Skip to main content

Overview

The Instances panel under Project → Compute → Instances is the primary interface for managing the full lifecycle of compute instances — from initial launch through day-to-day operations, scaling, and eventual termination.
Prerequisites
  • member role or higher in the active project
  • At least one image available in the project (see Images)
  • A network in the project to attach the instance to

Instance List View

The instance table displays all instances in the active project:
ColumnDescription
NameInstance display name (click to open detail view)
Image NameSource image or snapshot used at launch
IP AddressPrivate IP(s) — hover to see all addresses
SizeFlavor name (vCPU/RAM/disk)
Key PairSSH key pair name injected at launch
StatusActive, Shutoff, Error, Paused, Suspended, Resize, etc.
Availability ZoneAZ the instance was scheduled into
TaskIn-progress operation (e.g., Rebooting, Migrating)
AgeTime since instance was created
ActionsPer-row action dropdown

Launch an Instance

Open the launch dialog

Navigate to Project → Compute → Instances and click Launch Instance in the top-right corner.

Details tab

FieldDescription
Instance NameDisplay name for the instance
DescriptionOptional description
Availability ZoneSelect or leave on nova (auto-schedule)
CountNumber of instances to launch (bulk launch)

Source tab

Select the boot source:
SourceDescription
ImageLaunch from a platform image
Instance SnapshotRestore from a previous snapshot
VolumeBoot from an existing block storage volume
Volume SnapshotBoot from a volume snapshot
Enable Create New Volume to automatically create a persistent root volume (recommended for production). Set volume size ≥ minimum disk requirement of the image.
Enable Delete Volume on Instance Delete only for ephemeral/test workloads. For production, keep the root volume after instance deletion to preserve data.

Flavor tab

Select the instance flavor that defines vCPU, RAM, and disk allocation. Flavors with a root disk smaller than the image’s minimum disk requirement are shown but unavailable.For workloads that need elastic scaling, choose a flavor with hot-add extra specs — see Live vCPU/RAM Scaling.

Networks tab

Allocate at least one network to the instance. Drag networks from Available to Allocated. The order determines the interface assignment (eth0, eth1, etc.).

Security Groups tab

Select which security groups to apply. The default security group allows all egress and permits traffic between instances in the same group. Add additional groups as needed (e.g., web-http, ssh-access).

Key Pair tab

Select an existing key pair or create a new one. The public key is injected into the instance via cloud-init — you’ll need the corresponding private key to SSH in.
If you launch without a key pair and the image doesn’t have a pre-configured password, you may lose SSH access. Use the Console to recover.

Launch

Review the configuration summary on the right panel and click Launch Instance. The instance enters BUILD status and transitions to ACTIVE when ready (typically 30–90 seconds).
The instance appears in the list with Active status. The assigned IP address is visible in the IP Address column.

Instance Actions

Each instance row has an Actions dropdown with the following operations:
Captures the current state of the instance root disk as a platform image. The instance is briefly quiesced during snapshot creation.Snapshots appear under Project → Compute → Images with type Snapshot. Use them as restore points or as a base for new instances.
Changes the flavor — adjusting vCPU, RAM, and/or disk. Requires a reboot and a confirmation step (accept or revert within 24 hours).See Resize an Instance for the full workflow.
Adjusts vCPU count and/or RAM on a running instance with no downtime. Available only on instances running a flavor with hot-add extra specs configured.See Live vCPU/RAM Scaling for prerequisites and usage.
Soft Reboot: Sends ACPI shutdown signal to the guest OS — graceful shutdown and restart.Hard Reboot: Forces an immediate power cycle — equivalent to pulling the power cord. Use when soft reboot is unresponsive.
Shut Off: Sends a power-off signal to the guest. The instance moves to Shutoff status and releases vCPU/RAM on the hypervisor (storage persists and is billed).Start: Powers on a Shutoff instance. The instance boots from where it was shut off.
Suspend: Saves the full memory state to disk (like hibernation). vCPU/RAM are released but disk state is preserved. Transition to Suspended status.Resume: Restores the memory state and continues execution. Faster than a cold boot.
Pause: Freezes the instance in memory — all vCPUs halt but memory is retained. No disk write. Transition to Paused status.Unpause: Resumes execution from the frozen state instantly.
Pausing keeps the instance consuming vCPU and RAM on the host. Use Suspend for longer idle periods to free host resources.
Boots the instance from a rescue image (small Linux ISO) with the original root disk attached as a secondary disk. Use to recover from boot failures or lost root access.See Rescue an Instance for the full workflow.
Permanently terminates the instance. If the root volume was created with Delete Volume on Instance Delete enabled, the volume is also deleted.
Instance deletion is irreversible. Ensure all data is backed up or the root volume is set to persist before deleting.

Instance Metadata & Tags

and tags let you annotate instances for inventory management, cost attribution, automation, and search. Both are stored in the compute service and returned in API responses.

Metadata (Key-Value Pairs)

Open instance detail

Click the instance name to open its detail view.

Edit metadata

In the Overview tab, scroll to the Metadata section. Click Edit (pencil icon). Add key-value pairs — click + after each entry.
Example KeyExample ValueUse Case
environmentproductionEnvironment classification
teamplatform-engTeam/cost-center ownership
project_codePROJ-042Billing/chargeback code
backup_policydailyBackup automation trigger
patch_groupgroup-APatch management segmentation
roleweb-serverInstance role in the stack

Save

Click Save. Metadata is immediately available in API responses and automation tools.

Tags

Tags are simple string labels — no key, just a value. They enable fast instance filtering across projects.
Add tags to an instance
openstack server set --tag production --tag web-tier --tag pci-zone my-instance
List instances by tag
openstack server list --tags production,web-tier
Remove a tag
openstack server unset --tag pci-zone my-instance

Metadata vs Tags

FeatureMetadata (Properties)Tags
FormatKey-value pairsSimple strings
Filtering--property key=value--tags tag1,tag2
Hypervisor accessNot available inside guestNot available inside guest
Config driveAvailable via /openstack/latest/meta_data.jsonNot included
Use forStructured data, automation keysQuick labels, grouping
Metadata is accessible from inside the instance via the metadata service at http://169.254.169.254/openstack/latest/meta_data.json — useful for cloud-init scripts and bootstrap automation.

Instance Detail View

Click an instance name to open its detail page with tabs:
TabContents
OverviewFlavor, image, IP addresses, security groups, key pair, metadata
LogConsole output from the last boot (cloud-init, kernel messages)
ConsoleLive browser-based VNC console — no SSH required
Action LogHistory of all operations performed on this instance

Next Steps

Attach a Volume

Add persistent storage to running instances

Configure Networking

Assign floating IPs and manage security groups

Live vCPU/RAM Scaling

Scale compute resources without rebooting

Resize Instance

Change instance flavor with a controlled reboot