Process Monitoring Metrics
Sematext Agent gathers various metrics and metadata about running processes and brings them to Sematext Cloud so you can painlessly explore the system resources allocated by processes whether they're executing on bare metal machines, VMs, plain containers or scheduled to run on multi-cluster Kubernetes workloads.
To tweak the frequency of the process' data series collection and delivery to Sematext Cloud adjust the following options:
# st-agent.yml
process:
interval: 10s
PROCESS_INTERVAL=10s
Metrics¶
The agent collects top N processes by CPU and memory utilization. Once process is elected as a top resource consumer, it stays in the top N list for the time slice that has been allocated to it. It is relinquished from the top N list when there is another process with higher resource usage that displaces the current process or it leaves the top N list voluntarily (e.g. no longer detected as a relevant resource consumer).
The table below includes all available metrics.
| Metric name | Description |
|---|---|
| resident memory | Represents the resident memory allocated by the process. It includes the memory allocated for heap and stack areas |
| CPU usage | Designates the process' percentage of CPU utilization |
| fd count | Is the number of file descriptors allocated by the process |
| thread count | Represents the number of threads running in the process' address space |
| disk read bytes | Is the number of bytes that the process fetched from the storage layer |
| disk write bytes | Is the number of bytes which the process sent to the storage layer |
| read ops | Is the number of read operations (syscalls, e.g. read or pread) issued by the process |
| write ops | Is the number of write operations (syscalls, e.g. write or pwrite) issued by the process |