heartbeat crm 资源和资源组故障重启问题？

heartbeat 的crm模式下，一个资源组里的任意一个资源出现故障，其组内的其他任意资源都会stop掉，然后在重启。如何避免这个这样，让其只重启有故障的资源？？？


以下是debug里面抓到的一些信息

pengine[20149]: 2011/01/09_10:30:19 debug: update_action:    * (implies right) Marking action all_stopped mandatory because of ha_drbd.sh_1_stop_0
pengine[20149]: 2011/01/09_10:30:19 debug: update_action:    * Marking action group_1_running_0 un-runnable because of ha_drbd.sh_1_start_0
pengine[20149]: 2011/01/09_10:30:19 debug: update_action:    * Marking action group_1_start_0 un-runnable because of ha_drbd.sh_1_start_0


pengine[27271]: 2011/01/09_10:31:12 debug: update_action:       Ignoring implies left - ha_drbd.sh_2 already stopped
pengine[27271]: 2011/01/09_10:31:12 debug: update_action:       Ignoring implies left - ha_tms.sh_3 already stopped
pengine[27271]: 2011/01/09_10:31:12 debug: update_action:       Ignoring implies left - ha_bms.sh_4 already stopped
pengine[27271]: 2011/01/09_10:31:12 debug: update_action:       Ignoring implies left - IPaddr_192_168_37_100 already stopped


pengine[20149]: 2011/01/07_17:28:10 debug: update_action:       Ignoring implies left - ha_drbd.sh_2 already stopped
pengine[20149]: 2011/01/07_17:28:10 debug: update_action:    * (implies right) Marking action ha_drbd.sh_1_start_0 mandatory because of group_1_start_0
pengine[20149]: 2011/01/07_17:28:10 debug: update_action:       Ignoring implies left - ha_tms.sh_3 already stopped
pengine[20149]: 2011/01/07_17:28:10 debug: update_action:       Ignoring implies left - ha_bms.sh_4 already stopped
pengine[20149]: 2011/01/07_17:28:10 debug: update_action:       Ignoring implies left - IPaddr_192_168_37_100 already stopped

我顶顶顶顶。。。。。。。。。