RAC Instance Recovery and Auto Restart HD
When a Database Instance in a RAC Cluster dies abnormally, any other surviving Instance in the Cluster executes Instance Recovery. Here Instance 2 on node2 died abnormally and Instance 1 executed Recovery (applied Rollforward and Rollback using the Redo Logs of Thread 2 [Instance 2] ).
Furtheremore, ClusterWare on the node detects the failure of the Instance (the Instance is one of many resources monitored by it) and restarts it automatically. The CRS log captures the detection of the failure but does NOT log autorestart of the Database Instance --- for the restart, you have to check the alert log of the Instance.