CAS is broken

Dustin Sallings dustin at
Sun Nov 11 20:15:02 UTC 2007

On Nov 11, 2007, at 9:39, Tomash Brechko wrote:

> However internally (and this was mentioned in the thread) CAS value is
> actually a pointer (cast to uint32_t for some reason).  So I thought,
> how we know the pointer won't be reused?

	The pointers are definitely reused.  I found that in my testing a  
while back, but I thought we had fixed it by having a separate  
sequence for generating CAS identifiers.

Dustin Sallings

More information about the memcached mailing list