If a start operation of a resource fails, the failcount of this resource will be set to INFINITY. This translates to "never start the resource here again until the admin cleans up". This behaviour can be changed by the use of the crm attribute "start-failure-is-fatal":
start-failure-is-fatal [boolean: default=true]: Always treat start failures as fatal
This was the old default. However when set to FALSE, the cluster will instead use the resource's failcount and value for resource-failure-stickiness
This information provided courtesy of the Linux-HA project at http://linux-ha.org/