🚀 Looking for Testers – New Config Files Feature!

[michelroegl-brunner]

Feature got merged in the Main Repository, For feature request please use the correct Issue Template!

Hi everyone

We're looking for testers to try out our new config files feature! If you're interested in getting early access and helping us improve it, we'd love your feedback.

How to test

We have created a test branch for this feature so we do not disturb productive use of our scripts. Keep in mind that this test branch may not include latest bug fixes or improvements.
Do not use this for productive systems.
This is only for development testing and feedback. You can install the scripts as you are used to. There is no support for alpine at the moment. Please report all errors you encounter or feedback you have as answer to this thread, not as a new issue or disscusion please. This will get closed immediately. There is no support when using this scripts besides problems with the config files.

Bash command to start LXC install.

You can test it by using our scripts in the development Repo.
After starting the script use point 4 Use Config File

###Config file

You need to create a file on your PVE Host to supply the configs befor running the scripts. It is defined by us to be /opt/community-scripts/.settings but you can change that in the prompt to whatever you wish.

If all or some settings are not present in the file, default values get used.

What settings are available?

The following settings can be provided with this file at the moment (subject to changes).

Caution

There is some input sanity checking missing, if wrong values are entered the script may break

Name	Value
CT_ID	ID for LXC,can be a single number or a range 100-9999
var_os	debian or ubuntu
var_version	11 or 12 for Debian 20.04, 22.04, 24.04, 24.10 for Ubuntu
CT_TYPE	1 unprivileged or 0 privileged
PW	password for root
HN	hostname of the container
DISK_SIZE	size of the disk in GB
CORE_COUNT	number of cores
RAM_SIZE	size of ram in MiB
BRG	Name of the bridge (vmbr0)
NET	IP-Address/CIDR (192.168.0.1/24)
GATE	Gateway Address (192.168.0.254)
APT_CACHER_IP	Ip Addresse of APT-Cacher (192.168.0.2)
DISABLEIP6	yes or no
MTU	'1500'
SD	Search domain (example.com)
NS	IP of Nameserver (192.168.0.100)
MAC	Mac address of device
VLAN	VLAN ID (100)
TAGS	Custom Tags
SSH	yes or no
SSH_AUTHORIZED_KEY	SSH Key
VERB	yes or no

Example file

Each value needs to be in double qoutes.

DISK_SIZE="25"
CORE_COUNT="33"
RAM_SIZE="4500"
HN="TEST"
BRG="vmbr1"
NET="10.10.10.171/24"
GATE="10.10.10.254"
VLAN="100"
SSH="yes"
TAGS="dev;testing"
SSH_AUTHORIZED_KEY="REDACTED"

[mon5termatt]

Can DHCP be an option?

[michelroegl-brunner]

Dhcp is the default. If you dont set a ip address in the config file, dhcp is used.

[mon5termatt]

cool. i have a few changes i want to make to each machine and that makes it easier if it asks if not provided via file.

[mon5termatt]

Sorry for the formatting, its being dumb, Im having some small issues when providing the gateway but not the IP. i want it to grab a dhcp lease in the VLAN 2 Range of 10.2.0.0/24

BRG="vmbr0"
GATE="10.2.0.1"
VLAN="2"
SSH="yes"
TAGS="dev;testing"
PW="REDACTED"

💡 Using Config File on node pve
💡 Using config File: /root/.settings
🖥️ Operating System: debian
🌟 Version: 12
📦 Container Type: Unprivileged
🔐 Root Password: ********
🆔 Container ID: 100
🏠 Hostname: dockge
💾 Disk Size: 18 GB
🧠 CPU Cores: 2
🛠️ RAM Size: 2048 MiB
🌉 Bridge: vmbr0
📡 IP Address: DHCP
🌐 Gateway IP Address: Default
🚫 Disable IPv6: No
⚙️ Interface MTU Size: 1500
🔍 DNS Search Domain: HOST
📡 DNS Server IP Address: HOST
🏷️ Vlan: 2
📡 Tags: dev;testing
🔑 Root SSH Access: yes
🔑 SSH Authorized Key: None
🔍 Verbose Mode: No
🚀 Creating a Dockge LXC using the above settings
✔️ Using local for Template Storage.
✔️ Using local-lvm for Container Storage.
✔️ Updated LXC Template List

