PostgresDB PaaS for Azure

Setup Postgres Service on Azure Portal

An Azure Database for PostgreSQL server is created with a configured set of compute and storage resources. The server is created within an Azure resource group.

Prerequisites

The following prerequisites are necessary for setting up Postgres Database on Azure.

• A Microsoft account is required to work with Azure Cloud Platform. Sign up for a Microsoft account, if you do not have one.
• Ensure that you have an Azure subscription with a billing account.

PostgreSQL Database

Azure Database for PostgreSQL is a managed service that you use to run, manage, and scale highly available PostgreSQL databases in the cloud. To set up Postgres Database, refer to the official Azure Documentation.

Setup Postgres

Step 1: To set up Postgres, navigate to Azure Console, search for Azure Database for PostgreSQL flexible servers, and click it.

Step 2: Click Create.

PostgreSQL Basics

This section provides basic information about your Postgres database

Step 1: Enter the required details for the following mandatory fields.

• Subscription
• Resource Group
• Server Name
• Region
• PostgreSQL Version
• Admin Username
• Password
• Confirm Password

(contd.)

(contd.)

Step 2: Click Next: Networking>.

Networking

Networking in the below mentioned scenario describes securing the database and accessing it privately without exposing it to the outside world.

Step 1: Fill the required details for setting up the networking.

• Connectivity Method
• Subscription
• Virtual Network
• Subnet
• Private DNS Zone

(contd.)

(contd.)

Step 2: Click Next: Security (preview).

Security Preview

The storage used for database and backup is encrypted by default with service managed keys. Data Encryption additionally encrypts your databases, backups, and logs at rest without any changes to your application.

Step 1: Keep the default settings and click Next: Tags >.

Tags

Tags are name/value pairs that enable you to categorize resources and view consolidated billing by applying the same tag to multiple resources and resource groups.

Step 1: Add the tags as per your organization standards.

Step 2: Click Next: Review + Create >.

Review + Create

Verify the details and click Review + Create to complete the configurations and deploy the PostgreSQL Database.

(contd.)

High Availability

High Availability makes sure that the database service is always available and connectivity remains intact even in a scenario where one instance of the database is down or unavailable.

Infoworks recommends enabling High Availability to avoid downtime in case of an outage on the Postgres Machines.

Enabling High Availability

Enabling High Availability for a New Instance

Step 1: To enable High Availability for a new instance of PostgreSQL, navigate to Azure Console, search for Azure Database for PostgreSQL flexible servers, and click it.

Step 2: Click Create.

Step 3: Under the Basics tab, click Enable High Availability checkbox.

Enabling High Availability for an Existing Instance

Step 1: To enable High Availability for an existing instance of PostgreSQL, open an existing Postgres instance from the Azure portal,

Step 2: Under the Settings section, click High Availability, and click it.

Step 3: Select Enable High Availability checkbox.

User Permissions

For Infoworks to use Postgres PaaS, considering the variables set during the installation, the given Postgres user (set in POSTGRESDB_USERNAME) needs to have the LOGIN Role. It must also have all privileges on a given database (set in INFOWORKS_POSTGRESDB_DATABASE_NAME). For more information, refer to Infoworks Installation on Azure Kubernetes Service (AKS).

To set the privileges:

xxxxxxxxxx

CREATE USER <POSTGRESDB_USERNAME>;

GRANT ALL PRIVILEGES ON DATABASE <INFOWORKS_POSTGRESDB_DATABASE_NAME> TO <POSTGRESDB_USERNAME>;

Verifying Connection from PostgresDB to Infoworks

Verify that applications can access the PostgreSQL DB from Infoworks AKS cluster.

Use the following procedure as a reference only.

Step 1: Log in to your Bastion host where Kubernetes cluster is accessible.

Step 2: Create a pod with PostgreSQL image in the respective namespace.

xxxxxxxxxx

kubectl run postgres --image postgres --restart=Never --namespace <namespace> sleep 3000

Step 3: Run the exec command to run bash command inside the pod.

xxxxxxxxxx

kubectl exec -it postgres --namespace <namespace> -- bash

Step 4: Get the connection string from PostgreSQL DB. For example, consider the below given command.

xxxxxxxxxx

root@postgres:/# psql --host <postgres-db-name>.postgres.database.azure.com --port 5432 --username <username> --dbname <database_name>

Step 5: Provide the password.

Password for user <username>:

psql (15.1 (Debian 15.1-1.pgdg110+1), server 13.7)

SSL connection (protocol: TLSv1.3, cipher: TLS_AES_256_GCM_SHA384, compression: off)

Type "help" for help.

​

postgres=>

The PostgreSQL DB and Infoworks are successfully connected. In case of any errors, refer to official Postgres Documentation.

