Host Authorized with Differing/Unexpected Hostname

Problem

• A host authorized with Platform9 is displaying its name as "localhost", its FQDN is omitted, or it otherwise differs from that of its actual, expected hostname.
• If the host is authorized with Platform9 Managed OpenStack, instance operations for instances residing on the host may fail with the following error.

{u'message': u' [host_name] is not a valid node managed by this compute host.'

Environment

• Platform9 Managed OpenStack - All Versions
• Platform9 Managed Kubernetes - All Versions

Cause

• The pf9-hostagent service obtains the hostname using the Python function socket.getfqdn().

# python -c 'import socket; print socket.getfqdn()'

• Verify if the nslookup command is returning the updated hostname. This would indicate that there's no issue with DNS resolution

nslookup [updated hostname]

• If nslookup command returns the right hostname, check if there are any old entries associated with the older hostname in the /etc/hosts.

Resolution

Ensure the following:

1. socket.getfqdn() function returns the expected hostname.
2. DNS resolution is functional.
3. For the IPv4 localhost entry in /etc/hosts, the hostname is included as the first entry.

127.0.0.1 [hostname.domain.tld] localhost

4. Restart the pf9-hostagent service.

# systemctl restart pf9-hostagent

Additional Information

If using pf9-express, ensure the Manage Hostname option is set to true during the configuration process to automatically add the hostname entry for localhost on deployment.

# ./pf9-express -s...Manage Hostname [true false] [false]: --> accepted: true

The option may be set directly in pf9-express.conf. file as well.

manage_hostname|true