After reading up on new DNS features and some clarifications I've
asked people, I have to agree SRV records are something I definitely
have to implement, and John is totally right, but virtual ips are
still a more full proof solution, there are times that SRV records
won't work, using a cluster it will always work. I think that the very
best solution is to do both, and neither are complex to implement  ;)

Regards,
Shidan

On 6/15/06, John Lange <[EMAIL PROTECTED]> wrote:
At the risk of sounding redundant; you don't need any of this fancy
clustering, virtual IPs etc. etc.

All you need is to use SRV records in your DNS and tell your client
hardware to use SRV lookups.

John

On Thu, 2006-06-15 at 10:34 -0300, [EMAIL PROTECTED] wrote:
> Has anyone looked at using the CARP protocol to share virtual IPs between
> two servers?  This was developed on OpenBSD but looks like it has been
> ported to Linux as Ucarp.  I use it for our two OpenBSD firewalls and the
> switchover to the secondary server is instantaneous.  You would need a
> second NIC in each server to run the pfsync protocol between them.  This is
> great for maintenance too.  You just down the server you want to work on.
> Trickier if a T1 is attached of course.
>
> I don't know if Asterisk is working well on OpenBSD because of Zap driver
> problems but I would like to give it a shot some time.  I noticed that 
Asterisk
> is in the packages.
>
> Peter M.
>
> > The problem is those pesky TTLs. I believe the big resi providers (ie: 
Sympatico & Rogers) are
> > manipulating them, so even though you have yours set to 1 hour or whatever, 
your customers
> > who get their DNS via their providers will be given something that's almost 
a week old. I'm not
> > sure if they've stopped this practice or not, but last time I checked, 
that's the way they operated.
> >
> > Of course to get around it, you could easily ship your devices with your 
DNS servers set, but most
> > of that info is user configurable so they could bork it on you.
> >
> > - Ian
> >
> > On 6/14/06, Nabeel Jafferali <[EMAIL PROTECTED]> wrote:
> >     Simon:
> >
> >     Although it is true that DNS is cached and round-robin DNS would cause 
a 50%
> >     failure of calls in case one of the servers failed, DNS SRV is a 
different
> >     animal.
> >
> >     "Smart" SIP endpoints know to look at all SRV records and the RFC 
provides
> >     for priorities, so caching the DNS entry is not an issue.
> >
> >     Nabeel
> >
>
> ********************************************************
> Peter MacFarlane, ACP
> Network Administration &  Programming
> Target Call Center/ Message Centre P.E.I.
> *****************************************************************
> OpenBSD's PF Firewall: Now available with CARP Failover.
> Nothing to do with fish, but everything to do with security!
> *****************************************************************
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]


Reply via email to