Step 6: Once the connection between Postgres and Infoworks is successful. If you wish to delete Postgres pod, execute the following command.

xxxxxxxxxx

kubectl delete pods postgres --namespace <namespace>

PostgresDB PaaS Backup and Restore

PostgresDB Backup and Restore are necessary steps to support rollback, since the airflow engine specific to the Infoworks Version has been upgraded. Therefore, the PostgresDB data gets upgraded to support the newer Airflow version, leaving the data incompatible with the older versioned Airflow and by association Infoworks.

Backup

PostgresDB backup may either be saved as Azure supported scheduled backup or through the pg_dump binary.

First Method: Using Azure-provided Backup Options

Azure provides an automated, scheduled backup option with a defined retention period. For more information, refer to Backup and restore in Azure Database for PostgreSQL - Flexible Server.

However, this method has the following limitations:

Limitations

• This option takes a backup of the entire Server and not only of a specific Database inside the project.
• Currently, there is no on-demand backup available. Therefore any postgres data changes after a backup and before the upgrade will be lost.
• The retention period of these backups ranges from 7 to 35 days. Therefore any backup older that the user-defined retention period is lost.

Second Method: Using pg_dump commands for Backup

To avoid the limitations of the Cloud-based approach, Postgres data backup can be done through the following pod template, which uses the pg_dump command to backup Postgres data.

Limitation Postgres password has to be provided either as an environment variable mounted to the ephemeral pod, or to the backup script.

Step 1: It is assumed that the environment variable IW_HOME is set to /opt/infoworks, which is the path where your current 5.4.0 charts are stored. In case it is not set, execute the following command:

xxxxxxxxxx

export IW_HOME=/opt/infoworks

Step 2: Export the following Environment Variables.

xxxxxxxxxx

export IW_NAMESPACE="<your-infoworks-kubernetes-namespace>"

export POSTGRESDB_HOSTNAME="<postgres-host>"

export POSTGRESDB_USERNAME="<postgres-user>"

export POSTGRES_PLAINTEXT_PASSWORD="<postgres-plaintext-password>"

export INFOWORKS_POSTGRESDB_DATABASE_NAME="<postgres-database>"

For more details, refer to the Infoworks Installation on Azure Kubernetes Service (AKS) section.

NOTE The password provided as POSTGRES_PLAINTEXT_PASSWORD is plain text, and not encrypted.

NOTE All of the above values can be found in $IW_HOME/iw-k8s-installer/configure.sh file.

Step 3: Create a directory to store the database backup dump using the below command

xxxxxxxxxx

mkdir -p ${IW_HOME}/postgres_backup

cd ${IW_HOME}

Step 4: Clear any existing pods with the same name.

xxxxxxxxxx

kubectl -n ${IW_NAMESPACE} delete po postgres-backup

Step 5: Run the commands to create the required pod on the same namespace as the Kubernetes installation.

xxxxxxxxxx

kubectl apply -f - <<EOF

apiVersion: v1

kind: Pod

metadata:

  name: postgres-backup

  namespace: ${IW_NAMESPACE}

spec:

  containers:

  - name: postgres-backup

    env:

    - name: IW_NAMESPACE

      value: "${IW_NAMESPACE}"

    - name: POSTGRESDB_HOSTNAME

      value: "${POSTGRESDB_HOSTNAME}"

    - name: POSTGRESDB_USERNAME

      value: "${POSTGRESDB_USERNAME}"

    - name: POSTGRES_PLAINTEXT_PASSWORD

      value: "${POSTGRES_PLAINTEXT_PASSWORD}"

    - name: INFOWORKS_POSTGRESDB_DATABASE_NAME

      value: "${INFOWORKS_POSTGRESDB_DATABASE_NAME}"

    image: us-central1-docker.pkg.dev/iw-gcp-eng-dev/iwx-devops/infoworks-postgres-utils:psql-15

    imagePullPolicy: Always

    command: [ "/bin/bash", "-c", "--" ]

    args: [ "/opt/infoworks/scripts/postgres_backup_and_restore.sh -d" ]

  restartPolicy: Never

EOF

Step 6: Once the postgres-backup is up, run the following command to get logs

xxxxxxxxxx

kubectl -n $IW_NAMESPACE logs postgres-backup -f

After some time, you will receive the following message on your command prompt.

xxxxxxxxxx

Taking Dump of Postgres

Taking Postgres Dump - Enter the Parameters of the source PostgresDB Server

PG Data Dumped to postgres_backup/pg_dump_data.sql

Run kubectl -n aks-upgrade-540 cp postgres-backup:postgres_backup/pg_dump_data.sql postgres_backup/pg_dump_data.sql to copy to your local

Step 7: Copy the generated artifacts to your machine.

xxxxxxxxxx

date_time=$(date "+%Y_%m_%d_%H_%M_%S")

​

kubectl -n ${IW_NAMESPACE} cp postgres-backup:postgres_backup/pg_dump_data.sql 

postgres_backup/pg_dump_data.sql

​

cp postgres_backup/pg_dump_data.sql postgres_backup/pg_dump_data_${date_time}.sql

This should create the following files in the directory postgres _backup, $(date -I) being the current timestamp. To view the list of directories, execute the following command:

xxxxxxxxxx

ls postgres_backup

xxxxxxxxxx

pg_dump_data.sql pg_dump_data-${date_time}.sql

Restore

PostgresDB Restore is a necessary step to support rollback since the airflow engine specific to the Infoworks version has been upgraded. Therefore, the PostgresDB data gets upgraded to support the newer Airflow version, leaving the data incompatible with the older versioned Airflow and by association Infoworks.

