[Midnightbsd-users] MidnightBSD 2.1.1 now available

[Chris]

On 2021-10-05 07:50, Lucas Holt wrote:
>> First up. I just want to thank you for for this release and all the
>> touble you took to make it. I'm seeing a gptid/diskid warning in
>> dmesg(8) and /var/log/messages that I'm not encountering on FreeBSD
>> or Linux:
>> WARNING: Unable to alias gptid/e1117e44-22ad-11ec-a7aa-28d2440fd18a to 
>> enc at n3061686369656d30/type at 0/slot at 1/elmdesc at Slot_00/p2/gptid/e1117e44-22ad-11ec-a7aa-28d2440fd18a 
>> - path too long
>> WARNING: Unable to alias gpt/mswap to 
>> enc at n3061686369656d30/type at 0/slot at 1/elmdesc at Slot_00/p3/gpt/mswap - path too 
>> long
>> 
>> these are just 2 out of some 13 entries all grouped in one section of
>> the log. While it doesn't *appear* to be causing me any troubles (now).
>> Does this appear to be anything to be concerned about? Or better; any
>> thoughts as to where I might look, so as to *correct* this?
>> 
>> FWIW I'm attaching the excerpt from dmesg(8) which also includes the output
>> of gpart(8) list for the drive.
>> 
>> Thanks again!
>> 
>> --Chris
> 
> I'm still researching this, but I don't believe it's anything to worry 
> about.
> 
> The issue is that enc generates a long path and then geom picks it up and 
> tries to
> add even more additionally long things to hit the path limit for devfs.
> 
> Not sure if the right thing here is to modify what the enc driver is doing 
> or to
> extend the path length.
> 
> I've actually found old bug reports in several FreeBSD based products 
> including
> FreeNAS mentioning this very problem. See
> https://redmine.ixsystems.com/issues/3183
Thanks for the pointer, and your reply, Lucas.

My own investigation seems to concur with yours. One one hand; the overall 
length of
either hashes could be shortened to stifle the warning (keeping the overall 
length
under the current limit). But how might that affect disks already in use, 
that
somehow *depend* on what they've been assigned? Or 2; (as you already noted), 
simply
extend the current length limit.
In the end, if I'm not mistaken, geom(8) and friends only taste as much as 
they
need to ensure uniqueness. So the rest is just "noise". ;-)
I'll propose a patch to silence this when I can find the time. In the 
meantime
I'll just ignore the warning. :-)

Thanks again, Lucas!

--Chris