Offline deployment tutorial for openvidu3.4.1 single node version

Issues with deployments
1

Can you provide an offline deployment tutorial for the OpenVidu 3.4.1 single-node version in a Linux environment? It’s urgent!!!

2

Hello @wangsen0910 , what deployment do you want? Are you using CE

Right now it is not possible, but it is in our road map to do it. Not for the next release though… If you tell me which deployment you want and if you are using CE or PRO, I can try to guide you.

Also, what problem are you experiencing?

Best Regards.

3

Hello, thank you for your reply. I will elaborate on my thoughts in detail.

First of all, my requirements are as follows:
1.Deploy the single-node CE version of OpenVidu 3.4.1 on a public intranet server (LAN-only environment).
2.Conduct secondary development and testing on any local machine, including client-side and server-side development work.
3.Support access and testing from multiple devices.
4.The final deployment of the actual production environment will be as follows:
① The server-side will be integrated into the aforementioned intranet server along with OpenVidu 3.4.1;
② The client-side will be deployed on another intranet server.

Naturally, all the aforementioned intranet environments are interconnected—simply put, this is a LAN-only development, testing, and deployment setup.

Based on the above requirements, I performed the following split operations in a pure public network environment following the single-node deployment steps:
1.First, download the install.sh script from sh <(curl -fsSL http://get.openvidu.io/community/singlenode/latest/install.sh) to the local machine.
2.Then, uniformly download and package all images referenced in install.sh into a file (e.g., openvidu-single.tar).
3.Next, I commented out all image-pulling operations in the install.sh script, then copied both the modified script and openvidu-single.tar to the intranet server for deployment.
4.Subsequent deployment operations on the intranet were identical to those in the public network, but the deployment ultimately failed. The observed phenomena are:
① Both guided and non-guided installations can be initiated successfully.
② However, after configuring the basic services, when prompted to continue with the installation and I select “Yes”,
the console still displays error messages related to image pulling, and the process cannot proceed further.

Finally, my questions are:
Does OpenVidu support the scenario I described?
If it is supported, is there any mistake in my attempted deployment method? If so, I would appreciate your assistance.
If it is not supported, I may have to consider alternative solutions.

4

It is not possible right now, but now that I see it could be a problem I will allow a export PULL_IMAGES=false env variable in a future release…

For the moment, let’s try to workaround the problem:

  1. As you did uniformly download and package all images referenced in install.sh into a file (e.g., openvidu-single.tar)

  2. Create a temporary directory and cd into that directory

    mkdir -p /tmp/openvidu
cd /tmp/openvidu

  3. Download the installation script: http://get.openvidu.io/community/singlenode/3.4.1/install.sh and place it at /tmp/openvidu/install.sh via scp

  4. Create the following scirpt script in this directory /tmp/openvidu/patch_install.sh

    #!/usr/bin/env bash
set -euo pipefail

WORKDIR="${1:-$PWD}"
TARGET="${WORKDIR}/install.sh"
MARKER="# PATCH_APPLIED"

# Check if the patch is already applied
if grep -q "${MARKER}" "${TARGET}"; then
    echo "Patch already applied. Exiting."
    exit 0
fi

# Remove docker pull
sed -i '/docker pull "\${INSTALLER_IMAGE}"/d' "${TARGET}"
# Remove install.sh to modify it before installing
sed -i '/^\s*\.\/install\.sh/,$d' "${TARGET}"

# Modify generated install
echo "sed -i '/RUNNING_SYSTEMD=false docker-compose --env-file .* pull/d' ./install.sh" >> "${TARGET}"
echo './install.sh' >> "${TARGET}"
echo 'cat finish-message.txt' >> "${TARGET}"
echo 'echo' >> "${TARGET}"

# Add marker to indicate the patch has been applied
echo "${MARKER}" >> "${TARGET}"

    The script is going to modify the installation script to not pull images.

  5. Give both scripts permissions:

    chmod +x patch_install.sh install.sh

  6. The directory /tmp/openvidu should look like this
    image

  7. Execute the patch:

    ./patch_install.sh

  8. Execute the installation script

    ./install.sh

@wangsen0910 let me know if this worked for you.

5

Thank you very much for your help. Currently, this step can be successfully bypassed and the service can start normally
But there should still be some issues that cannot be accessed normally on the local development machine. It may be due to some problems with the domain name and certificate. I will try to solve them myself first. If it cannot be solved, I will ask for your help again. Thank you!
The development model of completely separating internal and external networks is too common for us, so I have to try to overcome all these difficulties, haha！

1 Like
6

No problem! Let me know if you continue having issues.