Franck Pachot for YugabyteDB Distributed PostgreSQL Database

Posted on Apr 11, 2022 • Edited on Aug 17

The cost of OKE for YugabyteDB - 2 - Install from Helm chart

#yugabytedb #kubernetes #oracle #cloud

In the previous post, I've created a Kubernetes Cluster on the Oracle Cloud. In this post I'll install YugabyteDB

Namespace

I'm creating a namespace to install YugabyteDB

dev@cloudshell:~ (uk-london-1)$

 kubectl create namespace yb-demo

namespace/yb-demo created

Storage class

I'm changing the default StorageClass to oci-bv, the recommended Container Storage Interface (CSI) volume plugin, rather than the default oci (the FlexVolume one). I need this also to ensure that the block volume are in the same availability domain as the compute instance of the worker.

dev@cloudshell:~ (uk-london-1)$

 kubectl patch storageclass oci     \
 -p '{"metadata": {"annotations":{"storageclass.beta.kubernetes.io/\
is-default-class":"false"}}}'

storageclass.storage.k8s.io/oci patched

dev@cloudshell:~ (uk-london-1)$

 kubectl patch storageclass oci-bv \
 -p '{"metadata": {"annotations":{"storageclass.kubernetes.io/\
is-default-class":"true"}}}'

storageclass.storage.k8s.io/oci-bv patched

Result:

dev@cloudshell:~ (uk-london-1)$

 kubectl get sc

NAME               PROVISIONER                       RECLAIMPOLICY   VOLUMEBINDINGMODE      ALLOWVOLUMEEXPANSION   AGE
oci                oracle.com/oci                    Delete          Immediate              false                   1h
oci-bv (default)   blockvolume.csi.oraclecloud.com   Delete          WaitForFirstConsumer   true                    1h

Helm chart

I'll install the latest version from the Helm Chart:

dev@cloudshell:~ (uk-london-1)$

 helm repo add yugabytedb https://charts.yugabyte.com

"yugabytedb" already exists with the same configuration, skipping

dev@cloudshell:~ (uk-london-1)$

 helm repo update

Hang tight while we grab the latest from your chart repositories...
...Successfully got an update from the "kubernetes-dashboard" chart repository
...Successfully got an update from the "yugabytedb" chart repository
Update Complete. ⎈Happy Helming!⎈

dev@cloudshell:~ (uk-london-1)$

 helm search repo yugabytedb/yugabyte

NAME                    CHART VERSION   APP VERSION     DESCRIPTION                                       
yugabytedb/yugabyte     2.13.0          2.13.0.0-b42    YugabyteDB is the high-performance distributed ...

YugabyteDB Configuration

I'll customize two things here from the values.yaml:

The storage (I'll put 1 TeraByte for each PersistentVolumeClaim)
The placement info (YugabyteDB must know on which Availability Domain it runs to ensure correct balancing of leaders and followers to be resilent to one AD failure).

The default is Replication Factor RF=3.

cat > yb-demo.yaml <<'YAML'
storage:
  master:
    size: 1Ti
  tserver:
    size: 1Ti
gflags:
  master:
    placement_cloud: $(curl -s http://169.254.169.254/opc/v1/instance/regionInfo/realmDomainComponent)
    placement_region: $(curl -s http://169.254.169.254/opc/v1/instance/region)
    placement_zone: $(curl -s http://169.254.169.254/opc/v1/instance/ociAdName)
  tserver:
    placement_cloud: $(curl -s http://169.254.169.254/opc/v1/instance/regionInfo/realmDomainComponent)
    placement_region: $(curl -s http://169.254.169.254/opc/v1/instance/region)
    placement_zone: $(curl -s http://169.254.169.254/opc/v1/instance/ociAdName)
YAML

Note that some screenshots later have been taked without this customization and show 50GB storage (the minimum in OKE) and cloud1/zone1/rack1 default placement. The right placement info is here: https://twitter.com/FranckPachot/status/1511976648807260162

YugabyteDB Install

The install takes a few minutes:

dev@cloudshell:~ (uk-london-1)$

 helm install yb-demo yugabytedb/yugabyte --namespace yb-demo \
 -f yb-demo.yaml --wait

NAME: yb-demo
LAST DEPLOYED: Wed Apr  6 08:40:02 2022
NAMESPACE: yb-demo
STATUS: deployed
REVISION: 1
TEST SUITE: None
NOTES:
1. Get YugabyteDB Pods by running this command:
  kubectl --namespace yb-demo get pods

2. Get list of YugabyteDB services that are running:
  kubectl --namespace yb-demo get services

3. Get information about the load balancer services:
  kubectl get svc --namespace yb-demo

4. Connect to one of the tablet server:
  kubectl exec --namespace yb-demo -it yb-tserver-0 bash

5. Run YSQL shell from inside of a tablet server:
  kubectl exec --namespace yb-demo -it yb-tserver-0 -- /home/yugabyte/bin/ysqlsh -h yb-tserver-0.yb-tservers.yb-demo

6. Cleanup YugabyteDB Pods
  For helm 2:
  helm delete yb-demo --purge
  For helm 3:
  helm delete yb-demo -n yb-demo
  NOTE: You need to manually delete the persistent volume
  kubectl delete pvc --namespace yb-demo -l app=yb-master
  kubectl delete pvc --namespace yb-demo -l app=yb-tserver

