Apologies for the delayed response. The following schedule should give you full cluster backups at minute 0 past every 6th hour.
CREATE SCHEDULE <label>
FOR BACKUP INTO <location>
RECURRING '0 */6 * * *' FULL BACKUP ALWAYS
There are further configurable schedule options you can append to the query above using a
WITH SCHEDULE OPTIONS clause as shown in our docs - CREATE SCHEDULE FOR BACKUP | CockroachDB Docs
Regarding retention of backups, we do not delete files once they have been written to
<location> by the backup. You would have to either manually delete backups every day or set an appropriate (24h) retention period on the bucket being backed up to.
Backup stores its data in SST files which are compressed.
To RESTORE a cluster from a cluster backup one needs to run a
RESTORE FROM <directory containing the backup>
SQL query on a single node. The execution itself will be internally distributed to all nodes by CockroachDB. You do not need to shut down the node, you require an active SQL connection to be able to run the RESTORE query. The only requirement is that the cluster does not have any user-created databases or tables when running a full cluster restore, otherwise the job will fail. Our RESTORE docs have all of this information for future reference - RESTORE | CockroachDB Docs