I've got it set cluster-wide currently, but I'd really like it to be set to 'true' by default and false for only my Slave DRBD resource.
Using Pacemaker 1.1.12 and Corosync 1.4.8.
start-failure-is-fatal
is a cluster option, so unfortunately, you can only apply it cluster-wide.
Perhaps you could set start-failure-is-fatal=false
cluster-wide, and then define meta migration-threshold=n
for each primitive in cluster to achieve similar results.
n=1 for resources you want to move quickly and n=3 for resources you'd like to recover without moving.