Wrong EC2 hostname set if DHCP Options Set has multiple domain-name

[SC]

Cloud Provider: AWS.
Cloud-init v. 19.3-2.amzn2.
Config says system_info: distro: amazon

Per https://docs.aws.amazon.com/vpc/latest/userguide/VPC_DHCP_Options.html the VPC DHCP Options set can have a domain-name value with multiple domain names separated by spaces. However, if using this approach, cloud-init sets an incorrect hostname. It appears to concatenate the unqualified host name to the Options Set host names without spaces, and truncating to a certain number of characters.

This is with "set-hostname" and "update-hostname" modules enabled, but no other relevant cloud-init config.

For example, if your unqualified hostname is ip-10-30-10-155 and your DHCP Options Set domain-name is "foo.bar.example.com baz.bat.example.com" you might get an incorrect hostname such as "ip-10-30-10-155.foo.bar.example.combaz.bat.ex"

This problem appears to be due to misinterpretation of the output of the EC2 metadata service for "local-hostname", which responds with output similar to above, but with spaces retained and no truncating, for example: "ip-10-30-10-155.foo.bar.example.com baz.bat.example.com"

The cloud-init log output also mentions this problem, for example you might have logs which look like:

Mar 18 00:31:33 cloud-init[3825]: cc_set_hostname.py[DEBUG]: Setting the hostname to ip-10-30-10-155.us-west-2.i.test.abcdefg.example.com i.test.abcdefg.example.com
us-west-2.test.abcdefg.example.com test.abcdefg.example.com us-west-2.compute.internal (ip-10-30-10-155)
Mar 18 00:31:33 cloud-init[3825]: util.py[DEBUG]: Running command ['hostnamectl', 'set-hostname', 'ip-10-30-10-155.us-west-2.i.test.abcdefg.example.com i.test.abcdefg.example.com us-west-2.test.abcdefg.example.com test.abcdefg.example.com us-west-2.compute.internal'] with allowed return codes [0] (shell=False, capture=True)
Mar 18 00:31:33 cloud-init[3825]: __init__.py[DEBUG]: Non-persistently setting the system hostname to ip-10-30-10-155.us-west-2.i.test.abcdefg.example.com i.test.abcdefg.example.com us-west-2.test.abcdefg.example.com test.abcdefg.example.com us-west-2.compute.internal
Mar 18 00:31:33 cloud-init[3825]: util.py[DEBUG]: Running command ['hostname', 'ip-10-30-10-155.us-west-2.i.test.abcdefg.example.com i.test.abcdefg.example.com us-we
st-2.test.abcdefg.example.com test.abcdefg.example.com us-west-2.compute.internal'] with allowed return codes [0] (shell=False, capture=True)
Mar 18 00:31:33 cloud-init[3825]: util.py[WARNING]: Failed to non-persistently adjust the system hostname to ip-10-30-10-155.us-west-2.i.test.abcdefg.example.com i.te
st.abcdefg.example.com us-west-2.test.abcdefg.example.com test.abcdefg.example.com us-west-2.compute.internal
Mar 18 00:31:33 cloud-init[3825]: util.py[DEBUG]: Failed to non-persistently adjust the system hostname to ip-10-30-10-155.us-west-2.i.test.abcdefg.example.com i.test.abcdefg.example.com us-west-2.test.abcdefg.example.com test.abcdefg.example.com us-west-2.compute.internal
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/cloudinit/distros/__init__.py", line 230, in _apply_hostname
    util.subp(['hostname', hostname])
  File "/usr/lib/python2.7/site-packages/cloudinit/util.py", line 2105, in subp
    cmd=args)
ProcessExecutionError: Unexpected error while running command.
Command: ['hostname', 'ip-10-30-10-155.us-west-2.i.test.abcdefg.example.com i.test.abcdefg.example.com us-west-2.test.abcdefg.example.com test.abcdefg.example.com us-west-2.compute.internal']
Exit code: 1
Reason: -
Stdout:
Stderr: hostname: the specified hostname is invalid

[Paride Legovini (paride)]

Thanks for taking the time to report this bug. Multiple domain names are currently not supported; the case falls in the warning given in the page you pointed to:

---
Some Linux operating systems accept multiple domain names separated by spaces. However, other Linux operating systems and Windows treat the value as a single domain, which results in unexpected behavior.
---

I'm not sure of what the Linux distributions supporting multiple domain names are doing; I imagine the first one is expected to become the true hostname of the instance, while all of them have to be added at least to /etc/hosts. In any case I agree it is something we should support.

[James Falcon (falcojr)]

Tracked in Github Issues as https://github.com/canonical/cloud-init/issues/3659