System
System 采集器收集系统负载、正常运行时间、CPU 核心数量以及登录的用户数。
配置¶
成功安装 DataKit 并启动后，会默认开启 System 采集器，无需手动开启。
采集器配置¶
进入 DataKit 安装目录下的 conf.d/host 目录，复制 system.conf.sample 并命名为 system.conf。示例如下：
[[inputs.system]]
##(optional) collect interval, default is 10 seconds
interval = '10s'
[inputs.system.tags]
# some_tag = "some_value"
# more_tag = "some_other_value"
配置好后，重启 DataKit 即可。
可通过 ConfigMap 方式注入采集器配置 或 配置 ENV_DATAKIT_INPUTS 开启采集器。
也支持以环境变量的方式修改配置参数（需要在 ENV_DEFAULT_ENABLED_INPUTS 中加为默认采集器）：
-
ENV_INPUT_SYSTEM_INTERVAL
采集器重复间隔时长
字段类型: Duration
采集器配置字段:
interval默认值: 10s
-
ENV_INPUT_SYSTEM_TAGS
自定义标签。如果配置文件有同名标签，将会覆盖它
字段类型: Map
采集器配置字段:
tags示例: tag1=value1,tag2=value2
指标¶
以下所有数据采集，默认会追加名为 host 的全局 tag（tag 值为 DataKit 所在主机名），也可以在配置中通过 [inputs.system.tags] 指定其它标签：
system¶
Basic information about system operation.
- 标签
| Tag | Description |
|---|---|
host |
hostname |
- 指标列表
| Metric | Description | Type | Unit |
|---|---|---|---|
cpu_total_usage |
The percentage of used CPU. | float | percent |
load1 |
CPU load average over the past 1 minute. | float | - |
load15 |
CPU load average over the past 15 minutes. | float | - |
load15_per_core |
CPU single core load average over the past 15 minutes. | float | - |
load1_per_core |
CPU single core load average over the past 1 minute. | float | - |
load5 |
CPU load average over the past 5 minutes. | float | - |
load5_per_core |
CPU single core load average over the last 5 minutes. | float | - |
memory_usage |
The percentage of used memory. | float | percent |
n_cpus |
CPU logical core count. | int | count |
n_users |
User number. | int | count |
process_count |
Number of Processes running on the machine. | int | - |
uptime |
System uptime. | int | s |
conntrack¶
Connection track metrics (Linux only).
- 标签
| Tag | Description |
|---|---|
host |
hostname |
- 指标列表
| Metric | Description | Type | Unit |
|---|---|---|---|
entries |
Current number of connections. | int | count |
entries_limit |
The size of the connection tracking table. | int | count |
stat_drop |
The number of packets dropped due to connection tracking failure. | int | count |
stat_early_drop |
The number of partially tracked packet entries dropped due to connection tracking table full. | int | count |
stat_found |
The number of successful search entries. | int | count |
stat_ignore |
The number of reports that have been tracked. | int | count |
stat_insert |
The number of packets inserted. | int | count |
stat_insert_failed |
The number of packages that failed to insert. | int | count |
stat_invalid |
The number of packets that cannot be tracked. | int | count |
stat_search_restart |
The number of connection tracking table query restarts due to hash table size modification. | int | count |
filefd¶
System file handle metrics (Linux only).
- 标签
| Tag | Description |
|---|---|
host |
hostname |
- 指标列表
| Metric | Description | Type | Unit |
|---|---|---|---|
allocated |
The number of allocated file handles. | int | count |
maximum_mega |
The maximum number of file handles, unit M(10^6). | float | count |
FAQ¶
为什么没有 cpu_total_usage 指标？¶
CPU 部分采集功能不支持部分平台，如 macOS。