How do I get a tarball into a Container?

[Bauerranger]

I have a piece of software that I want to run inside a proxmox container. This software is not available on github, but on a gitlab that requests special authorisation through a mobile app (not necessary for you to know, but I want to install the website evidence collector form the EU's gitlab). Therefore I cannot use the git commands or curl to get the tarball into the container. And I am not sure how to do it. My brain wants to tell me that there must be a protocoll like ftp to easily send files and folders to my server inside my network but so far searching the internet has not given me the answers I seek for.

Possible workarounds are: setting up a vm instead of a container and navigate to the website and authenticate with the mobile app, all via the vnc. However, I want to learn the ins and outs of proxmox virtualisation and would like to be able to get this done in a cotainer and not an unnecessarily big vm. Another workaround would be to set up a container, navigate to the download page, copy the link of the download button and curl that url from the container. But that is not what I would like, as I would have to do that every time I want to update the software.

I hope the forums can help me out. As I am still very new to proxmox, not that experienced in virtualisations and am far from mastering all things git I would also be very grateful if anyone could tell me why my idea may not be the best approach and why other solutions would be preferable. I don't know what these other possibilities would be, so please don't hesitate to enlight me!

Thanks!

 

[dherzig]

Hi Bauerranger,

what kind of container are you using? Does it have internet access -- here from a default debian12 LXC container I can do:

wget https://code.europa.eu/EDPS/website-evidence-collector/-/releases/v3.0.0/downloads/website-evidence-collector.tgz

I do not run into any authentication queries from the gitlab's side. The link is from here: https://code.europa.eu/EDPS/website-evidence-collector/-/releases.

To actually install the software (and follow the recommendation to use node > v20), you could do:

$ wget -O setup_node23.sh https://deb.nodesource.com/setup_23.x
$ bash setup_node23.sh
$ apt install nodejs
$ node -v
v23.6.1
# install the previously downloaded tarball (or use the command from the gitlab website)
$ npm install -g website-evidence-collector.tgz

I have to admit though, that the software fails at launching a browser process, but that's not so much of a PVE thing, I'd say...

Best regards,
Daniel

 

[dherzig]

I have to add, that the software seems to depend on a graphical desktop environment. I'm saying this as the `website-evidence-collector serve` command does not seem to allow for adding anything else than localhost as a server location. So reaching the server process from another machine is likely to be challenging and would involve port forwarding, eg via ssh.

One could go the way to set up a lightweight graphical desktop environment in the container (Mate, LXDE,...), install xrdp, add an unpriviledged user, and connect to the Container with an rdp client.

Or, quite possibly the easier way -- just use a VM with a Desktop environment and look at the software straight from your PVE host through the browser!

I hope this information helps,
Best regards,
Daniel

 

[Bauerranger]

Thanks for your assistance and valuable input! I think the software requires some sort of graphical interface as it depends on chromium and calls websites through that.

You are right that using a vm might be an easy way to do it, but here is what I want to do with that software: ideally, I would like to be able to call the container from inside another software/container, like a dokuwiki, i.e. send an http request that contains the url of the website that needs to be checked and receive the resulting pdf as the response. I want to automatise that. I don't want to have to log into a vm, run all the commands manually and then having to copy&paste the resulting pdf onto my machine.

The official repo of the website evidence collector also contains a docker image that I could use. However it feels "odd" to me to run a docker container inside a proxmox container. Does that come with tradeoffs like bad performance or is this not good practice? What would your thoughts be on that?

 

[Johannes S]

The official repo of the website evidence collector also contains a docker image that I could use. However it feels "odd" to me to run a docker container inside a proxmox container. Does that come with tradeoffs like bad performance or is this not good practice? What would your thoughts be on that?

It's known to break often after updates and thus not recommended by the ProxmoxVE developers. I would setup the software in a docker on a vm or try to install it without docker in this lxc:


https://git.osit.cc/public-projects/ubuntu-lxc-terminalserver-project

 

[dherzig]

I see two different topics here -- correct me, if I'm wrong!

The one is how to conveniently update the software -- I'm honestly not really sure what is more convenient. Be it changing the image version in a corresponding docker-compose.yaml file in a VM with docker installed, or running something like 'npm uninstall --global $SOFTWARE && npm install --global $NEW_TARBALL_OF_SOFTWARE && $COMMAND_THAT_YOU_NEED'. Both involve manual intervention! On PVE you can however reduce implications of an failed update by making snapshots of the VM/CT before changing it. If something goes wrong, you just rollback.

The other -- the automation of the url-check could be done with other techniques, but this all has very little to do with PVE. I have just very superficially looked at the software, but at first glance it does not seem to offer an API, that could help on doing what you describe. You could e.g. try to write a script that can take arguments (the URL) and try to remotely run it via eg ssh, where the output will be written to a shared storage, that you can access from another machine. Or have a script that runs in certain time intervals and checks a file or URL for new URLs to check!

Best,
Daniel