Getting Started TiDB on GKE

See another post also to set up a TiDB on minikube.

To get the basic concept and architecture of TiDB, there are many articles and videos on the Internet, like following:

Set up a TiDB cluster by Kubernetes Operator

Follow the official document for GKE and the official document for kubernetes operator for this document.

ext4 file system configuration

It’s recommended to configure following options for better I/O performance for ext4 file system.

• nodelalloc: Disable delayed allocation. See this option for why it’s enabled in the first place.
• noatime: Disable to store access time on files. See this article for the benefit for this performance improvement.

The options of ext4 file system can be seen in this man document, though there is no noatime option.

Install TiDB CRD and its operator

First, create the namespace at first.

kubectl create namespace tidb-cluster

Then following this document, installing kubernetes operator.

kubectl create -f https://raw.githubusercontent.com/pingcap/tidb-operator/v1.4.0/manifests/crd.yaml
helm repo add pingcap https://charts.pingcap.org/
kubectl create namespace tidb-admin
helm install --namespace tidb-admin tidb-operator pingcap/tidb-operator --version v1.4.0

Confirm an installation on the operator

kubectl get pods --namespace tidb-admin -l app.kubernetes.io/instance=tidb-operator

Create a cluster

Create followings

• the example of a cluster.
• tidb dashboard (for some reasons, it doesn’t work)
• tidb monitoring like prometheus and Grafana

kubectl -n tidb-cluster apply -f
https://raw.githubusercontent.com/pingcap/tidb-operator/v1.4.0/examples/basic/tidb-cluster
kubectl -n tidb-cluster apply -f https://raw.githubusercontent.com/pingcap/tidb-operator/v1.4.0/examples/basic/tidb-dashboard.yaml
kubectl -n tidb-cluster apply -f https://raw.githubusercontent.com/pingcap/tidb-operator/v1.4.0/examples/basic/tidb-monitor.yaml

Confirm the setup

watch kubectl get po -n tidb-cluster

Connect to TiDB

kubectl get svc -n tidb-cluster
kubectl port-forward -n tidb-cluster svc/basic-tidb 14000:4000 > pf14000.out &
mysql --comments -h 127.0.0.1 -P 14000 -u root mysql

Then create a table

use test;
create table hello_world (id int unsigned not null auto_increment primary key, v varchar(32));
select * from information_schema.tikv_region_status where db_name=database() and table_name='hello_world'\G

See the TiDB cluster information

select * from information_schema.cluster_info\G

Check Grafana

kubectl port-forward -n tidb-cluster svc/basic-grafana 3000 > pf3000.out &

Then go to [http://localhost:3000][] to see Grafana.

Check a TiDB Dashboard

I needed to update tidb-operator version to v1.4.2 because it was v1.3.5 and TiDBDashboard wasn’t created.

helm upgrade --namespace tidb-admin tidb-operator pingcap/tidb-operator --version v1.4.2 --set operatorImage='pingcap/tidb-operator:v1.4.2'

Once you make sure your TiDB Operator runs on v1.4.2, then run port-forward and access the TiDB dashboard

kubectl port-forward -n tidb-cluster svc/basic-tidb-dashboard-exposed 12333 > pf12333.out &

Access localhost:12333 on a browser.

If you want to use TopSQL and Continuous Profiling features, then deploy TidbNGMonitoring described in this document

Basic operations

Update the number of TiKV, TiDB, and PD nodes:

kubectl patch -n tidb-cluster tc basic --type merge --patch '{"spec":{"tikv":{"replicas":3}}}'
kubectl patch -n tidb-cluster tc basic --type merge --patch '{"spec":{"tidb":{"replicas":2}}}'
kubectl patch -n tidb-cluster tc basic --type merge --patch '{"spec":{"pd":{"replicas":2}}}'

Increase the amount of storage sizes of Network PV

Follow this document.

kubectl patch -n tidb-cluster tc basic --type merge --patch '{"spec":{"tikv":{"requests":{"storage":"100Gi"}}}}'

Note that Shrinking persistent volumes is not supported for Kubernetes Persistent Volumes.

Load an example data

At first, we’ll load the data from this example.

Before loading it, at first, check a storage engine on TiDB:

mysql> select * from information_schema.engines;
+--------+---------+------------------------------------------------------------+--------------+------+------------+
| ENGINE | SUPPORT | COMMENT                                                    | TRANSACTIONS | XA   | SAVEPOINTS |
+--------+---------+------------------------------------------------------------+--------------+------+------------+
| InnoDB | DEFAULT | Supports transactions, row-level locking, and foreign keys | YES          | YES  | YES        |
+--------+---------+------------------------------------------------------------+--------------+------+------------+
1 row in set (0.04 sec)

Then expand storage sizes of TiKV nodes

kubectl patch -n tidb-cluster tc basic --type merge --patch '{"spec":{"tikv":{"requests":{"storage":"100Gi"}}}}'

Now it’s time to load the data by

tar xzvf test_db-1.0.7.tar.gz
cd test_db/
mysql --comments -h 127.0.0.1 -P 14000 -u root < employees.sql

Validate if the data is loaded correctly

mysql --comments -h 127.0.0.1 -P 14000 -u root employees < test_employees_md5.sql

In order to manage some components directly, install tiup and a few CLIs

curl --proto '=https' --tlsv1.2 -sSf https://tiup-mirrors.pingcap.com/install.sh | sh
tiup ctl:v6.5.0 pd

Control PD metadata

Following this document,

kubectl port-forward -n tidb-cluster svc/basic-pd 2379:2379 &>/tmp/portforward-pd.log &

Following commands are one of pd-ctl, listed in this document, like tiup ctl:v6.5.0 pd -u http://127.0.0.1:2379 cluster.

Manage a cluster

• Confirm the cluster: cluster
• Confirm the configuration: config show all

Manage PD members

• Confirm the PD members: member
• Confirm the leader: member leader show

Identify and analyze slow queries

See this document

• Run queries on information_schema.SLOW LOG
• Run the query ADMIN SHOW SLOW recent 10

• 2024 2
• 2023 20
• 2022 11
• 2021 12
• 2020 21
• 2019 5