Beiträge von votdev

    Do you think this might be an option, too as replacement for recipes or are you afraid that the complexity will get higher when incorporating Helm?

    Recipes are using Helm charts. I don’t want to create Helm charts, I want to consume them and that is very simple with k3s. Check out the FileBrowser recipe, it is based on a Helm chart. As you can see, the recipe uses only three resources to get this app running behind a reverse proxy with SSL certificate and subpath. The user only needs to file the shared folder path. Voila, that’s near to a one-click installation.


    Recipes are building on Helm charts if possible plus adding everything else that the chart does not provide to give the user a more or less one-click solution.

    Following that, Kubernetes brings nothing new to OMV users, it simply asks to redo what has already been done with Docker.

    IMO Kubernetes is the future and much more powerful than Docker Compose. TrueNAS Scale is based on SUSE K3s for a reason.

    I find the current plugin system too limiting as it seems you can't even add your Javascript.

    This is a unique selling point and a feature of OMV. The UI can be implemented in a declarative way via YAML which is much simpler than doing that in Javascript. This allows users to contribute without having the knowledge of Javascript, Typescript, Angular and aIl the other things that would be necessary. IMO the UI already provides everything that is necessary for this type of UI.

    The watchdog package was a hard dependency long time ago. I'm sorry i do not remember everything what i've done in the last 15 years, but it seems that it is a soft dependency now. Nevertheless, with the commands mentioned above you can configure OMV so that the watchdog config is purged.


    But the core problem persists, the Raspi kernel does not ship the required modules to get the watchdog working.

    you wrote "This is a useful feature for headless servers. But it is not a service that is really required to run the NAS." Is this still valid for V7 and RPI? If so, I do not really understand the benefit of disabling watchdog.
    Please excuse if I misunderstand something, my knowledge is a bit low, but I'm trying to understand what is going on.

    Using a watchdog makes absolute sense on a headless NAS, but it seems Raspi OS does not contain the necessary kernel modules. So it is better to disable the OMV feature.


    Maybe it makes sense to raise a ticket for the Raspi OS that the project is aware of the problem.

    I think you shoudl raise the issue in the Raspi forum because OMV is using their OS but does not maintain it. If there is a problem, only the Raspi project can fix it. To me it looks like their kernel is missing some modules.


    You need to disable watchdog the OMV way, so the softdog module will not be configured. Set the OMV_WATCHDOG_ENABLED env var to false according to the docs. Instead of deploying everything as mentioned in the docs simply run

    Bash
    omv-salt stage run prepare
    omv-salt deploy run watchdog

    I think storage management in OMV is good, but there is room for getting better. Keep in mind that OMV is done by myself and omv-extras.org by Aaron and other contributors. It is unfair to compare it with companies that have employees that work on it full-time. We do it in our spare time as a hobby.


    OMV7 has now a Kubernetes plugin to run containers in a K8s environment without the need of Docker or Podman. I've also started a repository for it to collect recipes to easily install container apps from Helm Charts or K8s API resources. Standalone plugins require much much much time and maintenance. Therefore container based apps are much easier to maintain. The sad thing is that they need to be managed via Yaml and are not integrated very well in the UI. But i'm thinking about how to provide a UI for that because the whole OMV UI is done in a declarative manner, so it should be easy to provide that for container based plugins as well.


    To populate and increase this Kubernetes recipes it requires active participation of the community. When it comes to plugins/apps, i think this is the way to go because it reduces the implementation and maintenance time of apps/plugins.


    Users have to choose then between Kubernetes and Docker. I'm in favor of K8s because it is much easier when it comes to network config. I hate the Docker approach. But thats my personal point of view.


    Finally, a ecosystem like OMV can only increase when the community is contributing. This is something that is sadly going lim x->0.

    Or do I need to set a log size limit on my own?

    Yes



    Why are there still old fashioned log entries written? Why is there still a syslog file when syslogd journal is active? Is this a bug on my system?

    rsyslog is still installed because of features that are not available and possible with systemd. But afaik there shouldn't be written anything to the filesystem. This only applies to a fresh OMV7/Debian12, if it is migrated from OMV6/Debian11, then the rsyslog config is not touched and everything is duplicated.