> Adrian Chadd wrote: >>> Looks good. Just missing a cf.data.pre write-up for squid.conf >>> Adrians call on the commit though. > > Added the writeup to cf.data.pre, and I've also added a fall back, so if > name= hasn't been specified on the config line, it will use the old > hashing mechanism (out of interest, is there a reason p->name inherits > p->host if name= is not specified?).
Backwards compatibility. Before the name option the IP/fqdn was the UID of the cache peer. There are still many squid.conf out there using only a few unique peers and no name= at all. > > Latest patch is: http://www.squid-cache.org/bugs/attachment.cgi?id=1550 > >> I don't know enough about CARP or Hashing algorithms to know if thats >> a "correct" implementation or not. I'd like others' opinions. > > I agree with Adrian, that before it gets committed, we should wait and > see if anyone comes up with any reasons why it shouldn't go in. > Definitely thin ice here as far as whether we're going to break anything > outside of squid if it gets enabled. Amos
