Revive 4.1.4 on Kubernetes

[Hans]

Hi,

I am trying to install 4.1.4 on my k8s cluster using Chrome with plugins and extensions disabled in case some interfere with the installation process (google-chrome --disable-plugins --disable-extensions). I've also tried Chromium and Firefox. After specificying the admin user, password and timzeone (default) on the third page of the installation wizzard I am almost instantly redirected to the login page. Neihter the user is created nor are any plugins installed.

This is the Docker image I am trying to run https://github.com/hcvst/ReviveDockerImage (forked from this 4.1.3 revive version, which also does not work https://github.com/krish512/ReviveDockerImage)

The debug and install logs do not show anything suspicious.

I was once able to install revive 4.1.3 through port forwarding and running the wizzard via localhost against the same k8s mysql DB but even that I can no longer reproduce.

Under what circumstances does the wizzard redirect to the login page without completing the installation, please? Is there a way to get more meaningful debug info?

Regards,
Hans

 

 

[Hans]

Hi,

I suspect it might have something to do with the HTTPs ingress.

I spun up a temporary k8s pod now and, as mentioned in my original post, used port forwarding to install Revive via http://localhost. That worked.

Once the user and plugins were successfully installed I updated the .conf file of the failed deployment to use the database of the succesful one. 

Kind of ugly but better than nothing for now. Please see the steps I took below. If you figure out how to do it properly I am keen to find out.

Regards,
Hans

 

- Spin up a temp. pod: kubectl run -ti myrevive --image=hcvst/ReviveDockerImage
- Forward port: kubectl port-forward myrevive-... 3080:80
- Go to installation wizzard at: http://localhost:3080
- For the DB host specify the service name of a mysql DB in your k8s cluster
- Once installation finishes destroy the temp. pod deployment
- Edit installdir/var/your.domain.conf of the original failed deployemnt to point at the database of the successful one.
- Drop the databse of the original failed deployment.

 

[Krishna Modi]

Try the latest version for https://hub.docker.com/r/krish512/revive/

Alpine changes caused the image to break. It is currently updated to Revive AdServer 4.1.4 and will be actively maintained further.

[Hans]

Thanks. I forgot to post here, as I had planned, that revive was not to blame at all but rather me being a k8s newbie. I don't remember the details but the problem essentially was that I had used Helm to create a chart which had liveness probe enabled. As a result it kept recreating the pod halfway through the installation process.