Restoring a volume from a snapshot

Let's say I have an AMI with an attached EBS Volume.

I also have a snapshot.

I want to "restore" the EBS Volume to the snapshot.

What's th开发者_JS百科e best process to do this?

I don't know of a way that you can 'restore' an attached volume, but the way i would do it is to create a volume from the snapshot, then detach the original and attach the new one.

If you have a running EC2 instance, and you want to restore it to the state captured in an earlier snapshot, then you need to stop the instance, detach its current volume, create a new volume from the snapshot, attach the new volume to your instance, and restart your instance. Furthermore, there are a couple subtleties around specifying the availability zone of the new volume, and the device name when detaching/re-attaching the volume.

The logic might easier to see if you do it from the command line, instead of from the AWS web UI.

The following bash script is not fit for production use, since it lacks any error-checking and it just uses sleep instead of polling to ensure AWS commands have completed. But it does perform all these steps successfully:

#!/bin/bash

set -e

# IN PARAMS
INSTANCE_ID=<YOUR_INSTANCE_ID_HERE>
SNAPSHOT_ID=<YOUR_SNAPSHOT_ID_HERE>
# OUT PARAMS
VOLUME_ID=

# begin execution

echo "Gathering information about the instance"
DEVICE_NAME=`ec2-describe-instance-attribute ${INSTANCE_ID} --block-device-mapping | awk '{print $2}'`
OLD_VOLUME_ID=`ec2-describe-instance-attribute ${INSTANCE_ID} --block-device-mapping | awk '{print $3}'`
echo "Found instance ${INSTANCE_ID} has volume ${OLD_VOLUME_ID} on device ${DEVICE_NAME}"

echo "Creating new volume from snapshot"
AVAILABILITY_ZONE=`ec2-describe-availability-zones --filter state=available | head -n 1 | awk '{print $2}'`
VOLUME_ID=`ec2-create-volume --availability-zone ${AVAILABILITY_ZONE} --snapshot ${SNAPSHOT_ID} | awk '{print $2}'`
echo "Created new volume: ${VOLUME_ID}"

sleep 20

echo "Stopping the instance"
ec2-stop-instances $INSTANCE_ID

sleep 20

echo "Detaching current volume"
ec2-detach-volume $OLD_VOLUME_ID --instance $INSTANCE_ID --device $DEVICE_NAME

sleep 20

echo "Attaching new volume"
ec2-attach-volume $VOLUME_ID  --instance $INSTANCE_ID --device $DEVICE_NAME

sleep 20

echo "Starting the instance"
ec2-start-instances $INSTANCE_ID

I have touched up the script provided by @algal to use the aws cli and polling instead of sleep. It will also look for the latest snapshot of the given volume.

#!/bin/bash

set -e

# IN PARAMS
RECOVERY_INSTANCE_ID=
SNAPSHOT_VOLUME_ID=

echo "Gathering information about the instance"
BLOCK_DEVICE_MAPPING=`aws ec2 describe-instance-attribute --instance-id ${RECOVERY_INSTANCE_ID} --attribute blockDeviceMapping`
DEVICE_NAME=`echo ${BLOCK_DEVICE_MAPPING} | jq '.BlockDeviceMappings[0].DeviceName' | tr -d '"'`
OLD_VOLUME_ID=`echo ${BLOCK_DEVICE_MAPPING} | jq '.BlockDeviceMappings[0].Ebs.VolumeId' | tr -d '"'`
AVAILABILITY_ZONE=`aws ec2 describe-instances --filters "Name=instance-id,Values='${RECOVERY_INSTANCE_ID}'" | jq '.Reservations[0].Instances[0].Placement.AvailabilityZone' | tr -d '"'`
LATEST_SNAPSHOT_ID=`aws ec2 describe-snapshots --filter "Name=volume-id,Values='${SNAPSHOT_VOLUME_ID}'" | jq '.[]|max_by(.StartTime)|.SnapshotId' | tr -d '"'`
echo "Found instance ${RECOVERY_INSTANCE_ID} in ${AVAILABILITY_ZONE} has volume ${OLD_VOLUME_ID} on device ${DEVICE_NAME}"

echo "Creating new volume from snapshot ${LATEST_SNAPSHOT_ID}"
NEW_VOLUME_ID=`aws ec2 create-volume --region eu-west-1 --availability-zone ${AVAILABILITY_ZONE} --snapshot-id ${LATEST_SNAPSHOT_ID} | jq '.VolumeId' | tr -d '"'`
echo "Created new volume ${NEW_VOLUME_ID}"

aws ec2 wait volume-available --volume-ids $NEW_VOLUME_ID

echo "Stopping the instance"
aws ec2 stop-instances --instance-ids $RECOVERY_INSTANCE_ID

aws ec2 wait instance-stopped --instance-ids $RECOVERY_INSTANCE_ID

echo "Detaching current volume"
aws ec2 detach-volume --volume-id $OLD_VOLUME_ID --instance-id $RECOVERY_INSTANCE_ID

aws ec2 wait volume-available --volume-ids $OLD_VOLUME_ID

echo "Attaching new volume"
aws ec2 attach-volume --volume-id $NEW_VOLUME_ID --instance-id $RECOVERY_INSTANCE_ID --device $DEVICE_NAME

aws ec2 wait volume-in-use --volume-ids $NEW_VOLUME_ID

echo "Starting the instance"
aws ec2 start-instances --instance-ids $RECOVERY_INSTANCE_ID

If you'd like to stay up to date with this script or contribute:

https://github.com/karimtabet/ebs_snapshot_recovery

To replace a volume attached to an instance with a new volume created from a snapshot:

1. Create a volume from the snapshot in the same availability zone the instance is in (right click on snapshot and click "create volume from snapshot")
2. Best to stop the instance to avoid any application from crashing. Wait until instance is stopped.
3. Write down the exact device name of the original volume (it is written in the AWS console under instances view or volumes view)
4. Detach the old volume, delete it afterwards if you don't need it.
5. Attach the newly created volume (from the snapshot) to the instance with the same device name.
6. Start the instance again

Make a volume from the snapshot to mount the volume on an existing EC2 machine and copy files from it.

Check the EC2 machine.

1. Pick an instance. EC2 tab | INSTANCES | Instances.
2. Make a note of the EC2 machine’s availability zone.

Create a volume.

1. Find the snapshot you want to copy files from and tick the box. ELASTIC BLOCK STORE | Snapshots
2. Click the Create Volume button and fill in the fields. o The Size must be bigger than the snapshot size (free micro-instances get an 8GB volume). o The Availability Zone must be the same as the EC2 machine’s. o The Snapshot is already selected, more or less like snap12345678 - my description.
3. Click the Yes, Create button. A new line appears in the Volumes table. ELASTIC BLOCK STORE | Volumes

Attach the volume.

1. Click the Attach Volume button and fill in the fields.
2. The Volume value is already there.
3. Pick your machine name i-12345678 (running) from the drop-down list of Instances.
4. The Devices field shows the first available device name, like /dev/sdf. Does anyone bother changing this value?
5. Click the Yes, Create button. A new device magically appears on the EC2 machine.
6. Close the AWS console.