[Linux-cluster] CS4 Virtual Ip address Service re-starts

[Patton, Matthew F, CTR, OSD-PA&E]

Classification: UNCLASSIFIED

> [mailto:linux-cluster-bounces at redhat.com]On Behalf Of Lon Hohberger

> This is solved in current CVS - and will be solved with U3.  
> It's got to do with the router-pinging code being broken.

on a tangentially related note. I've noticed that if I define a default
gateway in at least 2 different ifcfg-eth? files since the interfaces are on
different nets, the routing breaks. The hack is to change the default
routing handler script so it doesn't clobber other entries. It would seem
somebody forgot that machines can be dual-homed and each have interface
specific default gateways. If I bind a socket to one or the other interface
the outbound packet needs to follow the default gateway for that interface.

Ok, so this may not be an altogether common scenario but shouldn't it be
allowed? Our heartbeat/backdoor management network is on a different net
from the customer side and each have their own notions of what the default
should be.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20060330/6a27ce43/attachment.htm>