Hostname inconsistent GUI / container

[cmonty14]

Hi,
I have configured this hostname in GUI (see screenshot):
ct102-haproxy.whl.meilocal.net

However, the CLI output for displaying Fully qualified domain name returns:
ct102-haproxy:~# hostname -f
ct102-haproxy.whl.meilocal.net.local

How can this inconsistency be fixed?

 

[Alwin]

You're using the domain name provided by your PVE host. Check /etc/hosts.

 

[Rhinox]

Hi,
I have configured this hostname in GUI (see screenshot):
ct102-haproxy.whl.meilocal.net

That it FQDN (fully qualified domain name), not hostname.
You should define hostaname as "ct102-haproxy" with dns-domain "whl.meilocal.net"...

 

[D0peX]

in DNS setting for the CT, you likely enabled 'Use host settings'. The host is probably using a .local domain.

 

[cmonty14]

This is /etc/hosts of Proxmox VE host:
root@pc4-svp:/etc/pve/local# cat /etc/hosts
127.0.0.1 pve.biszumbitterenen.de localhost.localdomain localhost
#127.0.1.1 pc4-pve.whl.meilocal.net pc4-svp
192.168.1.14 pc4-svp.whl.meilocal.net pc4-svp


# The following lines are desirable for IPv6 capable hosts
::1 localhost ip6-localhost ip6-loopback
ff02::1 ip6-allnodes
ff02::2 ip6-allrouters


What is wrong here?

And yes, in VM/CT configuration the setting in DNS is: Use host settings.
Where can I modify this for an existing VM/CT?
In the LXC config file?

FYI
This is the content of a LXC config file:
root@pc4-svp:/etc/pve/local# cat /etc/pve/lxc/102.conf
arch: amd64
cores: 1
hostname: ct102-haproxy.whl.meilocal.net
memory: 50
net0: name=eth0,bridge=vmbr0,gw=10.0.0.254,hwaddr=D6:B6:21:29:10:E6,ip=10.0.0.2/24,tag=1000,type=veth
onboot: 1
ostype: alpine
rootfs: images:102/vm-102-disk-1.raw,size=267596595
startup: order=1
swap: 8

 

[Alwin]

Check the searchdomain on the node, might be that the .local is there.

 

[cmonty14]

/etc/resolv.conf on Proxmox VE
root@pc4-svp:/etc/pve/local# cat /etc/resolv.conf
search whl.meilocal.net
nameserver 5.45.97.127
nameserver 82.141.39.32
nameserver 46.28.207.199

/etc/resolv.conf on CT
ct102-haproxy:~# cat /etc/resolv.conf
# --- BEGIN PVE ---
search whl.meilocal.net
nameserver 5.45.97.127
nameserver 82.141.39.32
nameserver 46.28.207.199
# --- END PVE ---

 

[Alwin]

If you use a other container, do you experience the same behavior?

 

[cmonty14]

Hm... interesting enough there are other containers where hostname -f returns correct FQHN.

 

[Alwin]

Hm... interesting enough there are other containers where hostname -f returns correct FQHN.

Are all of them alpine?

 

[cmonty14]

I collected this information:
ct102-haproxy:~# cat /etc/alpine-release
3.7.0
ct102-haproxy:~# hostname -f
ct102-haproxy.whl.meilocal.net.local

ct108-cal:~# cat /etc/alpine-release
3.6.2
ct108-cal:~# hostname -f
ct108-cal.whl.meilocal.net

This could indicate that the issue is related to Alpine release.

The LXC config is identical, though:
root@pc4-svp:/# cat /etc/pve/local/lxc/102.conf
arch: amd64
cores: 1
hostname: ct102-haproxy.whl.meilocal.net
memory: 50
net0: name=eth0,bridge=vmbr0,gw=10.0.0.254,hwaddr=D6:B6:21:29:10:E6,ip=10.0.0.2/24,tag=1000,type=veth
onboot: 1
ostype: alpine
rootfs: images:102/vm-102-disk-1.raw,size=267596595
startup: order=1
swap: 8

root@pc4-svp:/# cat /etc/pve/local/lxc/108.conf
arch: amd64
cores: 1
hostname: ct108-cal.whl.meilocal.net
memory: 32
net0: name=eth0,bridge=vmbr0,gw=10.0.0.254,hwaddr=1E:68:FE:40:78:57,ip=10.0.0.8/24,tag=1000,type=veth
onboot: 0
ostype: alpine
rootfs: images:108/vm-108-disk-1.raw,size=307M
startup: order=5
swap: 8
unprivileged: 1

 

[Alwin]

ct102-haproxy.whl.meilocal.net.local

Is that also what you see in the /etc/hosts and /etc/hostname?

Maybe alpine expects a dot on the end of the domain name?

 

[cmonty14]

ct108-cal:~# cat /etc/hosts

# --- BEGIN PVE ---
127.0.0.1 localhost.localnet localhost
::1 localhost.localnet localhost
10.0.0.8 ct108-cal.whl.meilocal.net ct108-cal
# --- END PVE ---
ct108-cal:~# cat /etc/hostname
ct108-cal
ct108-cal:~#
ct102-haproxy:~# cat /etc/hosts

# --- BEGIN PVE ---
10.0.0.2 ct102-haproxy.whl.meilocal.net ct102-haproxy
# --- END PVE ---
# begin generated by /etc/init.d/hosts
127.0.0.1 ct102-haproxy.whl.meilocal.net.local ct102-haproxy.whl.meilocal.net localhost
::1 ct102-haproxy.whl.meilocal.net.local ct102-haproxy.whl.meilocal.net localhost
# end generated

ct102-haproxy:~# cat /etc/hostname
ct102-haproxy

This output for ct102 shows an indication to the root cause: begin generated by /etc/init.d/hosts.

And this is the content of this file:
ct102-haproxy:~# cat /etc/init.d/hosts
#!/sbin/openrc-run

start() {
local start_tag='# begin generated'
local end_tag='# end generated'

local content=$(
cat <<-EOF
$start_tag by /etc/init.d/hosts
127.0.0.1 $(hostname).local $(hostname) localhost
::1 $(hostname).local $(hostname) localhost
$end_tag
EOF
)

if grep -q "^${start_tag}" /etc/hosts; then
# escape \n, busybox sed doesn't like them
content=${content//$'\n'/\\$'\n'}

sed -ni "/^${start_tag}/ {
a\\${content}
# read and discard next line and repeat until $end_tag or EOF
:a; n; /^${end_tag}/!ba; n
}; p" /etc/hosts
else
printf "$content" >> /etc/hosts
fi
}

Obviously this file was not executed on ct108.

 

[Alwin]

Which alpine versions are these two containers running?

To avoid the modifications try the following.

Modification of a file can be prevented by adding a .pve-ignore. file for it. For instance, if the file /etc/.pve-ignore.hosts exists then the /etc/hosts file will not be touched. This can be a simple empty file created via:

# touch /etc/.pve-ignore.hosts

Most modifications are OS dependent, so they differ between different distributions and versions. You can completely disable modifications by manually setting the ostype to unmanaged.

 

[cmonty14]

Hi,
both servers are running with same Alpine release 3.7.0.
However, ct102 was originally created with template "alpine-3.6-default_20171211_amd64" and then upgraded to Alpine 3.7, but ct108 was created with latest template "alpine-3.7-default_20171211_amd64".
This means the erroneous CT originates from a different template.

 

[Alwin]

This has apparently changed in that version, the template is pulled directly from LXC image repository. You could remove the init script or use the above mentioned ignore file.