The following procedure specifies how to add a Data Center (DC) to a live Scylla Cluster, in a single data center, multi-availability zone, or multi-datacenter. Adding a DC out-scales the cluster and provides higher availability (HA).
The procedure includes:
Install nodes on the new DC.
Add the new nodes one by one to the cluster.
Update the replication strategy of the selected keyspace/keyspaces to use with the new DC.
Rebuild new nodes
Run full cluster repair
Update the Monitoring stack
Note
Make sure to complete the full procedure before starting to read from the new datacenter. You should also be aware that the new DC by default will be used for reads.
Warning
The node/nodes you add must be clean (no data). Otherwise, you risk data loss. See Clean Data from Nodes below.
Log in to one of the nodes in the cluster, collect the following info from the node:
cluster_name - cat /etc/scylla/scylla.yaml | grep cluster_name
seeds - cat /etc/scylla/scylla.yaml | grep seeds:
endpoint_snitch - cat /etc/scylla/scylla.yaml | grep endpoint_snitch
Scylla version - scylla --version
Authentication status - cat /etc/scylla/scylla.yaml | grep authenticator
Note
If authenticator is set to PasswordAuthenticator - increase the system_auth table replication factor.
For example
ALTER KEYSPACE system_auth WITH REPLICATION = {'class' : 'NetworkTopologyStrategy', 'dc1' : <new_replication_factor>};
It is recommended to set system_auth replication factor to the number of nodes in each DC.
On all client applications, switch the consistency level to LOCAL_* (LOCAL_ONE, LOCAL_QUORUM,etc.) to prevent the coordinators from accessing the data center you’re adding.
Install the new clean Scylla nodes (See Clean Data from Nodes below) on the new datacenter, see Getting Started for further instructions, create as many nodes that you need.
Follow the Scylla install procedure up to scylla.yaml configuration phase.
In the case that the node starts during the installation process follow these instructions.
sudo rm -rf /var/lib/scylla/data
sudo find /var/lib/scylla/commitlog -type f -delete
sudo find /var/lib/scylla/hints -type f -delete
sudo find /var/lib/scylla/view_hints -type f -delete
Procedure
Warning
Make sure all your keyspaces are using NetworkTopologyStrategy. If this is not the case, follow the these instructions to fix that.
For each node in the existing data-center(s) edit the scylla.yaml file to use either
Ec2MultiRegionSnitch - for AWS cloud-based, multi-data-center deployments.
GossipingPropertyFileSnitch - for bare metal and cloud (other than AWS) deployments.
Set the DC and Rack Names.
In the cassandra-rackdc.properties file edit the parameters listed below.
The file can be found under /etc/scylla/.
Ec2MultiRegionSnitch
Ec2MultiRegionSnitch gives each DC and rack default names, the region name defined as datacenter name, and availability zones are defined as racks within a datacenter.
For example:
A node in the us-east-1 region,
us-east is the data center name, and 1 is the rack location.
To change the DC names, do the following:
Edit the cassandra-rackdc.properties file with the preferred datacenter name.
The file can be found under /etc/scylla/
The dc_suffix defines a suffix added to the datacenter name as described below.
For example:
Region us-east
dc_suffix=_1_scylla will be us-east_1_scylla
Or
Region us-west
dc_suffix=_1_scylla will be us-west_1_scylla
GossipingPropertyFileSnitch
dc - Set the datacenter name
rack - Set the rack name
For example:
# cassandra-rackdc.properties
#
# The lines may include white spaces at the beginning and the end.
# The rack and data center names may also include white spaces.
# All trailing and leading white spaces will be trimmed.
#
dc=thedatacentername
rack=therackname
# prefer_local=<false | true>
# dc_suffix=<Data Center name suffix, used by EC2SnitchXXX snitches>
In the existing datacenter(s) restart Scylla nodes one by one.
sudo systemctl restart scylla-server
docker exec -it some-scylla supervisorctl restart scylla
(without restarting some-scylla container)
For each node in the new datacenter edit the scylla.yaml file parameters listed below,
the file can be found under /etc/scylla/.
cluster_name - Set the selected cluster_name.
seeds - IP address of an existing node (or nodes).
listen_address - IP address that Scylla used to connect to the other Scylla nodes in the cluster.
auto_bootstrap - By default, this parameter is set to true, it allows new nodes to migrate data to themselves automatically.
endpoint_snitch - Set the selected snitch.
rpc_address - Address for client connections (Thrift, CQL).
The parameters seeds, cluster_name and endpoint_snitch need to match the existing cluster.
In the new datacenter, set the DC and Rack Names (see step number three for more details).
