<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'>Hi all,<br>I have a working setup with 2 CentOS 6.5 hosts and DRBD 8.4.5 with a dedicated DRBD-only connection created by bonding (round-robin, i.e. "mode=balance-rr") dual-port 10G Ethernet cards connected back to back (fiber cables with no switches interposed).<br><br>Currently it's working with the usual "miimon=100" option and so uses interface carrier detection (I suppose) to detect faults, but given the peculiar setup (back-to-back cables) I wonder whether it could be better to use something like "arp_interval=100 arp_ip_target=x.y.z.t"<br><br>A search did not reveal any definitive answer on the subject (but I understand that "usual" enterprise deployments tend to always involve switches...), so I thought I could ask here hoping to find suggestions/case-histories.<br><br>Many thanks in advance.<br><br>Regards,<br>Giuseppe<br> </div></body>
</html>