When there are multiple shard exceptions (caused by multiple node exceptions), there will be a large number of failed shard requests sent to the master. At this time, the master will be overwhelmed by such a large number of requests and cannot handle normal requests; in this case Only valid abnormal requests should be sent to the master, and other repeated requests can be discarded, so that the master can handle the abnormal requests without being busy processing a large number of repeated requests;
Relates to Avoid sending duplicate remote failed shard requests #31313