Redis [BUG] Join some new nodes to the cluster error

Describe the bug

When I use "redis-cli --cluster add-node" to join some new nodes to the cluster, an error occurs: "[ERR] No such master ID xxxxx"

To reproduce

The shell script is:

#!/bin/bash
EXIST_NODE_IP=$1
EXIST_NODE_PORT=$2
NEW_MASTER_IP=$3
NEW_MASTER_PORT=$4
NEW_SLAVE_IP=$5
NEW_SLAVE_PORT=$6

if (( $# != 6 )); then
    echo "Usage:"
    echo "    $0 <exist-node-ip> <exist-node-port> <new-master-ip> <new-master-port> <new-slave-ip> <new-slave-port>"
exit 1
fi

NEW_MASTER_ID=`../../src/redis-cli -h $NEW_MASTER_IP -p $NEW_MASTER_PORT  cluster myid`

../../src/redis-cli --cluster add-node $NEW_MASTER_IP:$NEW_MASTER_PORT $EXIST_NODE_IP:$EXIST_NODE_PORT
../../src/redis-cli --cluster add-node $NEW_SLAVE_IP:$NEW_SLAVE_PORT $EXIST_NODE_IP:$EXIST_NODE_PORT  --cluster-slave --cluster-master-id $NEW_MASTER_ID

Expected behavior

[OK] New node added correctly.

Comment From: rclijia

The reason should be: After the new master node joins, execute the "--cluster-slave --cluster-master-id" operation，but the gossip protocol has not been synchronized, and other nodes have not obtained the ID of the new master node.

Comment From: rclijia

To avoid the above problems, use the following command instead:

NEW_MASTER_ID=`redis-cli -h $EXIST_NODE_IP -p $EXIST_NODE_PORT cluster nodes | grep "$NEW_MASTER_IP:$NEW_MASTER_PORT" | awk {'print $1'}`

while [ -z "$NEW_MASTER_ID" ]
do
    echo "can't get master id, try again..."
    sleep 1
    NEW_MASTER_ID=`redis-cli -h $EXIST_NODE_IP -p $EXIST_NODE_PORT cluster nodes | grep "$NEW_MASTER_IP:$NEW_MASTER_PORT" | awk {'print $1'}`
done

However, this process should be implemented in the redis-cli ?