geom_fox vs gmultipath

[Eric Anderson]

John Nielsen wrote:
> I was just reading the (pre-) release notes for 7.0 at 
> http://people.freebsd.org/~bmah/relnotes/CURRENT/relnotes.html and learned 
> about mjacob's gmultipath(8) implementation, which seems very similar to 
> phk's older geom_fox(4) but perhaps a bit more polished.
> 
> First off, is that a correct assessment? I haven't used either, but 
> externally gmultipath is distinguished by having its own control utility 
> (rather than relying on geom(8)), a manpage (and perhaps other features) 
> more consistent with other geom utilities, and support for predictable 
> volume names under /dev/multipath/.
> 
> Assuming I'm basically on target so far, does geom_fox have any features 
> that are not in gmultipath? If not, should it be considered deprecated? (or 
> should gmultipath be considered experimental since it's newer?)
> 
> I'm just curious, esp. since I may have a need for one of the two in the 
> near future.


Your assessments are pretty dead on.  geom_fox from what I understood 
was more of a proof of concept, and geom_multipath is the 'real deal'. 
I played with both, and geom_multipath does well, whereas geom_fox did 
have some shortcomings.

 From my (maybe limited) experience, you should be able to use 
geom_multipath without much issue.

Eric