PostgresDB backup may either be restored through the cloud-supported scheduled backup or manually restoring the data, as picked by the user during Backup.

First Method: Using Azure-provided Backup Options

Azure provides an automated, scheduled backup option with a defined retention period. For more information, refer to Backup and restore in Azure Database for PostgreSQL - Flexible Server.

Limitations

• This option takes a backup of the entire project and not only of a specific Database inside the project.
• Currently, there is no on-demand backup available. Therefore any postgres data changes after a backup and before the upgrade will be lost.
• The retention period of these backups ranges from 7 to 35 days. Therefore any backup older than the user-defined retention period is lost.

Second Method: Using PSQL commands for Restore

To avoid the limitations of the Cloud-based approach, postgres data restore can be done through the following pod template, which uses the PSQL command to restore Postgres data.

Limitation Postgres Password has to be provided either as an environment variable mounted to the ephemeral pod or to the backup script.

Step 1: It is assumed that the environment variable IW_HOME is set to /opt/infoworks, which is the path where your current 5.4.0 charts are stored. In case it is not set, execute the following command

xxxxxxxxxx

export IW_HOME=/opt/infoworks

Step 2: Export the following environment variables.

xxxxxxxxxx

export IW_NAMESPACE="<your-infoworks-kubernetes-namespace>"

export POSTGRESDB_HOSTNAME="<postgres-host>"

export POSTGRESDB_USERNAME="<postgres-user>"

export POSTGRES_PLAINTEXT_PASSWORD="<postgres-plaintext-password>"

export INFOWORKS_POSTGRESDB_DATABASE_NAME="<postgres-database>"

For more details regarding the above mentioned commands, refer to Infoworks Installation on Azure Kubernetes Service (AKS).

NOTE The password provided as POSTGRES_PLAINTEXT_PASSWORD is plain text and not encrypted.

NOTE All of the above values can be found in $IW_HOME/iw-k8s-installer/configure.sh file.

It is recommended to take a backup of the current PostgresDB, by exporting the following variable. This requires the given user to have permission to create a new database. The pod will fail in case the backup fails, therefore do not export it to skip the backup.

xxxxxxxxxx

export INFOWORKS_POSTGRESDB_BACKUP_DATABASE_NAME="<postgres-backup-database>"

Step 3: Switch to the directory where PostgresDB backups were stored.

NOTE Assuming the backups were stored in ${IW_HOME}/postgres_backup as per the Backup section.

xxxxxxxxxx

cd ${IW_HOME}

Step 4: Clear any existing pods with the same name.

xxxxxxxxxx

kubectl -n ${IW_NAMESPACE} delete po postgres-backup

Step 5: Run the commands to create the required pod on the same namespace as the Kubernetes installation.

xxxxxxxxxx

kubectl apply -f - <<EOF

apiVersion: v1

kind: Pod

metadata:

  name: postgres-backup

  namespace: ${IW_NAMESPACE}

spec:

  containers:

  - name: postgres-backup

    env:

    - name: IW_NAMESPACE

      value: "${IW_NAMESPACE}"

    - name: POSTGRESDB_HOSTNAME

      value: "${POSTGRESDB_HOSTNAME}"

    - name: POSTGRESDB_USERNAME

      value: "${POSTGRESDB_USERNAME}"

    - name: POSTGRES_PLAINTEXT_PASSWORD

      value: "${POSTGRES_PLAINTEXT_PASSWORD}"

    - name: INFOWORKS_POSTGRESDB_DATABASE_NAME

      value: "${INFOWORKS_POSTGRESDB_DATABASE_NAME}"

    - name: INFOWORKS_POSTGRESDB_BACKUP_DATABASE_NAME

      value: "${INFOWORKS_POSTGRESDB_BACKUP_DATABASE_NAME}"

    image: us-central1-docker.pkg.dev/iw-gcp-eng-dev/iwx-devops/infoworks-postgres-utils:psql-15

    imagePullPolicy: Always

    command: [ "/bin/bash", "-c", "--" ]

    args: [ "/opt/infoworks/scripts/postgres_backup_and_restore.sh -r" ]

  restartPolicy: Never

EOF

Step 6: Once the pod is running, run the following command to copy the backup to the pod.

xxxxxxxxxx

kubectl -n ${IW_NAMESPACE} cp ${IW_HOME}/postgres_backup/pg_dump_data.sql postgres-backup:postgres_backup/pg_dump_data.sql

After copying the above file, the restore script will automatically restore from the backup files supplied.