⠼400 Parameter verification failed.
net0: invalid format - format error
net0.ip: invalid format - value does not look like a valid ipv4 network configuration

pct create <vmid> <ostemplate> [OPTIONS]
bash: line 201: exit: A problem occured while trying to create container.: numeric argument required

[mon5termatt]

I have a error in the script, i fix that tomorrow. But why would you want to spezifiy the GW when DHCP. I have a strong Network engineering background, and this makes no sense for me^^

Just skips me a step. I have everything on that vlan (2) with the GW of 10.(2).0.1, but I would still want to specify the IP

[michelroegl-brunner]

But you get the correct gateway with dhcp anyways? I dont see the benfit tbh.
And proxmox dont allows dhcp + gatway as per my testing.

[michelroegl-brunner]

nevermind, it is working now as you wanted.

[mon5termatt]

But you get the correct gateway with dhcp anyways? I dont see the benfit tbh. And proxmox dont allows dhcp + gatway as per my testing.

i was not aware of that.

[michelroegl-brunner]

But you get the correct gateway with dhcp anyways? I dont see the benfit tbh. And proxmox dont allows dhcp + gatway as per my testing.

i was not aware of that.

Nah, this was a bug? on my node, it worked after a few trys.

[mon5termatt]

Just an idea but only prefill and skip the options that are defined in the file and still prompt for the others.

[mon5termatt]

So example...

Script= read from script.

ID="200-299" SPECIFY RANGE FOR ID, IF 200 IS TAKEN USE 201 ETC.

DISK_SIZE="SCRIPT"
CORE_COUNT="SCRIPT"
RAM_SIZE="SCRIPT"
HN="SCRIPT"

BRG="vmbr0"
NET="DHCP"
GATE="10.2.0.1"

[michelroegl-brunner]

I will have a look at this an see what i can take from your feedback.
The id i dont see why you would ever set that manually, i would just leave this as automatic.
The Ip Range i have to see if this is possible somehow, but i dont think so.

[LilTrublMakr]

I could see ranges helpful for people that use dev/staging/prod containers. e.g. if the dev is in range 300-399, staging is in 200-299, and prod in 100-199. It could be helpful if someone uses multiple ENV (or settings in this case) files for those that have a setup like this.

As for IP address: in my instance, I use the machine ID as the last octet of the IP address. e.g. if the machine ID is 111, the IP would be 192.168.12.111/24 with a vlan tag. Though this may or may not be a common practice, it does work well and make things slightly easier to manage.

[mon5termatt]

Yeah I use a similar scheme.

If it has the ID of 211 its 10.2.0.11 (VLAN 2)

If the ID is 305 its 10.3.0.5 (VLAN 3)

[michelroegl-brunner]

ID Ranges are implemented. you can use CT_ID="100-200" for example. It checks for the next available ID.

[alexanderjerome]

Hey! I'm really glad to see this feature being tested, I'm happy to test it.

A couple of ideas that came to me when I read the initial post:

1. Post installation scripts - user can define commands to run afterwards. There could be a dedicated directory for post-install scripts to use, i.e. logging the root password to Vaultwarden, automating GPU passthrough, installing Tailscale / Netbird, bind mounting directories with proper permissions, etc. The user can provide their own scripts for this.

This feature can also piggy back onto @LilTrublMakr 's comment about ip addressing - I have the exact same practice, but instead of modifying the configuration script to handle this, the user could run their own custom script to update the LXC / VM settings to their preferences.

2. Settings / config files should be distinguished by script being run. It could be possible, for the user to contain configuration files for each individual automation script, so they can create a repeatable template that is specific to the service being installed. This way, should the user need to recreate their service, its quicker to reproduce and automate. The configuration file can then override the customization variables in the original build function, and possibly given a flag (-y to accept values on config file and complete the installation without interaction).

