Slow response from remaining RAC instance during recovery.
We recently had a maintenance event that required us to reboot one of the two nodes in our RAC cluster. We noticed that for about two or three minutes after the reboot of the node, the response time from instance on the primary node to queries was extremely slow. Almost to the point of looking like it was hung. Running top on the server didn't show any unusual activitiy and all other response times on the server were acceptable.
We believe that this performance drag was probably due to the remaining instance taking over nearly 700 connections from the "failed" node plus doing instance recovery and lock administration. The question we have is if anybody else has noticed this behavior and if there is anything we can do to improve the performance of the system during this time.