Re: FreeBSD 13.2R and OpenZFS bug #15933

On 2/29/24 22:22, Daniel Tameling wrote:
> On Wed, Feb 28, 2024 at 06:08:14PM -0800, David Christensen wrote:
> Here is a write-up of the person that fixed the bug containing all the
> gory details:
> https://despairlabs.com/blog/posts/2023-12-25-openzfs-data-corruption-bug/


Thank you for replying.


That is an interesting article.  I like the diagrams -- they remind me 
of a data structures course.  I am also reminded of an operating systems 
course -- specifically: shared resources, critical sections, and 
thread-safe programming techniques.  Failing to solve the mutual 
exclusion problem correctly in the design is going to result in race 
conditions in the implementation.  The article discusses concurrent 
operations, but only implies concurrent design.  I suspect this is where 
the root cause of the OpenZFS bugs is to be found.


>>>> Does this new OpenZFS bug [#15933] affect FreeBSD 13.2R?


The article states:

     "Another reason cloning was implicated was that the bug had been 
seen on FreeBSD."


So, I conclude that all FOSS OS's with FOSS ZFS were affected by the 
bugs, are still affected, and will remain affected until the root 
cause(s) are fixed.


>> So, is my data safe on up-to-date FreeBSD 13.2R ZFS with native encryption
>> disabled?
>>
> 
> It's safer than on FreeBSD 12 


I agree that running an EOL OS becomes more unsafe every day.  So, 
FreeBSD 13.2R for sure.


I am having tough time choosing between OpenZFS (low probability, 
catastrophic severity bugs) vs. GEOM RAID (giving up ZFS next-generation 
features).


> but nobody will give you any guarantees.

Of course.


David