Re: FreeBSD-provided .vhd with VirtualBox: gpart I/O errors after resizing the virtual hard disk

[Graham Perrin]

On 17/01/2021 18:22, Graham Perrin wrote:

VirtualBox 5.2.44 r139111 on FreeBSD-CURRENT.

01. Add FreeBSD-12.2-RELEASE-amd64.vhd to a machine
02. use Virtual Disk Manager to resize it to 2.0 TB
…



Long story short: it seems that the problem occurs only after resizing 
to (the maximum) 2.0 TB.


Nowadays I typically:

* resize to something less than 2.0 TB
* boot in single user mode
* gpart recover
* service start growfs
* exit single user mode.

___
freebsd-current@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"

Re: FreeBSD-provided .vhd with VirtualBox: gpart I/O errors after resizing the virtual hard disk

[Mark Millard via freebsd-current]

Graham Perrin grahamperrin at gmail.com wrote on
Mon Jan 25 22:08:53 UTC 2021 :

. . . omitted material, including steps 01..13 . . .

Your steps 01..13 make no mention of involving growfs
as indicated in "man 8 growfs". I'd guess that the
growfs step would be after step 8 (recover) and before
10 (show). (FYI: There is also a "man 7 growfs".)

> With FreeBSD-provided 'FreeBSD-12.1-RELEASE-amd64.vhd', things seem even 
> worse. Please see, for example, this screen recording:
> 
> 
> 
> What's going wrong?
> 
> Is it necessary to boot first from something _other_ than the resized 
> disk, for things such as gpart recover to proceed without I/O error for 
> the resized disk?

I looked at the screen recording's playback and . . .

You have made no mention of the cylinder checksum failure notices
or at what stage they first occurred. That might be important.
I've no clue if they might be an initial cause of issues vs.
them being an effect that might also contribute to later issues
when the checksums fail.

Unfortunately, it does not look like "man fsck_ffs" covers its
cylinder checksum failure handling:

** Phase 5 - Check Cyl groups

(At least I presume that phase is related to what the failure
notices are reporting.)

I'm afraid I'm of no general help but it looked like some extra
information might be appropriate --or the man 8 or 7 growfs
related activity being involved might avoid later getting cylinder
checksum failure notices.

I hope this proves of some help, but, if it is not, I'm unlikely
to have more information, sorry.

===
Mark Millard
marklmi at yahoo.com
( dsl-only.net went
away in early 2018-Mar)

___
freebsd-current@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"

Re: FreeBSD-provided .vhd with VirtualBox: gpart I/O errors after resizing the virtual hard disk

[Graham Perrin]

On 17/01/2021 18:22, Graham Perrin wrote:

VirtualBox 5.2.44 r139111 on FreeBSD-CURRENT.

01. Add FreeBSD-12.2-RELEASE-amd64.vhd to a machine
02. use Virtual Disk Manager to resize it to 2.0 TB
03. apply, close
04. boot single user
05. gpart show /dev/ada0
07. observe reported corruption
08. gpart recover /dev/ada0
09. I/O error 
10. gpart show /dev/ada0
11. no reported corruption
12. shutdown -r now
13. observe reported corruption:

> gptboot: invalid backup GPT header



For me, this seems to be consistently reproducible.

Thoughts?

Boot proceeds and  I guess that the 
UFS file system is automatically grown, but the reported I/O errors 
make me nervous.


% uname -v
FreeBSD 13.0-CURRENT #75 main-c572-g82397d791: Sun Jan  3 20:00:09 GMT 
2021 
root@mowa219-gjp4-8570p:/usr/obj/usr/src/amd64.amd64/sys/GENERIC-NODEBUG

%

With FreeBSD-provided 'FreeBSD-12.1-RELEASE-amd64.vhd', things seem even 
worse. Please see, for example, this screen recording:




What's going wrong?

Is it necessary to boot first from something _other_ than the resized 
disk, for things such as gpart recover to proceed without I/O error for 
the resized disk?


___
freebsd-current@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"

FreeBSD-provided .vhd with VirtualBox: gpart I/O errors after resizing the virtual hard disk

[Graham Perrin]

VirtualBox 5.2.44 r139111 on FreeBSD-CURRENT.

01. Add FreeBSD-12.2-RELEASE-amd64.vhd to a machine
02. use Virtual Disk Manager to resize it to 2.0 TB
03. apply, close
04. boot single user
05. gpart show /dev/ada0
07. observe reported corruption
08. gpart recover /dev/ada0
09. I/O error 
10. gpart show /dev/ada0
11. no reported corruption
12. shutdown -r now
13. observe reported corruption:

> gptboot: invalid backup GPT header



For me, this seems to be consistently reproducible.

Thoughts?

Boot proceeds and  I guess that the UFS 
file system is automatically grown, but the reported I/O errors make me 
nervous.


% uname -v
FreeBSD 13.0-CURRENT #75 main-c572-g82397d791: Sun Jan  3 20:00:09 GMT 
2021 
root@mowa219-gjp4-8570p:/usr/obj/usr/src/amd64.amd64/sys/GENERIC-NODEBUG

%

___
freebsd-current@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"