Introduction
Problem
Very large workspaces may fail during OPA policy evaluations on large workspaces that are in agent execution mode.
Prerequisites (if applicable)
- Terraform Enterprise v202309-1 through Terraform Enterprise v202311-1
Cause
- This issue is caused by a regression introduced in v202309-1 which prevents OPA policy evaluations from taking place on the agent rather than on the host. This can cause the host to come under resource pressure, particularly where infrastructure was provisioned assuming that policy evaluations will occur on agents as intended
Solution:
-
Upgrade to Terraform Enterprise v202312-1 or newer
Outcome
Upon upgrade, agents will perform OPA policy evaluations on the agent