AboutUsPerformanceMonitoring
The responsiveness and performance of the site makes a big difference in how many pages visitors will view, and how often they will come back. A poorly performing site will also wear out our active members causing some of them to leave.
Monitoring Resources Internally
We have chosen to try an open source performance monitoring tool called Zabbix [1]. Zabbix includes many features including the following.
- Real-time performance, availability, and integrity monitoring
- Web site monitoring
- Data visualization through use of graphs, maps, and user-defined views
- Condition notifications (including e-mail and jabber)
- Distributed monitoring and configuration
The dashboard view and notifications can easily point out immediate issues with the backend that could potentially affect site performance. Even benign issues such as occasional load spikes are logged and can used to indicate potential problems on the horizon. This allows us to be more proactive and stay in step with site growth.
The graphs allow for similar, user-friendly views into the AboutUs environment. These can be utilized to identify trends. They can also show equipment, service, and resource relationships by comparing multiple graphs on a single page.
You will find the AboutUs Zabbix site at the following URL:
https://admin.aboutus.org/zabbix/
The same credentials for the AboutUs stats can be used to log into Zabbix. Please contact the systems administrator if you require a personal user account.
Points Of Interest
The layout of Zabbix is fairly self explanatory. Most general use will be under the Monitoring tab. While more detailed documentation can be found in the Zabbix manual [2], a list of particularly useful items can be found below.
- Monitoring / Overview
The Overview is refreshed every 30 seconds and includes a dashboard view of all triggers. From the Zabbix manual, "a tigger is used to define constraints on items and provide notifications when these constraints are exceeded." A return of success is presented with a green box while a problem is presented as a red box. Grey and white boxes indicate that particular trigger is disabled.
- Monitoring / Web
This page provides the status of HTTP site requests. Clicking on each HTTP item will provide download speed and response time graphs. It is important to note each item is a single HTTP request or steps of several sequential requests. This data does not represent a user session as requests for all images, scripts, and external links are not followed. It does, however, display raw site performance trends related to backend performance and availability.
- Monitoring / Latest data
While you can retrieve graph data under the Overview section by clicking on the colored boxes, the "Latest data" area provides performance related graphs for various triggers such as CPU utilization, server load average, memory useage, and network throughput. A custom time period can be defined for any graph.
It is important to note that defining longer time periods will slow graph generation.
- Monitoring / Screens
Custom views can be created using graphs and other Zabbix objects. An example view has been put together that includes the CPU utilization of all the database servers.
- Reports / Most busy triggers top 100
Displays a list of the most common issues. This type of data can help isolate both existing and potential problems.
- Configuration / Screens
The screens mentioned above can be created here. Here are the steps used to defined the "Database server [CPU utilization]" screen.
- Make sure Screens is selected in the drop down box and click Create Screen
- Enter "Database server [CPU utilization]" for the screen name and select 2 columns and 3 rows. Click Save.
- Click the Edit link of the newly created screen.
- More to come...