gpg: failed to create temporary file '/var/lib/lurker/.#lk0x58245100.hantslug.org.uk.1957': Permission denied
gpg: keyblock resource '/var/lib/lurker/pubring.gpg': Permission denied
gpg: Signature made Fri May 16 16:12:59 2008 BST
gpg: using DSA key 2099B64CBF15490B
gpg: Can't check signature: No public key
Hi,
On Fri, May 16, 2008 at 03:38:19PM +0100, Stephen Rowles wrote:
> Just to be clear, are you suggesting that applications should use an IP
> address rather than a DNS name or am I misunderstanding?
I am suggesting that if as in the original poster's case, you have a
web server that talks to a local DB server, you would not want the web
server to be connecting to a hostname. You would want it connecting
to an IP address. Or at worst a hostname in /etc/hosts. You don't
want it doing DNS lookups.
> If so this isn't a very easy to maintain solution, what if you introduce a
> new subnet, change ip address, move a machine to another server room etc.
In any of these cases you have to change the DNS, so why is changing
an IP address any different? If you're going to say "it will have
to be done in multiple places" well sure, but if you have enough
config files for this to be an issue then I hope you are building
them automatically in which case this also becomes a non-issue.
> Certainly I would want to ensure that my resolver was very reliable and
> have appropriate fail-over in place, but don't use IP addresses, it makes
> networked applications far too dependant on what should be a transparent
> transport layer.
As a rule I would agree on using host names in the majority of
cases, but for some situations the performance implications outweigh
the convenience benefits and the described situation is one of them
in my opinion.
Cheers,
Andy
