Elasticsearch version : 6.8.11
Plugins installed: []
JVM version : 1.8.0_45
OS version : GNU/Linux

Description of the problem including expected versus actual behavior:
First the cluster version is 5.6.8, then we upgrade the cluster to 6.8.11. If we update cluster settings in the new cluster, it will throw the IllegalArgumentException:

"reason": "unknown setting [archived.indices.ttl.interval] please check that any required plugins are installed, or check the breaking changes documentation for removed settings"

The reason is clear, there exists a setting in the ES5 cluster, which is archived in ES6 cluster. After upgrading to ES6, if we update the cluster setting , ES6 server will validate the archived settings:indices.ttl.interval and not ignore it, which will throw exception.

Steps to reproduce:
step 1:
Push the cluster setting to ES5.6.8 cluster:

PUT _cluster/settings/admin/lc_gh_4
{
    "persistent": {
        "indices.ttl.interval":"6000s"
    }
}

step 2:
Upgrade the cluster to 6.8.11
step 3:
Push the cluster setting to the cluster:

PUT _cluster/settings
{
     "persistent": {
      "cluster": {
         "routing": {
            "allocation": {
               "cluster_concurrent_rebalance": "3"
            }
         }
      }
     }
}

ES will throw the exception

 "reason": "unknown setting [archived.indices.ttl.interval] please check that any required plugins are installed, or check the breaking changes documentation for removed settings"