terraform_update_timeout
Description
The terraform_update_timeout parameter controls how long Terraform waits for managed node group update operations to complete. On large clusters (50+ nodes), rack updates trigger rolling node replacements that can exceed the default 2-hour timeout. This parameter lets you extend that window.
Default Value
The default value for terraform_update_timeout is 2h (2 hours).
Use Cases
- Large Clusters: Extending the timeout for clusters with 50+ nodes where rolling updates take longer than 2 hours.
- Slow Rolling Updates: When node replacements are throttled by
node_max_unavailable_percentage, the total update time increases.
Setting Parameters
To set the terraform_update_timeout parameter, use the following command:
$ convox rack params set terraform_update_timeout=3h -r rackName
Setting parameters... OK
This command sets the Terraform update timeout to 3 hours.
Additional Information
The value must be a valid Go duration string (e.g., 2h, 90m, 2h30m). This timeout applies to all EKS node group resources, including primary, build, and additional node groups. The default value of 2h matches the previously hardcoded behavior, so existing racks are unaffected.
See Also
- node_max_unavailable_percentage controls how many nodes update simultaneously (affects total update duration)
- additional_node_groups_config for configuring additional node groups (all respect this timeout)
