The figure below shows the I/O operations when an I/O delay has occured on all slices that make up the mirror volume if system panic mode is activated.
When the specified I/O response time is exceeded, the I/O of disk1 and disk2 are timed out and the GDS I/O Monitoring Option panics the system. Cluster failover occurs because of this. The time necessary for failover is shorter than if system panic mode is disabled.
For cluster systems, it is recommended to set the system panic mode to activate.
Figure 8.2 I/O Response Time Assurance (When an I/O response delay occurs on all slices of the mirror volume)
GDS: Global Disk Services
Information
If system panic mode is set to disabled, if an I/O response delay or I/O error occurs on all the slices making up the mirror system, or if the class was closed, an error occurs in the volume stop processing on the previous operating node while cluster failover processing is being done and because of the double fault RMS panics the previous operating node.