Above all, thanks for keeping ttecks legacy alive and well. It seems like every other day a new script appears in the repo, so kudos to the maintainers!

[michelroegl-brunner]

1.) To be fair, this i think would get messy really quick, i have to think about that.

2.) you can set the name of the file during the dialouge, you can to that as you described. I dont think most user would create a file for each script the run? but i can add a check for that (If debian.conf in /opt/community-scripts/ is present use it by deafult, otherwise use the default one). About automation: we have decided we will not implement this due to security concerns.

[michelroegl-brunner]

number 2 was rather easy, is implemented.

[LilTrublMakr]

About automation: we have decided we will not implement this due to security concerns.

That is actually kind of disappointing to see but understandable from a security standpoint. I was thinking about suggesting this today because it would make some of my tasks easier. Right now, I manually set up a container, add it to Netbox, add info to an Ansible playbook, and add it to Uptime Kuma. It would be very nice if I was able to automate all of this when a container is created. I suppose that this could still be done from a manual run of a script though...

[viperfan7]

I want to add my +1 to post install scripts.

Personally, I use fish+tide for my console, and it would be nice if after a CT/VM is installed, it would run the script I have to install it.

Currently I have a script that takes the container ID and file as parameters, and when ran, it uploads the script, runs it, and delete the script on completion in order to do that, would be nice if that could be implemented into the install script itself.

[Drallas]

I like the idea of having a Config File on my Host, but is the password for accessing the LXC's secured?

[michelroegl-brunner]

As you write this i realize i dont even do the password, i have to add that. But to answer your question: No, this would be stored in plain text on your PVE Host. One option would be to not use the password feature and set it after install, and only use the rest of the options.

[hraphael]

It should be checked if jq is installed or directly install it. The script stopped because jq was not installed

[michelroegl-brunner]

where did it break exactly? wich script did u use?
ther is no jq used in the config file part?

[hraphael]

I have used the debian container.
bash -c "$(wget -qLO - https://raw.githubusercontent.com/community-scripts/ProxmoxVE/config_file/ct/debian.sh)"
Here is the output

    ____       __    _           
   / __ \___  / /_  (_)___ _____ 
  / / / / _ \/ __ \/ / __ `/ __ \
 / /_/ /  __/ /_/ / / /_/ / / / /
/_____/\___/_.___/_/\__,_/_/ /_/ 
                                 
  💡    Using Config File on node pve3
  💡  Using config File: /opt/community-scripts/.settings
  🖥️  Operating System: debian
  🌟  Version: 12
/dev/fd/63: line 760: jq: command not found
Unable to flush stdout: Broken pipe

[ERROR] in line 760: exit code 0: while executing command LIST_OF_IDS=$(pvesh get /cluster/resources --type vm --output-format json | jq '.[] .vmid' -r)

root@pve3:~#

Also I have found jq at line 747 and 760 in the build.func file https://github.com/community-scripts/ProxmoxVE/blob/config_file/misc/build.func

[michelroegl-brunner]

Ah i see, i have overlooked this one. i have rebuild it so it dose not use jq.

[mikecsmith]

Not sure if I've missed something but it would be great to be able to pass a flag to the installation script to trigger an unattended installation via config file. Great example of a use case would be triggering the scripts via Ansible playbooks to programmatically create/update containers.

[michelroegl-brunner]

Due to security concerns we will not provide the option for automatic installs.

[Drallas]

Does it have a safe config option?

[michelroegl-brunner]

No, you have to created the file yourself.

