Heartbeat is now running on a few production servers, which are very important in the manufacturing of many recent digital products. We have several Heartbeat clusters, in conjunction with DRBD, which are mainly constructed in two type in our systems. One is as a two node cluster running beside product lines, serving the equipment manufacturing process. Our other usage is as the load balancer of a big data warehouse server which run the Linux virtual server package. These clusters had run since Sep 2003, with almost no trouble occured.
As of July 2004, real failover has occured twice in our Heartbeat clusters. One is when the facility staff accidently pulled the power supply to the active server. When this happened, all services failed over to standby side successfully. The other is when a real hardware failure occured on active side. Although cluster did not fail over to the standby side automatically because of the nature of the hardware failure, we are happy that we could recover the services in a very short time.