» Bootstrapping a Datacenter
An agent can run in both client and server mode. Server nodes are responsible for running the consensus protocol and storing the cluster state. The client nodes are mostly stateless and rely heavily on the server nodes.
Before a Consul cluster can begin to service requests, a server node must be elected leader. Thus, the first nodes that are started are generally the server nodes. Bootstrapping is the process of joining these initial server nodes into a cluster.
The recommended way to bootstrap is to use the
configuration option. This option informs Consul of the expected number of
server nodes and automatically bootstraps when that many servers are available. To prevent
inconsistencies and split-brain situations (that is, clusters where multiple servers consider
themselves leader), all servers should either specify the same value for
or specify no value at all. Only servers that specify a value will attempt to bootstrap the cluster.
We recommend 3 or 5 total servers per datacenter. A single server deployment is highly discouraged as data loss is inevitable in a failure scenario. Please refer to the deployment table for more detail.
Suppose we are starting a 3 server cluster. We can start
Node B, and
Node C with each
-bootstrap-expect 3 flag. Once the nodes are started, you should see a message like:
[WARN] raft: EnableSingleNode disabled, and no known peers. Aborting election.
This indicates that the nodes are expecting 2 peers but none are known yet. To prevent a split-brain scenario, the servers will not elect themselves leader.
» Creating the cluster
To trigger leader election, we must join these machines together and create a cluster. There are multiple options for joining the machines:
- Manually specified list of machines with -join and start_join options
- Manually specified list of machines with -retry-join option
- Automatic joining by tag for supported cloud environments with the -retry-join option
Choose the method which best suits your environment and specific use case.
» Manually Creating a Cluster
To manually create a cluster, access one of the machines and run the following:
$ consul join <Node A Address> <Node B Address> <Node C Address> Successfully joined cluster by contacting 3 nodes.
Since a join operation is symmetric, it does not matter which node initiates it. Once the join is successful, one of the nodes will output something like:
[INFO] consul: adding server foo (Addr: 127.0.0.2:8300) (DC: dc1) [INFO] consul: adding server bar (Addr: 127.0.0.1:8300) (DC: dc1) [INFO] consul: Attempting bootstrap with nodes: [127.0.0.3:8300 127.0.0.2:8300 127.0.0.1:8300] ... [INFO] consul: cluster leadership acquired
» Verifying the Cluster
As a sanity check, the
consul info command
is a useful tool. It can be used to verify
raft.num_peers is now 2,
and you can view the latest log index under
consul info on the followers, you
raft.last_log_index converge to the same value once the
leader begins replication. That value represents the last log entry that
has been stored on disk.
Now that the servers are all started and replicating to each other, all the remaining clients can be joined. Clients are much easier as they can join against any existing node. All nodes participate in a gossip protocol to perform basic discovery, so once joined to any member of the cluster, new clients will automatically find the servers and register themselves.
Note: It is not strictly necessary to start the server nodes before the clients; however, most operations will fail until the servers are available.
» Manual Bootstrapping
Manual bootstrapping is not recommended as it is more error-prone than automatic bootstrapping