Create Local Cluster

1. Create a 3 node cluster with replication factor 3

We will use the yb-ctl utility located in the bin directory of the YugaByte DB package to create and administer a local cluster. The default data directory used is /tmp/yugabyte-local-cluster. You can change this directory with the --data_dir option. Detailed output for the create command is available in yb-ctl Reference.

$ ./bin/yb-ctl create

You can now check /tmp/yugabyte-local-cluster to see node-i directories created where i represents the node_id of the node. Inside each such directory, there will be 2 disks disk1 and disk2 to highlight the fact that YugaByte DB can work with multiple disks at the same time. Note that the IP address of node-i is by default set to 127.0.0.i.

2. Check cluster status with yb-ctl

Run the command below to see that we now have 3 yb-master processes and 3 yb-tserver processes running on this localhost. Roles played by these processes in a YugaByte cluster (aka Universe) is explained in detail here.

$ ./bin/yb-ctl status

2018-05-10 09:06:58,821 INFO: Server is running: type=master, node_id=1, PID=5243, admin service=http://127.0.0.1:7000
2018-05-10 09:06:58,845 INFO: Server is running: type=master, node_id=2, PID=5246, admin service=http://127.0.0.2:7000
2018-05-10 09:06:58,871 INFO: Server is running: type=master, node_id=3, PID=5249, admin service=http://127.0.0.3:7000
2018-05-10 09:06:58,897 INFO: Server is running: type=tserver, node_id=1, PID=5252, admin service=http://127.0.0.1:9000, cql service=127.0.0.1:9042, redis service=127.0.0.1:6379, pgsql service=127.0.0.1:5433
2018-05-10 09:06:58,922 INFO: Server is running: type=tserver, node_id=2, PID=5255, admin service=http://127.0.0.2:9000, cql service=127.0.0.2:9042, redis service=127.0.0.2:6379, pgsql service=127.0.0.2:5433
2018-05-10 09:06:58,945 INFO: Server is running: type=tserver, node_id=3, PID=5258, admin service=http://127.0.0.3:9000, cql service=127.0.0.3:9042, redis service=127.0.0.3:6379, pgsql service=127.0.0.3:5433

3. Check cluster status with Admin UI

Node 1’s master Admin UI is available at http://127.0.0.1:7000 and the tserver Admin UI is available at http://127.0.0.1:9000. You can visit the other nodes’ Admin UIs by using their corresponding IP addresses.

3.1 Overview and Master status

Node 1’s master Admin UI home page shows that we have a cluster (aka a Universe) with Replication Factor of 3 and Num Nodes (TServers) as 3. The Num User Tables is 0 since there are no user tables created yet. YugaByte DB version number is also shown for your reference.

The Masters section highlights the 3 masters along with their corresponding cloud, region and zone placement.

3.2 TServer status

Clicking on the See all nodes takes us to the Tablet Servers page where we can observe the 3 tservers along with the time since they last connected to this master via their regular heartbeats. Since there are no user tables created yet, we can see that the Load (Num Tablets) is 0 across all the 3 tservers. As new tables get added, new tablets (aka shards) will get automatically created and distributed evenly across all the available tablet servers.

1. Create a 3 node cluster with replication factor 3

We will use the yb-docker-ctl utility downloaded in the previous step to create and administer a containerized local cluster. Detailed output for the create command is available in yb-docker-ctl Reference.

$ ./yb-docker-ctl create

Clients can now connect to YugaByte DB’s Cassandra-compatible YCQL API at localhost:9042 and to the Redis-compatible YEDIS API at localhost:6379.

2. Check cluster status with yb-docker-ctl

Run the command below to see that we now have 3 yb-master (yb-master-n1,yb-master-n2,yb-master-n3) and 3 yb-tserver (yb-tserver-n1,yb-tserver-n2,yb-tserver-n3) containers running on this localhost. Roles played by these containers in a YugaByte cluster (aka Universe) is explained in detail here.

$ ./yb-docker-ctl status

PID        Type       Node       URL                       Status          Started At          
26132      tserver    n3         http://172.18.0.7:9000    Running         2017-10-20T17:54:54.99459154Z
25965      tserver    n2         http://172.18.0.6:9000    Running         2017-10-20T17:54:54.412377451Z
25846      tserver    n1         http://172.18.0.5:9000    Running         2017-10-20T17:54:53.806993683Z
25660      master     n3         http://172.18.0.4:7000    Running         2017-10-20T17:54:53.197652566Z
25549      master     n2         http://172.18.0.3:7000    Running         2017-10-20T17:54:52.640188158Z
25438      master     n1         http://172.18.0.2:7000    Running         2017-10-20T17:54:52.084772289Z

3. Check cluster status with Admin UI

The yb-master-n1 Admin UI is available at http://localhost:7000 and the yb-tserver-n1 Admin UI is available at http://localhost:9000. Other masters and tservers do not have their admin ports mapped to localhost to avoid port conflicts.

NOTE: Clients connecting to the cluster will connect to only yb-tserver-n1. In case of Docker for Mac, routing traffic directly to containers is not even possible today. Since only 1 node will receive the incoming client traffic, throughput expected for Docker-based local clusters can be significantly lower than binary-based local clusters.

