Utilities-and-troubleshooting.md

Utilities and Troubleshooting

Updating the parameter file

In case you need to modify the parameter, file or fix anything and rerun the pipeline, perform the following:

• Make your change to the parameter file
  
• Hijack the container
  
  • run docker ps to find the "nsx-t-install" container ID
  • run docker exec -it <nsx-t-install-id> bash
    
  • in the container run fly-reset (this basically will run set-pipeline for the nsx-t pipeline)

Cleaning up

If you want to re-run from scratch do the following to cleanup:

• Stop containers - docker stop $(docker ps -a -q)
  
• Remove the containers - docker rm $(docker ps -aq)
• Delete images - docker rmi -f $(docker images -a -q)
  
• clear cache - docker system prune --all
  
• unregister the vCenter extension - https://docs.vmware.com/en/VMware-NSX-T/2.2/com.vmware.nsxt.admin.doc/GUID-E6E2F017-1106-48C5-ABCA-3D3E9130A863.html

Commonly seen issues:

NSX manager OVA auto download did not work; container exited.
Solution: it's likely the myvmware credentials were misspelled and did not work. Make sure the username is an email address, and both username and password are enclosed by single quotes (e.g. -e VMWARE_USER='abc@efg.com' -e VMWARE_PASSWORD='pwd$pecial')

There should be 3 containers related to concourse running (web, worker, postgres). But one or more are missing from docker ps output.
Solution: use docker ps -a to find the exited container(s). Then check the logs with docker logs <container_id>. If the error message signals the container exited due to insufficient disk space on the jumphost, clean up unused docker volumes with:
docker volume rm $(docker volume ls -qf dangling=true)
Then clean up the existing containers:

docker exec -it nsx-t-install bash
cd /home/concourse
docker-compose down
exit
docker stop nginx-server nsx-t-install
docker rm nginx-server nsx-t-install

Finally, rerun the pipeline container.

Ovftool fails to deploy NSX manager with OVA. Error from pipeline: "transfer failed; failed to send http data".
Solution: chances are the EXTERNAL_DNS environment variable passed to the container was unable to resolve vCenter's name. Rerun the docker container with a proper DNS nameserver.