DataPains Trino Kubernetes

Medium Article For Reference

YouTube Guide Video

NOTE

• Old lakehouse catalog is now delta
• And iceberg also exists with this deployment.

Pre Requisites

• Docker Desktop (with k8s enabled)
• make

Setup

MinIO With Metastore

In this workshop I provide to you a setup for a MinIO with a Metastore (utilising postgres internally) to successfully query delta tables via trino.

Docker

If you are using M1 chip:

make build-image COMPONENT=metastore DOCKER_BUILD="buildx build --platform linux/amd64" CONTEXT="tools/docker"

Otherwise

make build-image COMPONENT=metastore CONTEXT="tools/docker"

Metastore - Vanilla - Local

make deploy-local-metastore ENV=local

To delete

make delete-local-metastore ENV=local

Statefulset

For Hive, Postgres, MinIO we choose to use StatefulSet as the deployment type over Deployment.

A StatefulSet should be chosen for stateful components that require predictable Pod identities, ordered rollouts, and stable storage access.

Deployment should be chosen instead when your service is stateless, doesn't require persistent storage, and won't be affected by changes to Pod identities.

Trino

Helm

First add the trino helm repo!

helm repo add trino https://trinodb.github.io/charts

Make sure you use version >= 0.21.0

helm search repo trino

Should give:

NAME       	CHART VERSION	APP VERSION	DESCRIPTION
trino/trino	0.21.0       	448        	Fast distributed SQL query engine for big data ...

values.yaml

We use default values for the deployment, but please note section:

additionalCatalogs:
  iceberg: |-
    connector.name=iceberg
    hive.metastore.uri=thrift://hive-service.metastore:9083
    hive.s3.endpoint=http://minio-service.metastore:9000
    hive.s3.path-style-access=true
    hive.s3.aws-access-key=minio
    hive.s3.aws-secret-key=minio123
    iceberg.file-format=parquet

  delta: |-
    connector.name=delta_lake
    hive.metastore.uri=thrift://hive-service.metastore:9083
    hive.s3.endpoint=http://minio-service.metastore:9000
    hive.s3.path-style-access=true
    hive.s3.aws-access-key=minio
    hive.s3.aws-secret-key=minio123
    delta.enable-non-concurrent-writes=true

  rdbms: |-
    connector.name=postgresql
    connection-url=jdbc:postgresql://postgres-service.metastore:5432/hivemetastore?allowPublicKeyRetrieval=true&useSSL=false&serverTimezone=UTC
    connection-user=metastore
    connection-password=password

This will ensure we can query our tables from our metastore.

make deploy-local-trino NAMESPACE=trino

And to delete

make delete-local-trino NAMESPACE=trino

Setup Delta Table And Query!

The trino cluster is now up and running!

We will have to execute ourselves into the coordinator and create a table.

kubectl get pods -n trino

kubectl exec -it datapains-trino-cluster-coordinator-747bc6ccff-b6tx2 -n trino -- /bin/bash

[trino@datapains-trino-cluster-coordinator-747bc6ccff-b6tx2 /]$ trino
trino> SHOW CATALOGS;
  Catalog
-----------
 delta
 iceberg
 system
 tpcds
 tpch
(4 rows)

Query 20240526_120408_00000_kyyrh, FINISHED, 2 nodes
Splits: 20 total, 20 done (100.00%)
2.57 [0 rows, 0B] [0 rows/s, 0B/s]

Delta

trino> CREATE SCHEMA delta.bronze;

trino> CREATE TABLE delta.bronze.products ( name VARCHAR, price DOUBLE, product_no BIGINT, ingest_date DATE, created TIMESTAMP WITH TIME ZONE);

trino> SELECT * FROM delta.bronze.products;
trino> INSERT INTO delta.bronze.products VALUES (CAST('Apple' AS VARCHAR), CAST(889.0 AS DOUBLE), CAST(1654523786273 AS BIGINT), CURRENT_DATE, CURRENT_TIMESTAMP);
trino> SELECT * FROM delta.bronze.products;

Iceberg

trino> CREATE SCHEMA iceberg.bronze;

trino> CREATE TABLE iceberg.bronze.products ( name VARCHAR, price DOUBLE, product_no BIGINT, ingest_date DATE, created TIMESTAMP WITH TIME ZONE);

trino> SELECT * FROM iceberg.bronze.products;
trino> INSERT INTO iceberg.bronze.products VALUES (CAST('Apple' AS VARCHAR), CAST(889.0 AS DOUBLE), CAST(1654523786273 AS BIGINT), CURRENT_DATE, CURRENT_TIMESTAMP);
trino> SELECT * FROM iceberg.bronze.products;