Troubleshooting
Problem
Terraform Cloud Self-hosted agents fail during the automatic upgrade process when upgrading from 1.28.6+ to 1.28.9.
The issue occurs after the core binary is automatically updated and restarted.
Symptom
Agents configured with auto-update enabled may fail immediately during restart while agent core upgrading from:
1.28.6→1.28.91.28.7→1.28.91.28.8→1.28.9
Observed logs:
2026-05-13T03:22:15.823Z [INFO] agent: Core update is available: version=1.28.9
2026-05-13T03:22:17.098Z [INFO] agent: Core successfully updated: version=1.28.9
2026-05-13T03:22:17.101Z [INFO] agent: Core plugin is shutting down
2026-05-13T03:22:17.452Z [INFO] core: Shutdown complete
2026-05-13T03:22:17.462Z [INFO] plugin: plugin process exited: plugin=/home/tfc-agent/bin/tfc-agent-core id=20
2026-05-13T03:22:17.578Z [INFO] plugin: plugin process exited: plugin=/home/tfc-agent/bin/tfc-agent-core id=41
2026-05-13T03:22:17.578Z [ERROR] agent: Failed starting core plugin: error="failed configuring core: failed evaluating config: invalid job type \"native_task\""
2026-05-13T03:22:17.578Z [INFO] agent: Shutting down
Graceful shutdown completeThe agent shuts down after failing to start the upgraded core plugin.
Cause
The issue appears to be related to the automatic core upgrade flow in certain 1.28.x versions.
This occurs specifically after the core replacement step during upgrades from 1.28.6+ versions to 1.28.9.
Direct installation of 1.28.9 is not affected.
Environment
- HCP Terraform / Terraform Self-hosted Agents
- Linux ARM64 environment
- Agent versions:
- Affected:
1.28.61.28.71.28.8
- Target version:
1.28.9
- Affected:
Diagnosing The Problem
Review the agent startup logs and verify whether the failure occurs immediately after the automatic core update.
Affected agents show:
[INFO] agent: Core successfully updated: version=1.28.9followed by:
[ERROR] agent: Failed starting core plugin: error="failed configuring core: failed evaluating config: invalid job type \"native_task\""
Findings:
- Direct
1.28.9deployment works - Auto-upgrade regression isolated to later
1.28.xversions - Earlier upgrade paths (
1.27.0,1.28.0,1.28.5) succeed
Resolving The Problem
Use one of the following workarounds.
Option 1 :
Deploy agent version 1.28.9 directly instead of relying on auto-upgrade from affected versions.
Direct installation of 1.28.9 works successfully.
Option 2 :
Disable automatic core upgrades by setting the following environment variable:
TFC_AGENT_AUTO_UPDATE=disabledExample:
docker run \
-e TFC_AGENT_AUTO_UPDATE=disabled \
hashicorp/tfc-agent:1.28.8After disabling auto-update, manually manage the agent version upgrades, it will not upgrade the core to 1.28.9 automatically.
Note :- For the VM hosted agents, only option 1 is applicable.
Related Information
Document Location
Worldwide
Was this topic helpful?
Document Information
Modified date:
08 June 2026
UID
ibm17272788