Kubernetes运维生态
Kubernetes的生态中，cAdvisor是作为容器监控数据采集的Agent，其部署在每个节点上，内部代码结构大致如下：代码结构很良好，collector和storage部分基本可做到增量扩展开发。https://images2015.cnblogs.com/blog/532603/201702/532603-20170226132430913-1046725660.png
关于cAdvisor支持自定义指标方式能力，其自身是通过容器部署的时候设置lable标签项：io.cadvisor.metric.开头的lable，而value则为自定义指标的配置文件，形如下：
{"endpoint" : {"protocol": "https","port": 8000,"path": "/nginx_status" },
"metrics_config": [ {
"name" : "activeConnections","metric_type" : "gauge","units" : "number of active connections","data_type" : "int","polling_frequency" : 10,"regex" : "Active connections: (+)" },
{
"name" : "reading","metric_type" : "gauge","units" : "number of reading connections","data_type" : "int","polling_frequency" : 10,"regex" : "Reading: (+) .*" },
{
"name" : "writing","metric_type" : "gauge","data_type" : "int","units" : "number of writing connections","polling_frequency" : 10,"regex" : ".*Writing: (+).*" },
{
"name" : "waiting","metric_type" : "gauge","units" : "number of waiting connections","data_type" : "int","polling_frequency" : 10,"regex" : ".*Waiting: (+)" }
]
}
当前cAdvisor只支持http接口方式，也就是被监控容器应用必须提供http接口，所以能力较弱，如果我们在collector这一层做扩展增强，提供数据库，mq等等标准应用的监控模式是很有价值的。在此之前的另一种方案就是如上图所示搭配promethuese（其内置有非常丰富的标准应用的插件涵盖了APM所需的采集大部分插件），但是这往往会导致系统更复杂（如果应用层并非想使用promethuse）
在Kubernetes监控生态中，一般是如下的搭配使用：
https://images2015.cnblogs.com/blog/532603/201702/532603-20170226132647398-1702986716.png
页:
[1]