Hello,
I’ve got a question regarding Kubernetes and the hardware it is running on. To my understanding, Kubernetes is not, by default, concerned about the hardware it deploys its nodes on. But is there a “standard” method of retrieving basic information that the the underlying hardware could provide via Kubernetes? Let’s say I deploy a Node on a Pi and I would want to monitor something about it, let’s say the CPU temperature.
Does Kubernetes allow for such a transfer as data in its basic form or would I need a second kind of infrastructure to receive such data? From my perspective, all the devices are already connected via Kubernetes in this way, so if I have a central point to orchestrate things from the software side, would it make sense to concentrate hardware information via this same channel as well? Or is it a bad idea to combine the two?
Run a daemonset that scrapes hardware stats and stuffs them into a
TSDB or something? I don’t know if the kubernetes API is what you
want for hardware alerting, as you almost certainly want history.
Yes, your understanding is correct. By default Kubernetes does not bother about hardware and we can deploy additional monitoring tools such as node-problem-detector. Of Course you can use the same channel as long as the hardware specs supports the additional overhead. i.e., by deploying additional tools to monitor nodes health it will consume some memory/cpu. If your infrastructure designed to handle this then it is ok I believe.