Manage virtual machine snapshots

This tutorial details how to create, revert and manually delete a virtual machine snapshot. Before proceeding, a brief talk about virtual machine snapshots and side effects on running instances is described below.

Snapshots

A virtual machine snapshot is not a copy so it cannot be treated as a direct backup. A snapshot file is simply a log of changes to the original virtual disk. The virtual machine is running on the most current snapshot, not the original vmdk disk files. To reiterate: snapshots are not copies of the VM’s original vmdk disk files. Taking a snapshot does not create a complete copy of the original vmdk disk file, rather it only copies the delta disks.

Side effects

Decreased performance if there are too many delta files in a chain (caused by having too many snapshots).
Decreased performance if delta files become too large.
Cannot increase Virtual Disk size while snapshots are active.
Delta files can grow to the same size as the original base disk file: a virtual machine’s provisioned storage size can grow to ( its original size ) multiplied by ( the number of snapshots ).

Lifecycle

The following diagram describes the ITS Private Cloud virtual machine snapshot lifecycle:

When a virtual machine snapshot request is submitted, the lifecycle starts. By default, a virtual machine snapshot in the ITS Private Cloud has a start and end date based on the timestamp and lifetime. For example, if the timestamp is set to 2017-03-15 21:00 and the lifetime to 24 hours, the snapshot will be created at 2017-03-15 21:00 and its deletion scheduled to timestamp + lifetime = 2017-03-16 21:00.

However, you still can rollback or delete the snapshot if required within the time frame and even extend the duration of the snapshot for a maximum of 3 times. For example, if by any reason you need to extend the snapshot for 72 hours, the resulting delete date would be old_delete + extension = 2017-03-16 21:00 + 72 = 2017-03-19 21:00.

With the VSS CLI you can run the vss-cli compute vm set <uuid-or-name> snapshot to either create, delete, or revert to a given snapshot:

vss-cli compute vm set <uuid-or-name> snapshot --help

Usage: vss-cli compute vm set snapshot [OPTIONS] COMMAND [ARGS]...

Manage virtual machine snapshots. Create, delete and revert virtual
machine snapshot on a given date and time.

Options:
--help  Show this message and exit.

Commands:
mk  Create snapshot
re  Revert snapshot
rm  Remove snapshot

Create

In order to create a virtual machine snapshot run the vss-cli compute vm set <name-or-vm-id> snapshot followed by the mk command, as well as the --description/-d with a brief description of what the snapshot will be used for, --timestamp/-t with the date and time the snapshot would need to run and --lifetime/-l with the number of hours for the snapshot to live.

Usage: vss-cli compute vm set snapshot mk [OPTIONS]

  Create virtual machine snapshot:

  vss-cli compute vm set <name-or-vm_id> snapshot mk --description 'Short
  description'

  Note: if -t/--timestamp not specified, the snapshot request timestamp is
  current time.

Options:
  -d, --description TEXT          A brief description of the snapshot.
                                  [required]

  -t, --timestamp [%Y-%m-%d %H:%M]
                                  Timestamp to create the snapshot from.
                                  [default: 2020-04-27 16:51]

  -l, --lifetime INTEGER RANGE    Number of hours the snapshot will live.
                                  [default: 24]

  -c, --consolidate / --no-consolidate
                                  Consolidate disks after snapshot deletion
                                  [default: True]

  --help                          Show this message and exit.

The following command submits a request to create a virtual machine snapshot starting from 2020-04-27 16:53 and valid until 2017-03-15 22:30 (timestamp + lifetime).

vss-cli --wait  compute vm set vm-123 snapshot mk \
--description 'Before doing a difficult upgrade' --timestamp '2020-04-27 16:53'
--consolidate

As a result, the command will return a snapshot request ID, as well as a confirmation email.

Check the state of the request made by running vss-cli request snapshot ls -s created_on desc -c 1 or vss-cli request snapshot get <id>.

Revert

Reverting to a given snapshot is executed right away. In order to revert to snapshot, obtain the snapshot ID by running vss-cli compute vm get <name-or-vm-id> snapshot, resulting in a similar output as shown below:

In this case, the snapshot ID is 4 and can be verified by running vss-cli compute vm get <name-or-vm-id> snapshot <id>.

Run vss-cli compute vm set <name-or-vm-id> snapshot re <snapshot-id> to submit a snapshot request to revert to a particular snapshot id:

Check the state of the request made by running vss-cli request snapshot ls -s created_on desc -c 1 or vss-cli request snapshot get <id>.

Delete

Virtual machine snapshots can also be deleted manually instead of waiting for the task to run. Run vss-cli compute vm set <name-or-vm-id> snapshot rm <snapshot-id> to submit a snapshot request to delete a given snapshot:

Check the state of the request made by running vss-cli request snapshot ls -s created_on desc -c 1 or vss-cli request snapshot get <id>.

Extend

Extend the snapshot lifetime for a maximum of 72 hours for only 3 occasions. Run vss-cli request snapshot set <request_id> duration --lifetime <hours> to extend a snapshot request:

A summary of the extended request will be returned.

Disk Consolidation

When you initiate a Delete or DeleteAll operation on snapshots, the snapshot is immediately deleted, then the backing virtual machine disk .vmdk files are consolidated on-disk. If the consolidation fails [during snapshot deletion], some Virtual Disk files may remain on disk and/or be actively used on the Datastore, consuming storage capacity. (Source: VMware).

Virtual machine disk consolidation is based on finding hierarchies of redo logs that can be combined without violating data dependency. The redundant redo logs after merging are then deleted. Consolidation improves I/O performance since less number of virtual disk files need to be traversed; it also reduces the storage usage.

Status

To validate whether a virtual machine requires disk consolidation, run vss-cli compute vm get <name-or-vm-id> consolidate as shown below:

Consolidate

Additional space is temporarily required to perform the operation. Consolidation can be I/O intensive, it is advisable to invoke this operation when guest is not under heavy I/O usage.

Disk consolidation is treated as a change request and can be scheduled with the --schedule/-s flag. If disk consolidation is required, run vss-cli compute vm set <name-or-vm-id> consolidate as shown below:

You can check the state of the request made by running vss-cli request change ls -s created_on desc -c 1 or vss-cli request change get <id>.