Skip to content

Latest commit

 

History

History
132 lines (93 loc) · 3.85 KB

README.md

File metadata and controls

132 lines (93 loc) · 3.85 KB

Persistent Storage

Expected outcome

In this lab you will use Kubernetes Persistent Storage constructs to mount a persistant volume store for a container. This exercise will use the Dynamic method instead of a static Persistent Volume

How to

  1. Create a Storage Class named azuredsc

    Create a storage class. Please use storage.k8s.io/v1beta1. You can see the spec in here

    If you created the cluster with managed disks you must use managed as the StorageClass kind. If your Agents are using Storage Accounts for DataDisks then you must use shared as the StorageClass kind.

    Example

    Deploy:

    kubectl create -f config\storage-class.yaml

  2. Create the Persistent Volume Claim named mssql-pv-claim

    Create a Persistent Volume Claim definition. You need to specify storageClassName on specification.

    Example

    Deploy:

    kubectl create -f config\pv-claim.yaml

  3. Deploy the microsoft/mssql-server-linux container.

    • Run with the volume /var/opt/mssql mounted from the PersistentVolumeClaim you created.
    • The mssql container requires 3 environment variables;
      • ACCEPT_EULA with a value of Y
      • MSSQL_SA_PASSWORD with a value of a complex password of at least 8 characters including uppercase, lowercase letters, base-10 digits and/or non-alphanumeric symbols
      • MSSQL_PID with a value of Developer

    Example

    Deploy:

    kubectl create -f config\mssql-store.yaml

  4. Use sqlcmd to create a new database and a new table and insert some data into the table.

    exec into the container by locating the pod name that mssql-deploy created

    $ kubectl exec mssql-store-***-*** -it bash
root@mssql-store-***-***:/#

    Connect to your service instance on port 1433

    /opt/mssql-tools/bin/sqlcmd -S <INTERNAL_IP_OF_SERVICE>,1433 -U SA -P '<YourPassword>'

    At the sqlcmd prompt create a new database

    CREATE DATABASE TestDB
GO

    Verify the table was created

    SELECT Name from sys.Databases
GO

    Set the DB context and create a new table called Inventory

    USE TestDB
CREATE TABLE Inventory (id INT, name NVARCHAR(50), quantity INT)
GO

    Insert data into the new table

    INSERT INTO Inventory VALUES (1, 'penguins', 150); INSERT INTO Inventory VALUES (2, 'whales', 154);
GO

    Query the table to verify the data was written

    SELECT * FROM Inventory WHERE quantity > 152;
GO

    Exit the sqlcmd tool

    QUIT

    Exit your container session

    exit

  5. Determine the node the msqql pod is running on and drain the node.

    You may have to use --ignore-daemonsets parameter to properly drain the node. You may have to use --delete-local-data if jenkins is still running from the previous step.

    kubectl get node

kubectl drain aks-nodepool1-***-* --ignore-daemonsets --delete-local-data

    Once the node is showing no errors

    kubectl uncordon aks-nodepool1-***-*

  6. Verify that the new pod can access the data written by the old pod.

    Use kubcetl get all to determine when the pod is finished recreating.

    Once the new pod is in a running state on a new node, connect to the database and query the Inventory table again to verify the data persisted.

Advanced areas to explore

  1. Try make a similar claim using AzureFiles.
  2. Use secrets to store the mssql credentials

Notes

  • In the drain pod step, if jenkins is still running from the previous challenge, it's local data will be lost.