Normally an HA system is designed to eliminate all SPOF[1]s, but is not designed to deal with multiple simultaneous failures. Therefore, a multiple simultaneous failure will likely cause a failure with a very long repair time, since it cannot be automatically recovered by the system.
| [1] | http://www.linux-ha.org/SPOF |
| [2] | http://www.linux-ha.org/MTBF |
| [3] | http://www.linux-ha.org/MTTR |
This information provided courtesy of the Linux-HA project at http://linux-ha.org/