Hi!

In your configuration I would strongly recommend to use one dedicated link 
for DRBD, the direct cross-link would be suited best. Thus you will avoid 
both network-saturation on your outside connection and a single point of 
failure. If one of your nodes in such a scenario looses its connection to 
the outside interface it will still get replicated data, protecting the 
data against disk failure on the other node, while on the other hand 
handing over all other services to the node, that has still connection to 
the network.
Are you using dopd and DRBD-peer-outdater? If not, start doing so. It 
greatly improves data security in case of a connection loss.
We use this setup quite often in combination with an Oracle-Database. The 
V2 config comes quite handy, when having a more complex set of constraints 
to control. In our case, we have 18 DRBD-Devices and 18 filesystems as 
well as about half a dozen of applications running on the machine. 
(Oracle, Oracle-Listener,, Apache, nagios, and some of our own apps). 
Robert Köppl

_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to