multiple-instances

Multiple Compute Instances

Boot a fleet of identical OpenStack compute instances (Nova) with a single count, naming them from a shared prefix and a 1-based index. This is the go-to pattern for stateless tiers — web front-ends, workers, batch nodes — where you want N copies of the same machine on one tenant network.

Primary search phrase: Terraform OpenStack multiple instances count

Architecture

flowchart LR
    TF[Terraform] -->|count = N| VM[openstack_compute_instance_v2]
    DN[data: networking_network_v2] -->|network id| VM
    DI[data: images_image_v2] -->|image id| VM
    DF[data: compute_flavor_v2] -->|flavor id| VM
    VM -->|web-1| N1[(Instance 1)]
    VM -->|web-2| N2[(Instance 2)]
    VM -->|web-N| NN[(Instance N)]

Loading

Network, image and flavor are each resolved once by name with a data source, then shared across every instance the count meta-argument creates. Outputs use splat expressions so you get lists of IDs and IPs back in creation order.

Usage

export OS_CLOUD=openstack          # or set `cloud` in terraform.tfvars
cp terraform.tfvars.example terraform.tfvars
terraform init
terraform plan
terraform apply

Inputs

Name	Description	Type	Default
cloud	clouds.yaml entry to use	string	"openstack"
instance_count	Number of instances to create	number	3
instance_name_prefix	Prefix for instance names	string	"web"
flavor_name	Flavor (size)	string	"m1.small"
image_name	Glance image to boot	string	"ubuntu-22.04"
network_name	Tenant network to attach	string	"private"
key_pair_name	Existing key pair for SSH (optional)	string	""
security_group_names	Security groups	list(string)	["default"]
tags	Instance tags	list(string)	see variables.tf

Outputs

Name	Description
instance_ids	UUIDs of all instances
instance_names	Names of all instances
access_ips	First IPv4 address of each instance
network_id	Network the instances are attached to

Best practices

• Why this approach: count is the simplest way to stamp out N identical, interchangeable instances. Data-source lookups by name keep the example portable (no UUIDs in code), and splat outputs give you tidy lists to feed into a load balancer or DNS record set.
• Common mistakes: Using count when instances are not interchangeable — reordering or removing a middle element renumbers everything after it and forces replacements. If each instance differs (per-AZ, per-name), prefer for_each (see availability-zones). Also avoid hard-coding the index into anything stateful (volumes, floating IPs) under count.
• Scaling considerations: Raise instance_count to scale out; mind project quotas (instances, cores, RAM). For large fleets put this behind the compute module and front it with a load balancer.
• Performance considerations: All instances schedule in parallel, so a big jump in instance_count can briefly stress the scheduler and image cache. Match the flavor to the workload rather than over-provisioning every node.
• Cost considerations: N instances bill N times while ACTIVE. Tag everything (done here) for cost attribution and terraform destroy dev fleets when idle.

Security considerations

• Every instance shares security_group_names; define least-privilege groups explicitly rather than relying on default — see security/security-group.
• Inject SSH access via a managed key pair, never passwords, and never bake secrets into user-data; use application credentials or a secrets manager.
• A larger fleet is a larger attack surface — keep images patched and rebuild rather than long-lived in-place upgrades.

Troubleshooting

Symptom	Likely cause	Fix
No valid host was found	No host has capacity for the flavor across the fleet	Lower instance_count, try a smaller flavor or another AZ; check openstack hypervisor stats show
Quota exceeded	Project instance/cores/RAM quota hit by the fleet size	Raise quota or lower instance_count (quotas examples)
Replacing instances on every apply	count index shifted after removing a middle element	Append/remove from the end, or move to for_each keyed by name
Image <name> not found	Wrong image_name or image not visible to the project	openstack image list; check image visibility
Network <name> not found	Wrong network_name or no access	openstack network list
Provider auth errors	Bad/missing clouds.yaml or OS_CLOUD	See provider configuration

Cleanup

terraform destroy

Further reading

• Provider configuration & clouds.yaml
• OpenStack provider — compute instance docs
• Terraform count meta-argument
• Single instance example