[distcc] PATCH: multi A records / very small timeout on connect

[Brad Hards]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Tue, 27 May 2003 17:00 pm, Martin Pool wrote:
> mDNS Rendezvous technologies possible avoid all three of them -- any
> machine can register to join a group without needing to control the
> nameserver.
mDNS has some problems with scaling to a very large number of hosts, because 
you need to check each host to see if it matches what you need, because:

> Whether that's a good thing or not depends on your point of view.
> Perhaps you don't want your jobs distributed to random machines that
> just happen to be on your broadcast domain.
There is a failure mode based on not specifying the compiler sufficiently, 
which is why you'd need to have DNS service discovery use a system like iChat 
does. Use mDNS to lookup SRV records, then use a PTR record / TXT record to 
specify which compilers are actually installed.

That is why I previously suggested using Service Location Protocol V2. That 
would eventually allow you to do something like "give me a list of machines 
with g++ v3.2 for ppc".

I did some cursory work on this a while back (patch probably won't apply, and 
it doesn't do enough to actually make things work properly, but I'll send it 
if anyone wants to play with this). The hard bit is figuring out what to 
register, and how to allocate tasks to machines in a smart way.

Brad
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE+1fckW6pHgIdAuOMRAoNDAJ9ht+l+MlowIuWrLscGjz8wJhbgMQCgihYl
467B01+xCOYXUehyWrBx88o=
=hIO0
-----END PGP SIGNATURE-----