[zephyr325]

This is a great idea - thanks for working on this! Two suggestions I'd like to offer:

1. First, for the DNS server entry, I've got a primary and secondary I usually add in the format of "192.168.1.11,192.168.1.13" (that's what I entered for the "NS=" line in the .settings file). That entry format works when creating an LXC manually. I get the following error:

    ____             __            
   / __ \____  _____/ /_____  _____
  / / / / __ \/ ___/ //_/ _ \/ ___/
 / /_/ / /_/ / /__/ ,< /  __/ /    
/_____/\____/\___/_/|_|\___/_/     
                                   
  💡    Using Config File on node zox1
  💡  Using config File: /opt/community-scripts/.settings
  🖥️  Operating System: debian
  🌟  Version: 12
  🆔  Container ID: 209
  📦  Container Type: Privileged
  🔐  Root Password: ********
  🆔  Container ID: 209
  🏠  Hostname: Crap01.verycarr.com
  💾  Disk Size: 4 GB
  🧠  CPU Cores: 1
  🛠️  RAM Size: 1024 MiB
  🌉  Bridge: vmbr0
  📡  IP Address: 192.168.1.24/24
  🌐  Gateway IP Address: 192.168.1.1
  🚫  Disable IPv6: Yes
  ⚙️  Interface MTU Size: Default
  🔍  DNS Search Domain: <********>
  ✖️  Invalid IP Address format for DNS Server. Needs to be 0.0.0.0, was 192.168.1.11,192.168.1.13

2. Second (and maybe it's a weird use case as this is an automation enhancement) but I always turn on verbose output - I like to see what the script is doing, where it's at in the process, etc. Maybe a config option of VERBOSE="(yes|no)"?

Thanks!

[michelroegl-brunner]

1.) I have a look, i did not consider this tbh and only check if one vaild IP is presented, if not it fails. But i can change the check.

2.) Yeah, this should be possible as well, good idea.

[tremor021]

@michelroegl-brunner can we make the path to .settings file part of .bashrc so if we say export PATH_TO_CUSTOM_SETTINGS= then script doesnt ask where the .setting file is?

[michelroegl-brunner]

We could i guess, but i want to have the question there so a user can have more then one config file if they want to. Mybee with two or more different ranges for IDs or some other settings. Also i do find it unclear for the user wich settings file gets used if we dont disply/ask for it befor the installation.

[tremor021]

The only reason i asked is because of our script testing. i usually set more cores/ram/hdd and apt cacher for the container than what i set for the final script while i'm testing, so having that kind of thing makes testing/installing go a lot faster than doing it the normal way. Making those default custom settings, without me typing or confirming anything is a huge helper

[viperfan7]

We could i guess, but i want to have the question there so a user can have more then one config file if they want to.

To expand on this, why not have it so that the location of the config files is a directory, and it auto selects the config via file/directory name eg.

┳ .settings.d
┣┳ debian.settings.d
┃┠─ <name1>.settings
┃┖─ <name2>.settings
┠─ .settings
┖─ n8n.settings

in this case, .settings would be the default used

if you install something with the name debian.sh, it would look for debian.settings, and if it's a directory, it would ask you which config file you want to use

if you install something like n8n.sh, since it's just a file, it just uses that settings file

Could also have something like global.settings where it gets used on-top of other sub settings, like if you were to import a file into a .bashrc file, but that seems more like something that should be implemented into the settings file itself

A potential alternative:

┳ .settings.d
┣┳ .debian.d
┃┠─ .<name1>
┃┖─ .<name2>
┠─ .global
┖─ .n8n

Same idea, different (Possibly better) naming convention

[viperfan7]

When running n8n.sh with the following config, it gives the error "Gateway Address cannot be empty"

I'm also using an OVS bridge, not the standard linux bridge

VLAN="2"
CT_TYPE="0"

Console Output:

          ____      
   ____  ( __ )____ 
  / __ \/ __  / __ \
 / / / / /_/ / / / /
/_/ /_/\____/_/ /_/ 
                    
  💡    Using Config File on node pve
  💡  Using config File: /opt/community-scripts/.settings
  🖥️  Operating System: debian
  🌟  Version: 12
  🆔  Container ID: 122
  📦  Container Type: Privileged
  🔐  Root Password: Automatic Login
  🆔  Container ID: 122
  🏠  Hostname: n8n
  💾  Disk Size: 6 GB
  🧠  CPU Cores: 2
  🛠️  RAM Size: 2048 MiB
  🌉  Bridge: vmbr0
  📡  IP Address: DHCP
  🌐  Gateway IP Address: Default
  ✖️  Gateway IP Address cannot be empty

Works as expected when using the config file install script but selecting "Advanced Configuration"

[BBArikL]

That is a pretty neat feature. Forgive me because it seems I am blind, I cannot find the config file feature branch anywhere? I checked this repo and the ProxmoxVED one, and I could not find it. Is it already merged?

