When tidb pushes down an insert request, the node where the region is located crashes during processing. What is the subsequent processing flow of the upper-layer tidb?

When tidb pushes down an insert request, the node where the region is located crashes during processing. What is the subsequent processing flow of the upper-layer tidb? Here are my assumptions

  1. tidb will receive a connection timeout or write failure feedback
  2. tidb will wait for a period of time, query the region information to PD again, and obtain the new node address (here, How to determine when to initiate a query? The region information may not be updated yet)
  3. tidb keeps repeating step 2 until it obtains the new leader node or time out. if tidb obtains new leader node, then pushes down the command to the new node again