odd. However --
> 
> ZFS is supplied via DKMS, which needs to have the proper version
> of kernel headers installed to build it. There are configuration
> choices which can be made so that:
> 
> - with kernel v1, headers v1 are installed and DKMS builds ZFS
> 
> - an apt upgrade occurs, installing kernel v2
> 
> - with kernel v2, headers v2 are not installed and DKMS cannot
> build ZFS. After reboot, ZFS pools cannot be found.
> 
> In which case installing v2 of the headers will allow DKMS to
> rebuild ZFS, and pools will either be automatically found or can
> be found with 'zfs import'.

Hmmm. Would installing linux-image-amd64 (or arch as appropriate) solve
that by bringing in the new headers at the same time it brought in a new
kernel?

-- 
Does anybody read signatures any more?

https://charlescurley.com
https://charlescurley.com/blog/

[toc] | [prev] | [next] | [standalone]

#247333

Charles Curley wrote: 
> On Sun, 17 Apr 2022 07:06:44 -0400
> Dan Ritter <dsr@randomstring.org> wrote:
> 
> > Doing this while running is odd. However --
> > 
> > ZFS is supplied via DKMS, which needs to have the proper version
> > of kernel headers installed to build it. There are configuration
> > choices which can be made so that:
> > 
> > - with kernel v1, headers v1 are installed and DKMS builds ZFS
> > 
> > - an apt upgrade occurs, installing kernel v2
> > 
> > - with kernel v2, headers v2 are not installed and DKMS cannot
> > build ZFS. After reboot, ZFS pools cannot be found.
> > 
> > In which case installing v2 of the headers will allow DKMS to
> > rebuild ZFS, and pools will either be automatically found or can
> > be found with 'zfs import'.
> 
> Hmmm. Would installing linux-image-amd64 (or arch as appropriate) solve
> that by bringing in the new headers at the same time it brought in a new
> kernel?

No, you need linux-image-amd64 and linux-headers-amd64

If you've specified versions of these instead, the headers don't
follow along automatically.

-dsr-

[toc] | [prev] | [next] | [standalone]

#247434 — Solved (partially) Re: What happened?

Dennis Wicks wrote on 4/16/22 18:16:
> When I first installed Debian 10, I installed Win 10 in a 
> virtual machine using KVM/QEMU and everything just worked. I 
> could copy/paste between host and vm and access host disks 
> in the vm. And zfs file systems worked.
> 
> The main problem was that dpkg would kill the system trying 
> to setup
> linux-image-4.19.0-18-amd64.
> 
> I decided to bite the bullet and upgrade to Debian 10.11. 
> That process managed  to install linux-image-4.19.0-19-amd64 
> and everything was fine. Then somewhere in the process of 
> installing additional software apt? killed the system trying 
> to setup linux-image-4.19.0-20-amd64. (I found out that I 
> could do dpkg --configure for each package individually and 
> skip the linux image and headers.)
> 
> And, now I can't copy/paste between host and vm, can't 
> access host disks from the vm, and the zfs file system 
> doesn't work! And then because dpkg killed the system so 
> many times the archive file is locked somehow and the 
> various apt(x) programs and dpkg won't run, so I can't 
> install any tools or different software to try and fix the 
> problems!!
> 
> ARGHHH!
> 
> If anybody has any hints or tips or pointers toward any 
> possible solutions I would appreciate hearing them!
> 
> TIA,
> Dennis
> 
> PS; dpkg says;
> 
>> E: dpkg was interrupted, you must manually run 'sudo dpkg 
>> --configure -a' to correct the problem. W: Could not lock 
>> the cache file; this usually means that dpkg or another 
>> apt tool is already installing packages.  Opening in 
>> read-only mode; any changes you make to the states of 
>> packages will NOT be preserved!
> 
> 

Mark this SOLVED (partially): software problem.

dpkg continued to kill the system trying to setup 
linux-headers. I discovered that dpkg will purge the 
linux-headers package and then I can try installing again. 
On one try I was running pg on dkms in an attempt to see 
what was going on that killed the machine. Two times I got a 
long screen full of tasks running various types of cc just 
before the crash.

I set up my pg to repeat every few secs and spooled the 
output to a file. Then I started a tail on that spool file. 
Surprise!! It ran to completion!

Here is my analysis: I have a big system. An AMD 8 core 
Ryzen running at 3.8GHz with 64GB of memory. I think that 
without any other load on the machine some task is getting 
started before it should and either steps on a task that 
should have finished by that time or can't find something 
that was supposed to be created by a task that hasn't run 
yet. It has to be something serious because all the 
screens/sessions are cleared and the system is frozen. I 
have to use the reset button to re-start and once or twice I 
have had to cycle the power to get re-booted.

No progress on other problems. I will open other threads for 
them.

Thanks for the comments. They got me thinking!

