Incorrectly reporting 100% CPU on Slackware (UnRAID) #3
Description
Platform: Slackware Linux (UnRAID OS)
Agent version: v0.1.0
I am running the agent as a native binary (not Docker). The load1_percent and load15_percent fields in /api/sysinfo/all always report 100 on UnRAID OS. I assume that this is a compatibility issue between the agent and Slackware. The incorrect reading persists across multiple polls and doesn't seem to settle over time. RAM and disk metrics report correctly. I am able to successfully get accurate metrics on my Proxmox host and Docker VM.
UnRAID has 18 days uptime and has been overwhelmingly idle. Evidence for this (that the host is not actually at 100% CPU):
/proc/stat output:
cpu 3714905 0 9979231 1250190711 27896254 0 832901 0 0 0
idle ticks: 1,250,190,711
iowait ticks: 27,896,254
non-idle ticks: ~14.5 million
UnRAID's dashboard confirm about 2% CPU usage:
API response:
{
"hostname": "UnRAID",
"platform": "slackware",
"cpu": {
"load_is_available": true,
"load1_percent": 100,
"load15_percent": 100
}
}
I tried running as a docker container inside UnRAID with the same results: incorrect 100% CPU reporting. Also, I couldn't get it to see outside of the docker container (even with --pid=host) so it was always (again, incorrectly) reporting 100% CPU:
{
"hostname": "39453a1690d1",
"platform": "alpine",
"cpu": {
"load_is_available": true,
"load1_percent": 100,
"load15_percent": 100
}
}