[michelroegl-brunner]

It is merged in the VED Repo. You can try it with any script there.

[michelroegl-brunner]

Merged with PR #3808

[jwgn]

How to set IPv6 configuration to SLAAC? Standard seems to be DHCP, so I have to manually change this on every deployed container.

[bourkey]

Awesome 👍

Might be an idea to have the menu driven configuration spit out a configuration file ? Build a container once manually, then automate it using this awesome feature.

Only other comment is if you exit after the config file is selected, it tries to build with the default options.

[zer0fade]

I use the SDN feature in Proxmox to define my networks and use that as the net name. This doesn't like that. It's looking at /etc/network/interfaces. Any plans to support that?

---

| | / /___ _____ __ / /
| | /| / / __ `/ / / / / / __
| |/ |/ / // / / // // / / / /
|/|/_,/ //_,// //

💡 Using Config File on node rmv-pve01
💡 Using config File: /opt/community-scripts/.wazuh_config
🖥️ Operating System: debian
🌟 Version: 12
🆔 Container ID: 810
📦 Container Type: Unprivileged
🔐 Root Password: ********
🆔 Container ID: 810
🏠 Hostname: wazuh
💾 Disk Size: 80 GB
🧠 CPU Cores: 8
🛠️ RAM Size: 8192 MiB
✖️ Bridge 'servers' does not exist in /etc/network/interfaces

[michelroegl-brunner]

I dont use the SDN feature. Can you elaborate a bit more on the topic or provide a link so i can have a look at it how we can implement that?

[zer0fade]

Hi there, thanks for the reply. At the Datacenter level I have a SDN Zone that is associated to my Trunk bridge (vmbr1), under that Zone I have my vnet's all named with the ID, so when creating a LXC or VM, for the network bridge I just select the name of the network I had defined in vnets (i.e. Servers, Mgmt, LAN, etc). That way I don't have to select the Bridge and enter the VLAN each time.

[JeGr]

One question and two ideas:

Q1) It's not really clear to me if the ability to have "advanced install" / ask stuff AND the settings file combined is possible right now? I'd really like to pre-define stuff like RootPW, SSH Key, DNS Servers etc. - infrastructure that won't change that often - and still ask about other things interactively.

F1) Not sure if that is already partially doable: but along the lines of @bourkey's idea
If I enter all the things in the advanced install and the installation fails whysoever (it happens) then it would be REALLY nice, if THAT installation would be saved in a .settings style from the last round. My idea would be: for every script you run that has advanced setup (e.g. checkmk) if you run through a complete advanced setup, save the settings under e.g.
/opt/community-scripts/.settings.<name> (e.g. checkmk - like the script name)
so you always have the settings from the last run. And if it goes wrong or you have to quickly redo the container/VM, you can easily run the script again and instead of having to type ALL over again, you just use "the last round" by either entering it or the install script could check for a file named like above already available and could offer you "settings from last run?".

F2) Second feature: when asking to disable IPv6 - how about offering to set it up, too if you say "no"? Would be nice to at least configure SLAAC or even static IP6 that way. Easy to do afterwards as Containers can be configured well over the GUI but for VMs it's more PITA.

I'd also had to check until I found this thread to even have the settings file being documented. Could we have at least a rudimentary docu/wiki page with the variables for the "now" version of it?

Cheers :)

[michelroegl-brunner]

Can you please open a feature request issue so we can better keep track of it? Thank you!