script to automatically snapshot local ebs volumes. Keep a pre-define number of snapshot and update them.

aws_ebs_snapshoting.py

import os
import sys
import time
import boto3
from botocore.exceptions import ClientError


def aws_ec2_session():
    try:
        session = boto3.Session(profile_name=none)
        return session.client('ec2')
    except ClientError as e:
        print e
        sys.exit()

def find_ec2_nodes(tag):
    filter = [{'Name':'tag:Name', 'Values':tag}]
    try:
        instances =  ec2.describe_instances(Filters=filter)
        return instances
    except ClientError as e:
        print e
        sys.exit()

def find_volumeids(nodes):
    volumeids = []
    if nodes:
        for instance in nodes["Reservations"]:
            for block_device in instance["Instances"]:
                for ebs in block_device["BlockDeviceMappings"]:
                    volumeids.append(ebs["Ebs"]["VolumeId"])
        return volumeids
    else:
        return 'no volume found for {}'.format(nodes)

def find_snapshots(tags):
    try:
        findsnapshots = ec2.describe_snapshots(Filters=[{'Name': 'tag:Name',
                                                         'Values': tags}])
        return findsnapshots['Snapshots']
    except ClientError as e:
        print e
        sys.exit()

def create_snapshot(volumesids, description, tags):
    try:
        snapshot = ec2.create_snapshot(VolumeId=volumeids[0], Description=description,
         TagSpecifications=[{'ResourceType':'snapshot',
                             'Tags': [{'Key': 'Name', 'Value': tags}]}])
        return snapshot['SnapshotId']
    except ClientError as e:
        print e
        sys.exit()

def delete_snapshot(snapshotid):
    try:
        ec2.delete_snapshot(SnapshotId=snapshotid)
        return 'deleted snapshot {}'.format(snapshotid)
    except ClientError as e:
        print e
        sys.exit()

def find_oldest_snapshot(snapshots):
    old = {}
    for snapshot in snapshots:
        if not old:
            old['starttime'] = snapshot['StartTime']
            old['snapid'] = snapshot['SnapshotId']
        elif old['starttime'] > snapshot['StartTime']:
            old['starttime'] = snapshot['StartTime']
            old['snapid'] = snapshot['SnapshotId']
    return old['snapid']


if __name__ == '__main__':

    ec2 = aws_ec2_session()
    airflow_nodes = find_ec2_nodes(['Data:Airflow-Scheduler'])
    volumeids = find_volumeids(airflow_nodes)
    found_snapshot = find_snapshots(['airflow'])
    num_found_snapshot = len(found_snapshot)
    max_snapshot = 3

    if num_found_snapshot == max_snapshot:
        print '{} snapshots: {}'.format(num_found_snapshot,
                                        [snapshot['SnapshotId'] for snapshot in found_snapshot])
        oldest = find_oldest_snapshot(found_snapshot)
        print 'deleting oldest snapshot: {}'.format(oldest)
        delete_snapshot(oldest)
        new_snap = create_snapshot(volumeids, 'Data:Airflow-Scheduler', 'airflow')
        print 'created a new snapshot: {}'.format(new_snap)
    elif num_found_snapshot < max_snapshot:
        new_snap = create_snapshot(volumeids, 'Data:Airflow-Scheduler', 'airflow')
        print '{} snapshots. Creating a new snapshot: {}'.format(num_found_snapshot, new_snap)
    elif num_found_snapshot > num_found_snapshot:
        oldest = find_oldest_snapshot(found_snapshot)
        delete_snapshot(oldest)
        print 'Too many snapshots({}).deleting oldest snapshot: {}'.format(num_found_snapshot, oldest)