Feature: Topology: replicaset parameter

Add --replicaset parameter to the binary.

• When two instances of picodata are run with the same --cluster-id and --replicaset parameters then these instances should become members of the same Tarantool replicaset.
• The instances' replicaset_uuid and instance_uuid should be generated and committed to the cluster-wide config using Raft.
• The instance_id of any instance should be associated with it's replicaset in the cluster-wide config. The corresponding record in the cluster-wide config is written when the instance is started as part of a cluster for the first time.
• When all data files of an instance of an existing replicaset are deleted and the instance is restarted with the same --replicaset parameter, then the instance should start in the same replicaset. The instance should receive and reuse the replicaset_uuid and instance_uuid from the cluster-wide config.
• If the --replicaset parameter for the instance is not the same as the one stored in the global cluster-wide config then the instance should stop with an error.
• When an instance with an instance_id which has previously been part of a replicaset is restarted with a --replicaset that mismatches the one from the global cluster-wide config, then the instance should stop with an error describing the mismatch of the parameter. This situation should be considered a configuration/parameter value error.

Motivation

• Manual topology configuration before automatic topology is implemented.
• Complex topology setup for advanced users.

changed milestone to %Release 0.1 - Demo

added 1 deleted label

marked this issue as related to #60 (closed)

marked this issue as related to #51 (closed)

marked this issue as related to #50 (closed)

marked this issue as related to #69 (closed)

marked this issue as related to #52 (closed)

marked this issue as related to #85 (closed)

marked this issue as related to #86 (closed)

The instance id an instance of a replicaset should be associated with the replicaset in the cluster-wide config.

@sv, could you reword it please, it's confusing.

done

When all data files of an instance of an existing replicaset are deleted and the instance is restarted, then the instance should start in the same replicaset. The instance should receive and reuse the replicaset_uuid and instance_uuid from the cluster-wide config.

The instance is restarted with what command-line arguments? Same I guess. What should happen if command-line arguments are different?

updated:

If the --replicaset parameter for the instance is not the same as the one stored in the global cluster-wide config then the instance should stop with an error.

changed the description

closed