[Linux-ha-dev] make --enable-libc-alloc the default?

[Alan Robertson]

Lars Marowsky-Bree wrote:
> On 2007-02-21T08:09:43, Alan Robertson <alanr at unix.sh> wrote:
> 
>> Which is EXACTLY what it's supposed to do for months or years at a time
>> - and in fact what it does do for months or years at a time.
>>
>> The fact that _under test conditions_ it consumes a bit more is a bit
>> artificial, IMHO.
> 
> This is incomplete. It neglects that 10% speedup mean 10% faster
> fail-over, give or take. People don't care whether we take 1% or even 2%
> at runtime, but they'll bother if we take 10% less time to fail-over.
> 
> (Given of course that the CCM implementation right now slows us down
> tons as well, this may not be significant.)

It doesn't mean that AT ALL.  Failover time is normally 90% dominated by
resource agent time.  And increasing CPU time in a multi-process,
multi-processor situation where networking delays and scheduling delays
are typically higher than CPU time means that even the wall-clock delay
that isn't due to resource agents isn't probably more than 5% at most.
If resource agents are 90% of that time, then the delay is probably more
like .5%.

> And, it also means that the code does not provide anything we need in
> practice; during testing, valgrind/Coverity/BEAM(?) provide more useful
> feedback. Occam's razor suggests the other half of the reason for
> disabling the code.

Really?  It has caught dozens of bugs.  Different tools find different
bugs.  None are perfect.  Somehow you're saying that we should take
weapons for finding bugs out of our arsenal because when Andrew disables
them, they don't find any bugs for him.

Hmmm... I wonder why that could be...

I think there's a little circular reasoning going on here...

In addition, Occam's razor does not apply in this case.  There is no
single explanation being sought here (the context for applying the
Occam's razor pattern).  If we were to apply your principle here, then
we would get rid of valgrind because we run Coverity -- or vice versa.

And, while Andrew's code is very important, it's still not by any means
all the code in the system.  Maybe you'll recall his discussion of why
valgrind (the only runtime tool besides the malloc library) wouldn't
work for the rest of the system.

-- 
    Alan Robertson <alanr at unix.sh>

"Openness is the foundation and preservative of friendship...  Let me
claim from you at all times your undisguised opinions." - William
Wilberforce