In the new datacenter, start Scylla nodes one by one using.
sudo systemctl start scylla-server
docker exec -it some-scylla supervisorctl start scylla
(with some-scylla container already running)
Verify that the nodes were added to the cluster using nodetool status.
For example:
$ nodetool status
Datacenter: US-DC
=========================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
-- Address Load Tokens Owns Host ID Rack
UN 54.191.2.121 120.97 KB 256 ? c84b80ea-cb60-422b-bc72-fa86ede4ac2e RACK1
UN 54.191.72.56 109.54 KB 256 ? 129087eb-9aea-4af6-92c6-99fdadb39c33 RACK1
UN 54.187.25.99 104.94 KB 256 ? 0540c7d7-2622-4f1f-a3f0-acb39282e0fc RACK1
Datacenter: ASIA-DC
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
-- Address Load Tokens Owns Host ID Rack
UN 54.160.174.243 109.54 KB 256 ? c7686ffd-7a5b-4124-858e-df2e61130aaa RACK1
UN 54.235.9.159 109.75 KB 256 ? 39798227-9f6f-4868-8193-08570856c09a RACK1
UN 54.146.228.25 128.33 KB 256 ? 7a4957a1-9590-4434-9746-9c8a6f796a0c RACK1
When all nodes are up and running ALTER the following Keyspaces in the new nodes:
Keyspace created by the user (which needed to replicate to the new DC).
System: system_auth, system_distributed, system_traces For example, replicate the data to three nodes in the new DC.
For example:
Before
DESCRIBE KEYSPACE mykeyspace;
CREATE KEYSPACE mykeyspace WITH replication = { 'class' : 'NetworkTopologyStrategy', '<exiting_dc>' : 3};
ALTER Command
ALTER KEYSPACE mykeyspace WITH replication = { 'class' : 'NetworkTopologyStrategy', '<exiting_dc>' : 3, <new_dc> : 3};
ALTER KEYSPACE system_auth WITH replication = { 'class' : 'NetworkTopologyStrategy', '<exiting_dc>' : 3, <new_dc> : 3};
ALTER KEYSPACE system_distributed WITH replication = { 'class' : 'NetworkTopologyStrategy', '<exiting_dc>' : 3, <new_dc> : 3};
ALTER KEYSPACE system_traces WITH replication = { 'class' : 'NetworkTopologyStrategy', '<exiting_dc>' : 3, <new_dc> : 3};
After
DESCRIBE KEYSPACE mykeyspace;
CREATE KEYSPACE mykeyspace WITH REPLICATION = {'class’: 'NetworkTopologyStrategy', <exiting_dc>:3, <new_dc>: 3};
CREATE KEYSPACE system_auth WITH replication = { 'class' : 'NetworkTopologyStrategy', '<exiting_dc>' : 3, <new_dc> : 3};
CREATE KEYSPACE system_distributed WITH replication = { 'class' : 'NetworkTopologyStrategy', '<exiting_dc>' : 3, <new_dc> : 3};
CREATE KEYSPACE system_traces WITH replication = { 'class' : 'NetworkTopologyStrategy', '<exiting_dc>' : 3, <new_dc> : 3};
Run nodetool rebuild on each node in the new datacenter, specify the existing datacenter name in the rebuild command.
For example:
nodetool rebuild -- <existing_data_center_name>
The rebuild ensures that the new nodes that were just added to the cluster will recognize the existing datacenters in the cluster.
Run a full cluster repair, using nodetool repair -pr on each node, or using Scylla Manager ad-hoc repair
If you are using Scylla Monitoring, update the monitoring stack to monitor it. If you are using Scylla Manager, make sure you install the Manager Agent and Manager can access the new DC.
This procedure will help your clients not to connect to the DC you just added.
The example below is for clients using the Java driver. Modify this example to suit your needs.
One way to prevent clients from connecting to the new DC is to temporarily restrict them to only use the nodes which are in the same DC as the clients.
This can be done by ensuring the operations are performed with CL=LOCAL_* (for example LOCAL_QUORUM) and using
DCAwareRoundRobinPolicy created by calling withLocalDc("dcLocalToTheClient) and withUsedHostsPerRemoteDc(0) on DCAwareRoundRobinPolicy.Builder.
Example of DCAwareRoundRobinPolicy creation:
variable = DCAwareRoundRobinPolicy.builder()
.withLocalDc("<name of DC local to the client being configured>")
.withUsedHostsPerRemoteDc(0)
.build();
Once the new DC is operational you can remove “withUsedHostsPerRemoteDc(0)” from the configuration and change CL to the previous value.