Our kube api server in master node has one problem recently, its memory keeps growing and will cause OOM kill, now I am looking into this issue and using go pprof tool to check the
memory status, but one thing that confused me is that the inuse_space memory is far less than the memory that kube api server actually used:
it shows 9124.54MB memory is in using, right? but from top command in master node and from grafana monitoring, shows the kube api server used nearly 24G.
grafana:
top cmd:
idk why there is this difference, Do I misunderstand the inuse_space in pprof profliing?
