[CentOS] Hot swap CPU

[alex at milivojevic.org]

Quoting Nikos Zaharioudakis <nzahar at gmail.com>:

> Let's assume that I have a machine with at least 4 CPUs.
> In case that one of these fails what are my options.
> I mean, could I take that cpu offline and maybe ( I say maybe) I could
> change it without taking down the server ?
> What is the suggested procedure?
> Is this the same procedure no matter what cpu architecture I have
> (Itanium, or PPc or whatever )
> Thanks in advance for the attention

I'm not aware of any Intel based servers that would allow something 
like that. The smallest machines I saw that allowed hot-swap of 
processors were mid-range
Sun servers (but they are not Intel, they are Sparc processor based).  Of
course, the OS must also support it.

Also, note that if processor fails (and system detects it), in some cases the
system must be halted anyhow.  The reason is correctness requirement.  If the
failure introduced non-repairable data corruption (and CPU failure is 
likely to
produce it), you can't continue with execution of current environment.  
There is
a way around that, but systems that implement it are very expensive.  Usually
implemented only in critical applicatins, such as for example computers that
controll airplanes.  Basically, in such critical systems you have at least 3
redundant computers executing same code in parallel, and watching the 
output of
each other.  If one produces different output than the others, system 
concludes
that the one with differing output failed and stops using it.  Obviously, you
need more than 2, so that system can conclude which one failed.  Expensive,
very specialized and rare stuff.  You are not going to do something like that
for database or file server ;-)

Anyhow, the processor failures are so rare, that for vast majority of
applications it does not make economical sense to invest into CPU 
hot-swappable
systems.  Look at it as additional nice feature of higher-end servers.  
Not the
reason to buy higher-end servers.  Unless of course if we are talking about
life supporting systems, or some other systems where failure could cause loss
of human lifes.

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.