If, in a replication group, the expected DNS scheduling sequence is different from the site sequence, you can perform the operations described in this section to change the DNS scheduling sequence.
The DNS scheduling sequence can be modified only for the default cluster.
When you access the object service in DNS mode, the DNS server in the storage system schedules requests based on the site sequence in the replication group by default. For example, the site sequence in replication group 1 is <cluster1,cluster2>, and both of the clusters are of the active role. In DNS mode, application requests are scheduled to <cluster1>. If <cluster1> is faulty, application requests will be scheduled to <cluster2>.
If the expected DNS scheduling sequence is different from the site sequence in the replication group, perform the following steps to change the sequence.
If applications access replication buckets in DNS mode, changing the DNS scheduling sequence in the replication group may cause objects written before the change to be temporarily unreadable in a new cluster. For example, if the DNS scheduling sequence in replication group 1 is <cluster1,cluster2>, application requests are scheduled to cluster1. After the DNS scheduling sequence is changed to <cluster2,cluster1>, new application requests are scheduled to cluster2. As a result, objects that have been written to cluster1 may fail to be read before replication is complete.