Re: kern/57409: Feature request: Make GPT detection on VND configuration optional (and/or disable it if configuration is originated from xbdback)

[Matthias Petermann <mp%petermann-it.de@localhost>]

Hello,

On 16.05.23 07:10, Michael van Elst wrote:
> The following reply was made to PR kern/57409; it has been noted by GNATS.
>
> From: mlelstv%serpens.de@localhost (Michael van Elst)
> To: gnats-bugs%netbsd.org@localhost
> Cc:
> Subject: Re: kern/57409: Feature request: Make GPT detection on VND configuration optional (and/or disable it if configuration is originated from xbdback)
> Date: Tue, 16 May 2023 05:05:57 -0000 (UTC)
>
>   mp%petermann-it.de@localhost writes:
>   
>   >This can lead to problems if:
>   
>   > - the GUID of the GPT is already known in the system
>   > - the GUID of the GPT partition is known in the system
>   
>   >Functionally this does not lead to any limitation, it is rather the error message that confuses and the feeling that the host should not try to interfere with the file images unless it is explicitly desired.
>   
>   It will "interfere" the same, if any other kind of label is used, just less noisy.
>   
>   
>   >For testing it should be sufficient to create a new image with dd, configure it as VND, create a GPT with a partition on it and format it with FFS. Then unconfigure, copy the image to a second file and now copy both images one after the other with vndconfig. When configuring the second image, the error message should appear.
>   
>   When you clone a GPT based image, you should use 'gpt uuid -a' to create unique UUIDs on
>   the copy to avoid such conflicts.
>   

Thanks for the tip with gpt uuid -a and the objective assessment. In the 
end, I came to the conclusion that I might want a complex solution here 
for a very specific problem for which there are quite a few better 
solutions.

For the backing store images of my VMs, I have now decided that I will 
generally do without labels for the GPT partitions, in order to exclude 
this as a source of collisions. I now rely exclusively on uuids, which I 
either randomly assign during image creation, or - if the image is 
created as a copy - are reassigned with gpt uuid -a as suggested by you.

Consequently, this PR is no longer necessary and I would like to close it.

Many thanks & best regards
Matthias

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature