Process

The process collector can monitor various running processes in the system, acquire and analyze various metrics when the process is running, Including memory utilization rate, CPU time occupied, current state of the process, port of process monitoring, etc. According to various index information of process running, users can configure relevant alarms in Guance, so that users can know the state of the process, and maintain the failed process in time when the process fails.

Attention

Process collectors (whether objects or metrics) may consume a lot on macOS, causing CPU to soar, so you can turn them off manually. At present, the default collector still turns on the process object collector (it runs once every 5min by default).

Configuration¶

Preconditions¶

The process collector does not collect process metrics by default. To collect metrics-related data, set open_metric to true in host_processes.conf. For example:

[[inputs.host_processes]]
    ...
     open_metric = true

Collector Configuration¶

Host InstallationKubernetes

Go to the conf.d/host directory under the DataKit installation directory, copy host_processes.conf.sample and name it host_processes.conf. Examples are as follows:

[[inputs.host_processes]]
  # Only collect these matched process' metrics. For process objects
  # these white list not applied. Process name support regexp.
  # process_name = [".*nginx.*", ".*mysql.*"]

  # Process minimal run time(default 10m)
  # If process running time less than the setting, we ignore it(both for metric and object)
  min_run_time = "10m"

  ## Enable process metric collecting
  open_metric = false

  ## Enable listen ports tag, default is false
  enable_listen_ports = false

  ## Enable open files field, default is false
  enable_open_files = false

  ## only collect container-based process(object and metric)
  only_container_processes = false

  # Extra tags
  [inputs.host_processes.tags]
  # some_tag = "some_value"
  # more_tag = "some_other_value"
  # ...

Once configured, restart DataKit.

Can be turned on by ConfigMap Injection Collector Configuration or Config ENV_DATAKIT_INPUTS .

Can also be turned on by environment variables, (needs to be added as the default collector in ENV_DEFAULT_ENABLED_INPUTS):

ENV_INPUT_HOST_PROCESSES_OPEN_METRIC

Enable process metric collecting

Type: Boolean

input.conf: open_metric

Default: false
ENV_INPUT_HOST_PROCESSES_PROCESS_NAME

Whitelist of process

Type: List

input.conf: process_name

Example: .*datakit.*,guance
ENV_INPUT_HOST_PROCESSES_MIN_RUN_TIME

Process minimal run time

Type: Duration

input.conf: min_run_time

Default: 10m
ENV_INPUT_HOST_PROCESSES_ENABLE_LISTEN_PORTS

Enable listen ports tag

Type: Boolean

input.conf: enable_listen_ports

Default: false
ENV_INPUT_HOST_PROCESSES_TAGS

Customize tags. If there is a tag with the same name in the configuration file, it will be overwritten

Type: Map

input.conf: tags

Example: tag1=value1,tag2=value2
ENV_INPUT_HOST_PROCESSES_ONLY_CONTAINER_PROCESSES

Only collect container process for metric and object

Type: Boolean

input.conf: only_container_processes

Default: false
ENV_INPUT_HOST_PROCESSES_METRIC_INTERVAL

Collect interval on metric

Type: Duration

input.conf: metric_interval

Default: 30s
ENV_INPUT_HOST_PROCESSES_object_interval

Collect interval on object

Type: Duration

input.conf: object_interval

Default: 300s

Metric¶

For all of the following data collections, a global tag named host is appended by default (the tag value is the host name of the DataKit), or other tags can be specified in the configuration by [inputs.host_processes.tags]:

 [inputs.host_processes.tags]
  # some_tag = "some_value"
  # more_tag = "some_other_value"
  # ...

`host_processes`¶

Collect process metrics, including CPU/memory usage, etc.

Tags

Tag	Description
`container_id`	Container ID of the process, only supported Linux
`host`	Host name
`pid`	Process ID
`process_name`	Process name
`username`	Username

Metrics

Metric	Description	Type	Unit
`cpu_usage`	CPU usage, the percentage of CPU occupied by the process since it was started. This value will be more stable (different from the instantaneous percentage of `top`)	float	percent
`cpu_usage_top`	CPU usage, the average CPU usage of the process within a collection cycle	float	percent
`mem_used_percent`	Memory usage percentage	float	percent
`nonvoluntary_ctxt_switches`	From /proc/[PID]/status. Context switches that nonvoluntary drop the CPU. Linux only	int	count
`open_files`	Number of open files (only supports Linux)	int	count
`page_children_major_faults`	Linux from /proc/[PID]/stat. The number of major page faults for this process. Linux only	int	B
`page_children_minor_faults`	Linux from /proc/[PID]/stat. The number of minor page faults for this process. Linux only	int	B
`page_major_faults`	Linux from /proc/[PID]/stat. The number of major page faults. Linux only	int	B
`page_minor_faults`	Linux from /proc/[PID]/stat. The number of minor page faults. Linux only	int	B
`proc_read_bytes`	Linux from /proc/[PID]/io, Windows from `GetProcessIoCounters()`. Read bytes from disk	int	B
`proc_syscr`	Linux from /proc/[PID]/io, Windows from `GetProcessIoCounters()`. Count of `read()` like syscall`. Linux&Windows only	int	count
`proc_syscw`	Linux from /proc/[PID]/io, Windows from `GetProcessIoCounters()`. Count of `write()` like syscall`. Linux&Windows only	int	count
`proc_write_bytes`	Linux from /proc/[PID]/io, Windows from `GetProcessIoCounters()`. Written bytes to disk	int	B
`rss`	Resident Set Size (resident memory size)	int	B
`threads`	Total number of threads	int	count
`voluntary_ctxt_switches`	From /proc/[PID]/status. Context switches that voluntary drop the CPU, such as `sleep()/read()/sched_yield()`. Linux only	int	count

Object¶

`host_processes`¶

Collect data on process objects, including process names, process commands, etc.

Tags

Tag	Description
`container_id`	Container ID of the process, only supported Linux
`host`	Host name
`name`	Name field, consisting of `[host-name]_[pid]`
`process_name`	Process name
`state`	Process status, currently not supported on Windows
`username`	Username

Metrics

Metric	Description	Type	Unit
`cmdline`	Command line parameters for the process	string	-
`cpu_usage`	CPU usage, the percentage of CPU occupied by the process since it was started. This value will be more stable (different from the instantaneous percentage of `top`)	float	percent
`cpu_usage_top`	CPU usage, the average CPU usage of the process within a collection cycle	float	percent
`listen_ports`	The port the process is listening on	string	-
`mem_used_percent`	Memory usage percentage	float	percent
`message`	Process details	string	-
`nonvoluntary_ctxt_switches`	From /proc/[PID]/status. Context switches that nonvoluntary drop the CPU. Linux only	int	count
`open_files`	Number of open files (only supports Linux, and the `enable_open_files` option needs to be turned on)	int	count
`page_children_major_faults`	Linux from /proc/[PID]/stat. The number of major page faults of it's child processes. Linux only	int	B
`page_children_minor_faults`	Linux from /proc/[PID]/stat. The number of minor page faults of it's child processes. Linux only	int	B
`page_major_faults`	Linux from /proc/[PID]/stat. The number of major page faults. Linux only	int	B
`page_minor_faults`	Linux from /proc/[PID]/stat. The number of minor page faults. Linux only	int	B
`pid`	Process ID	int	-
`proc_read_bytes`	Linux from /proc/[PID]/io, Windows from `GetProcessIoCounters()`. Read bytes from disk	int	B
`proc_syscr`	Linux from /proc/[PID]/io, Windows from `GetProcessIoCounters()`. Count of `read()` like syscall`. Linux&Windows only	int	count
`proc_syscw`	Linux from /proc/[PID]/io, Windows from `GetProcessIoCounters()`. Count of `write()` like syscall`. Linux&Windows only	int	count
`proc_write_bytes`	Linux from /proc/[PID]/io, Windows from `GetProcessIoCounters()`. Written bytes to disk	int	B
`rss`	Resident Set Size (resident memory size)	int	B
`start_time`	process start time	int	msec
`started_duration`	Process startup time	int	sec
`state_zombie`	Whether it is a zombie process	bool	-
`threads`	Total number of threads	int	count
`voluntary_ctxt_switches`	From /proc/[PID]/status. Context switches that voluntary drop the CPU, such as `sleep()/read()/sched_yield()`. Linux only	int	count
`work_directory`	Working directory (Linux only)	string	-

Process

Configuration¶

Preconditions¶

Collector Configuration¶

Metric¶

host_processes¶

Object¶

host_processes¶

Is this page helpful? ×

`host_processes`¶

`host_processes`¶