Setting up a working Master/Slave replication requires at least two ArangoDB instances:
master: this is the instance where all data-modification operations should be directed to.
slave: this is the instance that replicates, in an asynchronous way, the data from the master. For the replication to happen, a replication applier has to be started on the slave. The replication applier will fetch data from the master's write-ahead log and apply its operations locally. One or more slaves can replicate from the same master.
Generally, one deploys the master on a machine and each slave on an additional, separate, machine (one per slave). In case the master and the slaves are running on the same machine (tests only), please make sure you use different ports (and data directories) for the master and the slaves.
Please install the master and the slaves as they were, separate, single instances. There are no specific differences, at this stage, between a master a slave and a single instance.
Once the ArangoDB master and slaves have been deployed, the replication has to be started on each of the available slaves. This can be done at database level, or globally.
For further information on how to set up the replication in master/slave environment, please refer to this Section.