Dennis

[toc] | [prev] | [next] | [standalone]

#247441 — Re: Solved (partially) Re: What happened?

[Multipart message — attachments visible in raw view] — view raw

On 2022-04-21 at 14:37, Dennis Wicks wrote:

> Dennis Wicks wrote on 4/16/22 18:16:
>
>> When I first installed Debian 10, I installed Win 10 in a 
>> virtual machine using KVM/QEMU and everything just worked. I 
>> could copy/paste between host and vm and access host disks 
>> in the vm. And zfs file systems worked.
>> 
>> The main problem was that dpkg would kill the system trying 
>> to setup
>> linux-image-4.19.0-18-amd64.

> Mark this SOLVED (partially): software problem.
> 
> dpkg continued to kill the system trying to setup 
> linux-headers. I discovered that dpkg will purge the 
> linux-headers package and then I can try installing again. 
> On one try I was running pg on dkms in an attempt to see 
> what was going on that killed the machine. Two times I got a 
> long screen full of tasks running various types of cc just 
> before the crash.

What's "pg"? It doesn't appear to be e.g. a shell builtin ('type pg'
reports 'bash: type: pg: not found'), and I don't find it in the
archive. For example,

$ apt-file search -x /pg$

finds only two results, from the package grass-core, which appear to be
(parts of? related to?) database drivers.

> I set up my pg to repeat every few secs and spooled the 
> output to a file. Then I started a tail on that spool file. 
> Surprise!! It ran to completion!
> 
> Here is my analysis: I have a big system. An AMD 8 core 
> Ryzen running at 3.8GHz with 64GB of memory. I think that 
> without any other load on the machine some task is getting 
> started before it should and either steps on a task that 
> should have finished by that time or can't find something 
> that was supposed to be created by a task that hasn't run 
> yet. It has to be something serious because all the 
> screens/sessions are cleared and the system is frozen. I 
> have to use the reset button to re-start and once or twice I 
> have had to cycle the power to get re-booted.

This doesn't look like a terribly likely scenario to me, but if it *is*
what's happening, that's definitely a bug - though a bug in what is less
than clear.

-- 
   The Wanderer

The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man.         -- George Bernard Shaw

[toc] | [prev] | [next] | [standalone]

#247463 — Re: Solved (partially) Re: What happened?

The Wanderer wrote on 4/21/22 17:14:
> On 2022-04-21 at 14:37, Dennis Wicks wrote:
> 
>> Dennis Wicks wrote on 4/16/22 18:16:
>>
>>> When I first installed Debian 10, I installed Win 10 in a
>>> virtual machine using KVM/QEMU and everything just worked. I
>>> could copy/paste between host and vm and access host disks
>>> in the vm. And zfs file systems worked.
>>>
>>> The main problem was that dpkg would kill the system trying
>>> to setup
>>> linux-image-4.19.0-18-amd64.
> 
>> Mark this SOLVED (partially): software problem.
>>
>> dpkg continued to kill the system trying to setup
>> linux-headers. I discovered that dpkg will purge the
>> linux-headers package and then I can try installing again.
>> On one try I was running pg on dkms in an attempt to see
>> what was going on that killed the machine. Two times I got a
>> long screen full of tasks running various types of cc just
>> before the crash.
> 
> What's "pg"? It doesn't appear to be e.g. a shell builtin ('type pg'
> reports 'bash: type: pg: not found'), and I don't find it in the
> archive. For example,
> 
> $ apt-file search -x /pg$
> 
> finds only two results, from the package grass-core, which appear to be
> (parts of? related to?) database drivers.
> 
>> I set up my pg to repeat every few secs and spooled the
>> output to a file. Then I started a tail on that spool file.
>> Surprise!! It ran to completion!
>>
>> Here is my analysis: I have a big system. An AMD 8 core
>> Ryzen running at 3.8GHz with 64GB of memory. I think that
>> without any other load on the machine some task is getting
>> started before it should and either steps on a task that
>> should have finished by that time or can't find something
>> that was supposed to be created by a task that hasn't run
>> yet. It has to be something serious because all the
>> screens/sessions are cleared and the system is frozen. I
>> have to use the reset button to re-start and once or twice I
>> have had to cycle the power to get re-booted.
> 
> This doesn't look like a terribly likely scenario to me, but if it *is*
> what's happening, that's definitely a bug - though a bug in what is less
> than clear.
> 
Sorry, pg = ps

That is the only thing I can think of that it might be. I 
have run dpkg --configure -a or with a specific package name 
and every time the package was linux-image-... or 
linux-headers-... it killed my machine. It happened with 
kernel 4.19.0-18-amd64 and 4.19.0-20-amd64. I ran it a dozen 
or more times with the same results. The two times that I 
caught any output from ps that was useful it was running 
many tasks of cc.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web