<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 12pt;
font-family:Calibri
}
--></style></head>
<body class='hmmessage'><div dir='ltr'>> Using purely DNS for zone replication (supermaster) is nice and sounds <br><div>> great, but also has disadvantages, e.g. it is not reliable: If the <br>> NOTIFY could not be delivered to the slave, then the slave is <br>> inconsistent. So, you need another mechanism to verify and update slaves <br>> which failed to receive the NOTIFY.<br><br>Does this scenario not hold true for any distributed service which relies on network connectivity?<br><br>I suppose this will happen when the slave node goes down, which is why the plan is to put slave nodes in a high availability configuration like a Solaris 10 with Sun cluster.<br><br>This is why we banged on getting pdns working on Solaris at all costs, so that we could exploit the Sun clustering software in tandem with Oracle real application clusters and ASM, basically getting us very high availability for doing a little extra work in advance.<br><br>Having regular NOTIFY on 53 is our escape hatch: in case we want a different solution, we can switch the storage engine and the OS underneath for anything else we want, when we want, or even mix and match different solutions.<br></div> </div></body>
</html>