Will 5.4 be an "Extended Life" release?

[Danny Pansters]

On Monday 18 April 2005 03:05, Brett Glass wrote:
> At 03:33 PM 4/12/2005, Clifton Royston wrote:
> >  If 5.4 is expected to be an extended-life branch, I would consider
> >moving them up to 5.4 instead, to get a leap on current technology.
> >Has that decision been made yet?
>
> I have a similar dilemma. Currently, I am building all production
> servers with 4.11. But this means that I can't take advantage of
> AMD64 processors or some other things that are available in 5.x.
> And 5.x does some nice things, such as sandboxing BIND by default.
> Finally, it seems as if CPU manufacturers are rapidly moving toward
> multiple core processors, which need sophisticated SMP to work well.
> So, I'd really like to move to 5.4 when it ships.
>
> Trouble is, from the reports I'm seeing on the -STABLE list and
> my own experiments, I don't yet know if 5.4 is going to be as
> fast (especially at disk access) or stable as 4.11. (Many of the
> systems I am building will need very fast disk access, because they
> will be used as database servers and caches.) What's more, here we
> are at RC2, and there are still a number of open issues, as shown at
>
> http://www.freebsd.org/releases/5.4R/todo.html
>
> So, I am wondering if I should stick with 4.11, favoring fast single
> CPUs over multiprocessor systems, for production machines -- and then
> jump to 6.0 when it's released. Will security fixes be available long
> enough for me to do this if need be?


Let me just boldly insert that IMHO, if 6.X is going to become stable this 
autumn already that indeed 5.4 or maybe 5.5 at least one of those must be 
long-term-supported. I'm sure one of the two will, as one of the two will 
reflcet ultimately the walk-of-life of 5-STABLE, won't it?

FYI, as far as I have noticed 5.4 is mostly a bugfix/stabilize efford over 
5.3, so yes, it should be a lot better. I'm currently testing it with a UP P4 
using HTT and SMP (don't use SCHED_ULE for this ;-) and it seems to go fine. 
Dual/multiple core CPUs are going to be the norm. In that respect we're doing 
really well. I wish ULE got promoted/fixed though...

Hmm, I seem to have diverged from -security but so has Brett already.. :)

I think amd64 is still a bit fragile. I also am interested to see how the jump 
to 6-STABLE will go, but inevitably there will be some 5.X maintanance 
release that will go on for a long time. Does anyone really doubt that? I 
don't, also consider that 4.X will have to be phased out ASAP (or be prepared 
to support 3 system compilers, I don't think so, no one would like that for a 
long time). There's not much need for concern IMHO altogether.

Dan