How to Create and use Snapshots in Kubernetes

Objective

Today I am going to show how to create snapshots in Kubernetes. This is really helpful in case you want to manage or mitigate the risk of losing your PODs.

From UI:

From the command line: 

  

Creating periodic snapshots 

When you create a StorageClass, you can specify a snapshot schedule on the volume as specified below. This allows snapshotting the persistent data of the running pod using the volume. 

Above spec will take snapshots of the portworx-repl-1-snap-internal PVC every 190 minutes.

Creating a snapshot on demand 

You can also trigger a new snapshot on a running POD by creating a PersistentVolumeClaim as specified in the following spec:

Note the format of the “name” field. The format is name.<new_snap_name>-source.<old_volume_name>. Above example references the parent (source) persistent volume claim pvc013 and creates a snapshot by the name snap013. 

Listing snapshots

To list snapshots taken by Portworx, use the /opt/pwx/bin/pxctl volume snapshot list command.

 For example: 

# /opt/pwx/bin/pxctl volume snapshot list 

ID                                   NAME   SIZE  HA   SHARED   IO_PRIORITY SCALE    STATUS 

1056733319296008813 snap013 1 GiB  2      no               LOW                 1               up - detached  

You can use the ID or NAME of the snapshots when using them to restore a volume. 

Restoring a pod from a snapshot 

To restore a pod to use the created snapshot, use the pvc name.snap013-source.pvc013 in the pod spec. 

Managing snapshots through pxctl 

You can provide the SAN capabilities, using portworx and then create a snapshot of a volume, let's say using an MYSQL DB.

First, create a database and a demo table in your MYSQL container. 

# mysql --user=tleoncio --password=password
MySQL [(none)]> create database pxThiagoLeoncio;
Query OK, 1 row affected (0.00 sec)
MySQL [(none)]> use pxThiagoLeoncio;
Database changed
MySQL [pxdemo]> create table  K8SThiagoLeoncioSample (counter int unsigned);
Query OK, 0 rows affected (0.04 sec)
MySQL [pxdemo]> quit;
Bye

Now create a snapshot of this database using pxctl. 

First, use pxctl volume list to see what volume you want to snapshot 

# /opt/pwx/bin/pxctl v l
ID     NAME          SIZE HA SHARED ENCRYPTED IO_PRIORITY SCALE STATUS
381983511213673988 pvc-e7e66f98-0915-11e7-94ca-7cd30ac1a138 20 GiB 2 no  no   LOW   0  up - attached on 147.75.105.196

Then use pxctl to snapshot your volume 

opt/pwx/bin/pxctl snap create 381983511213673988 --name snap-013
Volume successfully snapped: 835956864616765999

Also, you can use pxctl to see your snapshot created

# /opt/pwx/bin/pxctl snap list
ID     NAME SIZE HA SHARED ENCRYPTED IO_PRIORITY SCALE STATUS
835956864616765999 snap-01 20 GiB 2 no  no   LOW   0  up - detached

Then we create a mysql Pod to mount the snapshot discussed in this article:

kubectl create -f portworx-mysql-snap-pod.yaml




apiVersion: v1
kind: Pod
metadata:
  name: test-portworx-snapped-volume-pod
spec:
  containers:
  - image: mysql:5.6
    name: mysql-snap13
    env:
      # Use secret in real usage
    - name: MYSQL_ROOT_PASSWORD
      value: password
    ports:
    - containerPort: 3306
      name: mysql
    volumeMounts:
    - name: snap-013
      mountPath: /var/lib/mysql
  volumes:
  - name: snap-013
    # This Portworx volume must already exist.
    portworxVolume:
      volumeID: "vol1"

Inspect that the database shows the cloned tables in the new MYSQL instance. 

# mysql --user=tleoncio --password=password
mysql> show databases;
+--------------------+
| Database           |
+--------------------+ 
| mysql              |
| performance_schema |
| pxThiagoLeoncio    |
+--------------------+
3 rows in set (0.00 sec)

mysql> use pxThiagoLeoncio;
Reading table information for completion of table and column names
You can turn off this feature to get a quicker startup with -A

Database changed
mysql> show tables;
+-----------------------+
| Tables_in_pxdemo      |
+-----------------------+
| K8SThiagoLeoncioSample|
+-----------------------+
1 row in set (0.00 sec)

I hope this helps and happy coding,

Thiago Leoncio.