Containerization?
vsoch opened this issue · comments
Hiya! I have a possibly dumb question. I've built this into a Docker container (and it seems to work) and I'm wondering what interfaces are used to derive the metrics/data? E.g., I'm wondering if I run this in Kubernetes in a sidecar pod, if the metrics would reflect the entire node the pod is on (or just the pod in question). Thanks!
The samplers will all use the same interfaces that they normally use to retrieve data, and whether or not those interfaces are available inside the pod at all, or if the information changes between the pod or the host OS will depend entirely on the specific sampler. Different samplers go about retrieving information in different ways.
Is there documentation somewhere that describes the interfaces used by each sampler?
The man page typically specifies which file/source from which it derives its data. Most of the /proc samplers use open/read. The ibmad, slingshot, GPU, ... samplers use whatever interfaces are required to query that data. In most cases, this is not called out in the man pages.
Gotcha - thanks @tom95858! One more question - do you have an example config file that will output a json structure (as opposed to the tabular / terminal output I see by default)?
@vsoch are you tacking about ldms_ls output?
Yes - and I think I know the answer to that (no) so happy to close the issue. Thanks for the help @tom95858 !