Failed to execute `flush synced`

Elasticsearch version 6.2.3
I execute curl -XPOST ''http://192.168.2.201:9203/myindex/_flush/synced" .
The result shows as follows:

{
	"_shards":{
		"total": 10,
		"successful": 6,
		"failed": 4
   },
	"myindex" : {
		"total": 10,
		"successful": 6,
		"failed": 4,
		"failures": [
		{
			"shard": 3,
			"reason": "no commit id from pre-sync flush",
			"routing": {
				"state": "STARTED",
				"primary": false,
				"node": "YK3h1SM5SZK-JloBdVCg",
				"relocating_node": null,
				"shard": 3,
				"index": "myindex",
				"allocation_id": {
					"id": "bobR0AfWQiWv24S6GzblkA"
				}
			}
		}
		···
		···
    ]
    }
}

The myindex has 5 primary shards and 1 replica. Each shards is aboud 30GB. And cluster status is GREEN.
I am not clear how to fix this problem.