Release v1.6.0-rc.1
SDKv2 to v3 migration
The provider's API layer has been migrated from osc-sdk-go@v2 to osc-sdk-go@v3. This migration should be fully backward-compatible, no changes needed and no impact on your Terraform files.
Note: This is a release candidate. Please report any issues you encounter.
Provider improvements
-
Read batching: During
planandapply, the provider now collects individualReadcalls for VMs, Volumes, Security Groups, Nets, and Subnets and merges them into single bulk API requests. Instead of one API call per resource, reads arriving within a short time window are grouped together. On large infrastructures, this means dramatically fewer API calls and faster operations. -
Exponential backoff on throttling: When the API returns throttling responses, retries now use exponential backoff instead of fixed intervals. This reduces the chance of hitting rate limits and resulting timeouts.
-
409 conflict handling: Conflict errors are now handled with finer granularity, eliminating cases where the provider could enter infinite retry loops and eventually time out.
-
Cleaner debug logs: HTTP request/response logs (
TF_LOG=DEBUG) are now formatted as multi-line output, making debugging requests easier.
About read batching
Increase Terraform's parallelism so the batcher can group more reads into single API calls:
terraform plan -parallelism=20 # (defaults to 10)To make it permanent, set it via environment variables:
export TF_CLI_ARGS_plan="-parallelism=20"See the Terraform CLI documentation for more details on TF_CLI_ARGS.