解决kill大事务导致的undo争用

wait for a undo record是cpu并行恢复的资源争用undo记录的等待事件

调整相关参数

fast_start_parallel_rollback 为false，进行串行恢复。

调整参数  fast_start_parallel_rollback 为false，这样回滚的进程就只会有一个，速度有时候反而比并行回滚还快。

SQL>show parameter fast_start_parallel_rollback

SQL>alter system set fast_start_parallel_rollback=false;

查看回滚进度

select  undoblockstotal "Total", undoblocksdone "Done", undoblockstotal-undoblocksdone
      "ToDo",decode(cputime,0,'unknown',to_char(sysdate+(((undoblockstotal-undoblocksdone)
       / (undoblocksdone / cputime)) / 86400),'yyyy-mm-dd hh24:mi:ss'))
        "Estimated time to complete"，to_char(sysdate,'yyyy-mm-dd hh24:mi:ss')
        from v$fast_start_transactions;