Skip to main content

​
Troubleshooting

Identify which layer failed before changing anything:

​
Mirror Manager does not reopen after SSH disconnect

Reconnect and run:
Do not rerun install.sh simply to reopen the GUI.

​
Download and Prepare was interrupted

Run Menu 2 again after confirming another copy is not already active. The workflow can reuse completed files and partial data. Do not delete ACPS .part files merely to force a clean retry. Useful checks:

​
Readiness is NOT READY or FAIL

Recheck in this order:
  1. saved Configuration
  2. Mirror Server IP
  3. prepared artifact status
  4. nginx configuration and service state
  5. live HTTP access to the files required by the selected mode
  6. Menu 4 again
Do not start the DP upgrade until readiness reports PASS.

​
Mirror Server IP changed

Update the IP in Configuration and run the required sequence again:
Do not reuse old Menu 7 commands because generated client artifacts contain the saved Mirror Server IP.

​
Phase 1 OS upgrade is blocked

The current Phase 1 engine distinguishes retryable conditions from integrity or evidence problems. Transient examples include:
  • DNS failure
  • HTTP/network failure
  • NTP failure
  • active apt/dpkg lock
Correct the cause and follow the current status/recovery guidance. Do not delete /opt/aelladata/os-upgrade to make the state disappear.

​
Ubuntu 22.04 looks unhealthy

Jammy is an intermediate OS state in a Full upgrade. On affected DP builds, the product runtime may be unavailable or incompatible during this hop. Do not repair kubelet, Docker, or the DP runtime merely to make the intermediate state healthy. Continue the validated OS workflow to Ubuntu 24.04, then complete Phase 2.

​
Worker bringup fails

Check:
  • the correct DL/DA worker list was saved in Mirror Manager
  • master-to-worker SSH reachability
  • the configured common aella worker password
  • node IP addressing and MTU
  • NTP and DNS
  • Kubernetes node state
  • that common staging completed on all nodes before final master orchestration
Regenerate Menu 7 output after any relevant configuration change.

​
Rollback

The project does not provide OS or DP-runtime downgrade commands. If a destructive failure cannot be safely recovered using the current release tooling, restore the approved hypervisor snapshot.