Docker Operations Essentials: monitoring The Treasure Docker monitor trial notes

This article is a report by Sho Yuanhao deep practice Docker monitoring

It is a great honor to have a surveillance treasure invitation to try out their recently launched new product--docker monitoring.

September 7, China APM manufacturer Cloud Wisdom Cloudwise officially released on-line Docker monitoring, the product from the deployment to use, the whole process is very simple. Not only can real-time monitor the performance information of the host and Docker containers (including CPU, MEM, disk, Net in/out), but also can customize the corresponding alarm policy. The following will be from the deployment, monitoring information, alarm these aspects of the trial experience. You can [register] monitoring treasure, free use of docker monitoring.

Deployment process

Read the article on the Dockone "grilled a grilled monitoring treasure Docker monitoring technology principles", understand that the implementation of Docker monitoring is based on the smartagent architecture, the entire deployment process can be completed within a few minutes.

The first step, click "Create Monitoring Project", enter the basic information, including the name and frequency of monitoring, you can see the specific deployment steps.

The second step is to install the agent and Docker plugin on the monitoring machine.

L first download, unzip and start Sendproxy,sendproxy is a proxy, as the sending engine, can be deployed in the LAN, the monitoring information of the machine in the LAN is efficiently transferred to the Cloud Intelligent SaaS platform. Sendproxy can be started by sendproxy.sh script, command './sendproxy.sh start ', after execution, you can view sendproxy status by command './sendproxy.sh status ' if " States "is a" OK "state, it means that Sendproxy started successfully.

Second, download, unzip, and launch the Docker plugin--docker Agent,docker Agent is the main module for Docker monitoring, which collects data on the monitoring machine and transmits it via SENDPROXY to the Cloud Smart SaaS platform. You can launch the Docker Agent directly using the start.sh script.

After these two steps, on the monitoring Treasure's Docker monitoring page, you can see that the monitoring project just created has acquired the data on the monitoring machine.

I do not know whether people will have a question, "How is the data located to the monitoring project just created?" "Venture to speculate that when a monitoring project is created, the name entered and the frequency of the set monitoring are stored in the monitoring project and the Monitoring project information is written to the Docker agent's configuration file so that the monitoring project can be mapped." But there is one question, "What if there are two monitoring items on that machine?" , think about a single monitoring project on a single machine, and a monitoring project can monitor multiple machines.

> Based on hands-on practice, there are a few small suggestions for the problems encountered in the deployment process.

> (1) In the deployment prompt, explain to the user the type of operating system supported by the Docker agent now;

> (2) Prompt the user, monitor the relationship between the project and the Docker plugin;

> (3) Prompt the user, if the Docker monitoring page has not received data, you can use the Docker_py script in the bin directory to try push monitoring information.

Monitoring information Presentation

After the deployment is complete, you can enter a specific information display interface. The information presented is comprehensive, including monitoring information on CPU, memory, disk and network traffic, and monitoring the statistics of Docker containers in different states on the machine (so comprehensive information that mom no longer has to worry about Docker operations). The following is the overall presentation of the monitoring information.

Monitoring information Unified use of line chart display, more intuitive, if you want to know the value of specific data, you can switch to the Data view on the right side of the chart. If you want to know the data for a certain period of time, you can choose at the top of the page, the default is "Today", "Yesterday" and "last seven days" three options, of course, you can also customize the time range as needed. The data lines on the icon will inevitably overlap, and you can turn off some of the data lines by clicking on the title above. These details are convenient and commendable.

The specific data values are displayed as follows:

In the specific monitoring information display page, such as net In/out category, you can see the monitoring data of different containers and some statistics (maximum, average and minimum value), by default, will show "resource consumption Top10" 10 containers, Of course, you can also select a specific container's data display by the drop-down box in the upper left corner.

Alarm

Alarm function is undoubtedly one of the most important features for OPS and developers. The user can customize the alarm settings in the Docker monitoring of the cloud intelligence monitoring treasure. The alarm object is mainly for the resource usage of the container and the survival rate of the container.

The alarm policy is set according to the statistic data (mean, sum value), which is higher, lower than or equal to the set threshold, and the corresponding alarm is made. For resource usage, alarm settings can be set for all containers or for individual containers.

It is worth mentioning that the surveillance Bao's alarm method is very comprehensive, can be emailed, SMS, phone voice, app push, and other ways to notify, especially through the telephone voice and the way to ensure that you do not miss any important alarm information.

Overall experience

Since the advent of Docker, OPS has been a pain point for Docker users. Cloud Intelligence launches the Docker monitoring that fills the gap in domestic docker monitoring. From deployment to monitoring, the overall feel is easy to understand and easy to use. The entire deployment process, simply download the two zip installation package, and then modify the appropriate permissions, start is good, for novice or small white users, this is very simple operation.

The monitoring information is displayed using a chart, but it does not forget to provide the data view interface to the user who needs the specific data, and the details are in place. Into the CPU, mem and other specific monitoring Information section, can see the specific individual container monitoring information, the effect is good. The alarm part, the strategy is clear, through the statistic average or the value to measure, sets the condition and the threshold value to trigger the alarm, may the individual container performance monitoring information to do the alarm, or is more meticulous.

Finally, to make a suggestion, whether to join the "group" or "cluster" related concepts, in the monitoring display, you can choose to display the "cluster" monitoring information, in the alarm settings, you can set the "cluster" alarm information. Sometimes a host will use several containers as a cluster, such as a Hadoop cluster. It is cumbersome and inconvenient to observe the monitoring information individually or set the alarm information for each container.

Finally, thank you very much for monitoring the opportunity to give this trial Docker monitoring, hope that the monitoring treasure can bring us more surprises.

Monitoring treasure Docker monitoring is being used free of charge, welcome to experience

Docker Operations Essentials: monitoring The Treasure Docker monitor trial notes