July is marked by us starting the Alpha Test. Email us if you want to help us test!
Furthermore, this month we added a new team member, we wrote a document about auto update mechanisms, and we improved our Nextcloud setup.
We approximate testing OpenAppStack will take 1 to 2 hours of your time. The test will include testing:
- The setup process. We have been testing with our own configurations and exclusively on Greenhost VPSs. If you have some other VPS lying around on which you can run Debian Buster and which you can dedicate to a test run, that would be really helpful for us! Alternatively we can provide you with a test VPS.
- The Nextcloud/ONLYOFFICE setup and integration. We can not guarantee that Nextcloud will stay up to date and that you won’t lose data during this test. To get there, though, we would really appreciate it if you could setup OpenAppStack and try to use the NextCloud.
If you contact us, we will send you more elaborate test instructions.
New team member
Starting this month, we have a new team member who is responsible for setting up OpenAppStack’s single sign-on and user management.
In the past month, most of his focus was on designing how exactly the sign in process of OpenAppStack should look: which components there are and how they should be connected. We now have a crude, but working, prototype of the proposed setup.
Auto update mechanisms
Because OpenAppStack wants to be able to automatically update the applications that are running on it, we need some system that checks for updates and applies them. At the same time, you want to be sure that those updates do not break anything.
As a first step towards this goal, we have investigated the different solutions, and wrote an article about it. Click this link to read it.
In short, we have made the decision to try using Flux, because of the following features:
- It seems to be well documented
- It updates Docker containers as well as Kubernetes components and Helm charts
- It can run as part of OpenAppStack (it is “pull based”)
- It is installed with our application management system, Helm
- It seems possible to roll a broken system back to an older version
- It exposes an endpoint to our monitoring system, Prometheus
Nextcloud was updated to the newest Helm chart version. Additionally we fixed a few warnings that showed up in the Nextcloud administration screen. Lastly, we made sure that regular jobs that are required for Nextcloud now run in Kubernetes.