BOSH loop during installation, caused by tile.service failure - Tanzu Evaluation Appliance 10.4.3

Products

• VMware Tanzu Platform Core

Issue/Introduction

• After installing the Tanzu Evaluation Appliance and powering on, the Opsman deploys and starts up successfully and can be accessed by the web client.
• You see the Bosh Director for Docker deployment task successfully installing in a loop, however, no other tiles deploy in the Opsman.
• From an SSH to the Eval Appliance, you see the tiler.service in exited state:
  
  sudo systemctl status tiler.service
  
  
• The tiler.service logging shows failures downloading files from https://usw1.packages.broadcom.com:
  
  

  Aug 14 18:14:29 tanzu-app run-tiler.sh[3461791]: {"type":"step_started","id":"bosh_product.deploying","description":"Installing BOSH"}[Tiler] [Steps] UploadFiles(cf/10.4.3) Max retry 6 reached giving up: Mover.Get: could not cache file: failed to download file: https://usw1.packages.broadcom.com/artifactory/tas-rel-eng-generic-prod-local/tas-releases/srt-10.4.3-build.3.pivotal got: Response status code is not ok: 403
  
  

  Aug 14 18:13:22 tanzu-app run-tiler.sh[3441538]: Exited with 0.

  Aug 14 18:13:22 tanzu-app run-tiler.sh[3441538]: {"type":"step_finished","id":"clean_up_bosh.cleaning_up","description":"Cleaning up BOSH director"}

  Aug 14 18:13:22 tanzu-app run-tiler.sh[3441538]: {"type":"exit","code":0}UploadFiles(cf/10.4.3): step UploadFiles(cf/10.4.3) failed: Mover.Get: could not cache file: failed to download file:

  Aug 14 18:13:22 tanzu-app systemd[1]: tiler.service: Main process exited, code=exited, status=1/FAILURE

Environment

Tanzu Evaluation Appliance 10.4.3

Cause

The appliance pre-loads install packages so it doesn't need internet access. A recent build change left an extra log file next to one of those packages, which makes the installer think the package is corrupted and tries to re-download it from a repo that is not accessible to the appliance.

 

Resolution

This is resolved in the Tanzu Evaluation Appliance 10.4.4.

 

Workaround:

The workaround for this requires identification of the problem log files, and removal of the log files during to first boot of the appliance.

 

1. Identify the problem log files using the following command from an SSH to the appliance:
   
   
   ls -laR /mnt/disks/docker-data/tiler/cache/
   
   
2. Reboot the appliance VM, as soon as it boots, SSH back into the VM and delete the *.aria2.log from any directories under /mnt/disks/docker-data/tiler/cache identified in Step 1.

 

NOTE: It is preferred to use another version of the Tanzu Evaluation Appliance as rebooting is not supported, see the Release Notes for Tanzu Platform Evaluation Appliances