LXC appliance: DHCP discover request does not include hostname

[Dude4Linux]

For appliance versions up through v15.0, the DHCP request issued by udhcpc does not include the hostname. This prevents the upstream DHCP server (usually dnsmasq) from dynamically registering the hostname in DNS. In the past, this was largely an inconvenience which I overcame by pre-defining static entries in the upstream DHCP/DNS. Since developing the Ansible appliance, I've become increasingly frustrated by the need to work around the problem in Ansible playbooks. Lately I've tried adding mods to the LXC containers so that the containers will dynamical register their hostname.

When I recently began looking at fixing a problem in confconsole, I realized that there was a way of addressing the issue by modifying the confconsole code.

From man interfaces:

The dhcp Method
This method may be used to obtain an address via DHCP with any of the tools: dhclient, pump, udhcpc, dhcpcd. (They have been listed in their order of precedence.) If you have a complicated DHCP setup you should note that some of these clients use their own configuration files and do not obtain their configuration information via ifup.
Options
        hostname hostname
                Hostname to be requested (pump, dhcpcd, udhcpc)
        metric metric
                Metric for added routes (dhclient)
        leasehours leasehours
                Preferred lease time in hours (pump)
        leasetime leasetime
                Preferred lease time in seconds (dhcpcd)
        vendor vendor
                Vendor class identifier (dhcpcd)
        client client
                Client identifier (dhcpcd)
        hwaddress address
                Hardware address.

In the testing I've done, adding the hostname to any interface using DHCP in /etc/network/interfaces results in udhcpc including the hostname in the DHCP discover request. In my testing, I also found that when the hostname is changed by confconsole, it must also change /etc/network/interfaces and restart networking to issue a new DHCP request.

[Dude4Linux]

@JedMeister - I've issued a new PR that should close this issue.
Update confconsole for v15.1 #25

[JedMeister]

I just realised that I hadn't tagged this issue properly. Apologies on that. Also apologies on the really slow turnaround on this... I have a few things that I'd like to tidy up first, but I aim to get back to this sometime within the next week or 2 and tidy up (i.e. finish going over the PR properly and hopefully merge).

[JedMeister]

The v16.0 LXC appliance has been put aside for now as there are significant issues running privileged Debian Buster LXC containers on a Buster host (see #1520).

As such this issue is also deferred for now...

[JedMeister]

This should now be resolved as of v18.x.

Having said that, we don't currently have an LXC appliance, so it's untested in this context. However, I'm feeling very confident, so closing this one.

[fanyangto]

I can confirm this problem still exists in the v18.x. My debian-12-turnkey-wordpress_18.2-1_amd64.tar.gz still could not obtain DHCP hostname.

I also find that LXC created with debian-12-standard_12.2-1_amd64.tar.gz works with DHCP names. It's the Turnkey Core templates that has the DHCP no name issue.

[JedMeister]

Hi @fanyangto

Thanks for reporting.

Although I'm not completely sure that it's exactly the same issue that was discussed here?

This discussion was in context of vanilla LXC (and the LXC host appliance that we produced at the time). On Proxmox, unless you touch /etc/network/.pve-ignore.interfaces inside the guest, the Proxmox host handles the LXC guest's /etc/network/interfaces file.

I've done some quick and dirty testing and I couldn't see any obvious difference between the interfaces file between debian-12-turnkey-wordpress_18.2-1_amd64.tar.gz and debian-12-standard_12.2-1_amd64.tar.gz? If DHCP is chosen in Proxmox at launch, then the hostname isn't added to the interfaces file of either of them. In either guest, if I edit the interfaces file to add the hostname, both are reset to the Proxmox default on guest reboot.

So could you please open a new issue describing the problematic behavior in as much detail as possible, ideally with step by step instructions on how I can reproduce it locally. Personally I always use static IPs for my servers but if I can reproduce your problem on my local Proxmox server, then I'm confident that I can fix it - or at the least, document a workaround...

[fanyangto]

Hi @fanyangto

Thanks for reporting.

Although I'm not completely sure that it's exactly the same issue that was discussed here?

This discussion was in context of vanilla LXC (and the LXC host appliance that we produced at the time). On Proxmox, unless you touch /etc/network/.pve-ignore.interfaces inside the guest, the Proxmox host handles the LXC guest's /etc/network/interfaces file.

I've done some quick and dirty testing and I couldn't see any obvious difference between the interfaces file between debian-12-turnkey-wordpress_18.2-1_amd64.tar.gz and debian-12-standard_12.2-1_amd64.tar.gz? If DHCP is chosen in Proxmox at launch, then the hostname isn't added to the interfaces file of either of them. In either guest, if I edit the interfaces file to add the hostname, both are reset to the Proxmox default on guest reboot.

So could you please open a new issue describing the problematic behavior in as much detail as possible, ideally with step by step instructions on how I can reproduce it locally. Personally I always use static IPs for my servers but if I can reproduce your problem on my local Proxmox server, then I'm confident that I can fix it - or at the least, document a workaround...

Good point Jeremy. I have opened a new case #1992 per your suggestion.