CSPACE-LAUNCH-05: Workload does not auto-restart on failure (tee-restart-policy)¶
| Property | Value |
|---|---|
| Severity | medium |
| Platform | GCP Confidential Space (AMD SEV-SNP) |
| Lifecycle stage | deploy |
| Mode | Passive: read-only cloud API calls. |
What it checks¶
Example failures:
- tee-restart-policy is 'Always' — a single poisoned input re-runs indefinitely.
How to fix¶
- Set tee-restart-policy to Never (or a bounded OnFailure).
After the fix, scan again: the check should report PASS.
Scan inputs¶
The untrust scan options this check reads:
| Option | Description |
|---|---|
--gcp-instance |
Confidential VM instance name. |
--gcp-project |
GCP project ID. |
--gcp-zone |
Zone of the Confidential VM. |
Permissions¶
The cloud API calls this check makes, read from the scanner's source:
| Cloud | API call | Permission | Access |
|---|---|---|---|
| Google Cloud | compute.instances.get |
compute.instances.get |
read |
See Required permissions for every check on this platform.
Generated from untrust 1.3.0 (commit 7ef462d9a0c3).