Hello,
empowering developers to manage and deploy their software themselves is a cornerstone of the DevOps movement. In highly dynamic environments powered by Kubernetes it is especially important to provide a way for development teams to define the monitored endpoints themselves. At work, we currently ask developers to send a pull request for the Xymon config when something changes. This works well enough, but forces a strong coupling between Xymon configuration and application deployment, which is not represented in the resulting files!
The Kubernetes orchestrator can be extended using Custom Resource Definitions 1, even in running clusters. Using this facility, cluster users can create/read/update/delete resources just as they are used to for normal workloads.
xycrd bridges the gap between Xymon and Kubernetes by watching custom resources, generating a Xymon hosts.cfg snippet and sending a SIGHUP to Xymon as soon as the resources are changed.
Please be aware that this is an early release, not suited for production use without extensive testing!
See the code here (Debian package available on the latest release): <https://github.com/thriqon/xycrd>
Future roadmap: Xycrd does not cleanup monitoring data for deleted endpoints. This creates waste that should be cleaned up. In the future, xycrd should learn to tell Xymon to drop data prior to deleting the entry from the hosts.cfg snippet. Additionally, when nearing v1, the group name (xycrd.thriqon.github.io) will be shortened considerably; still looking for a fitting domain though.
If you want something changed, PRs/Issues are open, but I'm happy to answer each and every question on this ML as well.
Best regards, Jonas