Data for the kepler_container_bpf_cpu_time_ms_total metric is zero

[andetlei]

Hi

Sorry, I am a bit new to this, but what has happened to this item?
[https://github.com/sustainable-computing-io/kepler-model-server/discussions/109]
Has it become an issue?

The reason for asking is that I see something very similar, almost all metrics provided by Kepler is zero:

For example:

I have peeked into the eBPF maps and here I find a reading like this:

},{
"key": 1816415,
"value": {
"cgroup_id": 574811,
"pid": 1816415,
"process_run_time": 2329528,
"cpu_cycles": 0,
"cpu_instr": 0,
"cache_miss": 0,
"page_cache_hit": 0,
"vec_nr": [0,21,0,0,0,0,0,195,0,179
],
"comm": "cpu-stress"
}
},{

I guess this must be the reading from the process in pod I have applied to load the system.

Still the output from Kepler for this one is zero:

kepler_container_bpf_cpu_time_ms_total{container_id="11efba3342e67c3232d9c74e27a9cd380794f94750eed87c614a01509c1f28d3",container_name="cpu-stressor",container_namespace="monitoring",pod_name="cpu-stressor-deployment-6dfd7758d-bfjrc",source="bpf"} 0

To me it seems like the mapping between the eBPF readings and the pods is failing?

I am using a Kind cluster inside WSL2 on laptop.

BR