3.1 Overview and Master status

The yb-master-n1 home page shows that we have a cluster (aka a Universe) with Replication Factor of 3 and Num Nodes (TServers) as 3. The Num User Tables is 0 since there are no user tables created yet. YugaByte DB version number is also shown for your reference.

The Masters section highlights the 3 masters along with their corresponding cloud, region and zone placement.

3.2 TServer status

Clicking on the See all nodes takes us to the Tablet Servers page where we can observe the 3 tservers along with the time since they last connected to this master via their regular heartbeats. Additionally, we can see that the Load (Num Tablets) is balanced across all the 3 tservers. These tablets are the shards of the user tables currently managed by the cluster (which in this case is the system_redis.redis table). As new tables get added, new tablets will get automatically created and distributed evenly across all the available tablet servers.

1. Create a 3 node cluster with replication factor 3

Run the following command to create the cluster.

$ kubectl apply -f yugabyte-statefulset.yaml

service "yb-masters" created
statefulset "yb-master" created
service "yb-tservers" created
statefulset "yb-tserver" created

2. Check cluster status

Run the command below to see that we now have two services with 3 pods each - 3 yb-master pods (yb-master-1,yb-master-2,yb-master-3) and 3 yb-tserver pods (yb-tserver-1,yb-tserver-2,yb-tserver-3) running. Roles played by these pods in a YugaByte DB cluster (aka Universe) is explained in detail here.

$ kubectl get pods

NAME           READY     STATUS              RESTARTS   AGE
yb-master-0    0/1       ContainerCreating   0          5s
yb-master-1    0/1       ContainerCreating   0          5s
yb-master-2    1/1       Running             0          5s
yb-tserver-0   0/1       ContainerCreating   0          4s
yb-tserver-1   0/1       ContainerCreating   0          4s
yb-tserver-2   0/1       ContainerCreating   0          4s

Eventually all the pods will have the Running state.

$ kubectl get pods

NAME           READY     STATUS    RESTARTS   AGE
yb-master-0    1/1       Running   0          13s
yb-master-1    1/1       Running   0          13s
yb-master-2    1/1       Running   0          13s
yb-tserver-0   1/1       Running   1          12s
yb-tserver-1   1/1       Running   1          12s
yb-tserver-2   1/1       Running   1          12s

3. Initialize the Redis-compatible YEDIS API

Initialize Redis-compatible YEDIS API in the YugaByte DB Universe we just setup by running the following yb-admin command.

$ kubectl exec -it yb-master-0 /home/yugabyte/bin/yb-admin -- --master_addresses yb-master-0.yb-masters.default.svc.cluster.local:7100,yb-master-1.yb-masters.default.svc.cluster.local:7100,yb-master-2.yb-masters.default.svc.cluster.local:7100 setup_redis_table

...
I0127 19:38:10.358551   115 client.cc:1292] Created table system_redis.redis of type REDIS_TABLE_TYPE
I0127 19:38:10.358872   115 yb-admin_client.cc:400] Table 'system_redis.redis' created.

Clients can now connect to this YugaByte DB universe using Cassandra and Redis APIs on the 9042 and 6379 ports respectively.

4. Check cluster status via Kubernetes

You can see the status of the 3 services by simply running the following command.

$ kubectl get services

NAME           TYPE           CLUSTER-IP      EXTERNAL-IP   PORT(S)                               AGE
kubernetes     ClusterIP      10.96.0.1       <none>        443/TCP                               10m
yb-master-ui   LoadBalancer   10.102.121.64   <pending>     7000:31283/TCP                        8m
yb-masters     ClusterIP      None            <none>        7000/TCP,7100/TCP                     8m
yb-tservers    ClusterIP      None            <none>        9000/TCP,9100/TCP,9042/TCP,6379/TCP   8m

5. Check cluster status with Admin UI

In order to do this, we would need to access the UI on port 7000 exposed by any of the pods in the yb-master service (one of yb-master-0, yb-master-1 or yb-master-2). In order to do so, we find the URL for the yb-master-ui LoadBalancer service.

$ minikube service  yb-master-ui --url

http://192.168.99.100:31283

Now, you can view the yb-master-0 Admin UI is available at the above URL.

5.1 Overview and Master status

The yb-master-0 home page shows that we have a cluster (aka a Universe) with Replication Factor of 3 and Num Nodes (TServers) as 3. The Num User Tables is 0 since there are no user tables created yet. YugaByte DB version is also shown for your reference.

The Masters section highlights the 3 masters along with their corresponding cloud, region and zone placement.

5.2 TServer status

Clicking on the See all nodes takes us to the Tablet Servers page where we can observe the 3 tservers along with the time since they last connected to this master via their regular heartbeats. Additionally, we can see that the Load (Num Tablets) is balanced across all the 3 tservers. These tablets are the shards of the user tables currently managed by the cluster (which in this case is the system_redis.redis table). As new tables get added, new tablets will get automatically created and distributed evenly across all the available tablet servers.