Pods

Here are the pods created in my yb-demo namespace:

dev@cloudshell:~ (uk-london-1)$

 kubectl get pods -n yb-demo -o wide

NAME           READY   STATUS    RESTARTS   AGE   IP             NODE          NOMINATED NODE   READINESS GATES
yb-master-0    2/2     Running   0          12m   10.244.0.4     10.0.10.51    <none>           <none>
yb-master-1    2/2     Running   0          12m   10.244.1.3     10.0.10.152   <none>           <none>
yb-master-2    2/2     Running   0          12m   10.244.0.131   10.0.10.103   <none>           <none>
yb-tserver-0   2/2     Running   0          12m   10.244.0.132   10.0.10.103   <none>           <none>
yb-tserver-1   2/2     Running   0          12m   10.244.1.4     10.0.10.152   <none>           <none>
yb-tserver-2   2/2     Running   0          12m   10.244.0.5     10.0.10.51    <none>           <none>

I have 3 master which are the control plane of the YugabyteDB cluster, and 3 tserver which are the data plane. 3 is the minimum because, for high availability, the cluster is in replication factor RF=3 (You can change that in the helm install with --set gflags.tserver.replication_factor=5 for example). This means 3 master where one is the leader and the two others are follower, ready to become leader in case of an Availability Domain outage. And at least 3 tserver for the same reason, be we can have more to distribute connections, load and data to many nodes.

Services

The master and tserver expose a web console, and the SQL endpoints, through a headless service ClusterIP that is accessible though a LoadBalancer:


dev@cloudshell:~ (uk-london-1)$
 kubectl get services -n yb-demo -o wide

NAME                 TYPE           CLUSTER-IP      EXTERNAL-IP       PORT(S)                                                                      AGE   SELECTOR
yb-master-ui         LoadBalancer   10.96.187.203   141.147.107.246   7000:31290/TCP                                                               17m   app=yb-master
yb-masters           ClusterIP      None            <none>            7000/TCP,7100/TCP                                                            17m   app=yb-master
yb-tserver-service   LoadBalancer   10.96.122.104   141.147.106.110   6379:32100/TCP,9042:30198/TCP,5433:31343/TCP                                 17m   app=yb-tserver
yb-tservers          ClusterIP      None            <none>            9000/TCP,12000/TCP,11000/TCP,13000/TCP,9100/TCP,6379/TCP,9042/TCP,5433/TCP   17m   app=yb-tserver

Port 7000 is the master web console, which I can access by its extenal IP, which I've put on the public network for this lab: http://141.147.107.246:7000

for the moment, the database is empty but in the next post I'll run some SQL. The PostgreSQL-compatible enpoint is exposed by the tserverload balancer: postgresql://141.147.106.110:5433/yugabyte:

$ psql postgresql://141.147.106.110:5433/yugabyte
psql (13.5, server 11.2-YB-2.13.0.0-b0)
Type "help" for help.

yugabyte=# select * from yb_servers();

                        host                        | port | num_connections | node_type | cloud  |   region    | zone  |                     public_ip
----------------------------------------------------+------+-----------------+-----------+--------+-------------+-------+----------------------------------------------------
 yb-tserver-1.yb-tservers.yb-demo.svc.cluster.local | 5433 |               0 | primary   | cloud1 | datacenter1 | rack1 | yb-tserver-1.yb-tservers.yb-demo.svc.cluster.local
 yb-tserver-0.yb-tservers.yb-demo.svc.cluster.local | 5433 |               0 | primary   | cloud1 | datacenter1 | rack1 | yb-tserver-0.yb-tservers.yb-demo.svc.cluster.local
 yb-tserver-2.yb-tservers.yb-demo.svc.cluster.local | 5433 |               0 | primary   | cloud1 | datacenter1 | rack1 | yb-tserver-2.yb-tservers.yb-demo.svc.cluster.local

I have queried the yb_servers() function that returns the information about all nodes. Note that the public_ip here is the address within the Kubernetes cluster. From outside, we use the load balancers that have been automatically created by OKE, also tagged with OKEclusterName: yugabytedb:

On is the master console, 141.147.107.246, listening on port 7000:

The other is the tserver service, 141.147.106.110, listening on:

5433 the YSQL endpoint (the PostgreSQL-compatible API)
9042 the YCQL endpoint (the Cassandra-compatible API)
6379 the YEDIS enspoint (the REDIS-compatible API)

Because YugabyteDB, in addition to be shared-nothing distributed SQL database, is also multi-API.

I'll use the PostgreSQL compatible API in the next post to generate some load. My goal is to use some storage and network transfer to evaluate the cost on the Oracle Cloud

DEV Community

The cost of OKE for YugabyteDB - 2 - Install from Helm chart

Namespace

Storage class

Helm chart

YugabyteDB Configuration

YugabyteDB Install

Pods

Services

Top comments (0)

Read next

Migrating AWS Organizations: How I Did It and Why

Navigating Through the Latest AWS Pricing Increase: What You Need to Know

Run Kubernetes Like a Pro—Without the Expertise! Introducing EKS Auto Mode

Get thread dump of Java container running on JRE Docker image