Combining Zabbix with an external monitoring tool
In our environment we want to use VCops as central monitoring tool and use several dedicated monitoring tools to monitor their specific set of servers / services / tools, such as Zabbix for Linux/Unix servers and Scom for Windows servers. The idea behind VCops is that it can easily display whole customer services which incorporate several Windows and Linux servers and perhaps one or more Oracle database, network components and so on. Therefore a problem with one component can be linked easily to a problem for a customer service.
With Zabbix storing its data in a MySQL database, it can be accessed from outside, and since VCops can fire off SQL statements, it can receive all the data from the database. However we are currently running into a problem with either the SQL statement or interpretation of the received data on the VCops side, which I would like to demonstrate with the following example.
Let's say we are monitoring a Linux server with an apache server on it. Since the application management team only interests in the apache process being up, Zabbix only monitors the apache process. In the apache configuration file it states that 8 server processes should be started up, so if the process count of the httpd process is less than 8 we have a problem. Ofcourse monitoring an apache webserver goes further than just monitoring the apache process, but in sake of this example let's just stick with the process monitoring.
In Zabbix we add the server, the item apache process counter and the trigger where we define the error situation to be less than 8.
Zabbix receives the data from the Linux server and starts storing the data for trend analyses and such. On the same time VCops starts to pull the stored data from the MySQL database. However, all VCops sees is the stored value (say 20) for this item. VCops doesn't know if this value is right or wrong, since that interpertating is done by Zabbix.
Our SQL statement needs tweaking, that is for sure, but I am wondering if we are not asking more from either Zabbix or VCops than we can expect. Is there a table in the Zabbix database where occured errors are stored? When Zabbix determines an item to be faulty, say less than required diskspace and such, is that stored somewhere in the database? In the Zabbix dashboard we can acknowledge an error, so I am wondering is that information can be retrieved from the MySQL database, or if that is just Zabbix internal technology.
We are using Zabbix 2.0 on RHEL 6.
With Zabbix storing its data in a MySQL database, it can be accessed from outside, and since VCops can fire off SQL statements, it can receive all the data from the database. However we are currently running into a problem with either the SQL statement or interpretation of the received data on the VCops side, which I would like to demonstrate with the following example.
Let's say we are monitoring a Linux server with an apache server on it. Since the application management team only interests in the apache process being up, Zabbix only monitors the apache process. In the apache configuration file it states that 8 server processes should be started up, so if the process count of the httpd process is less than 8 we have a problem. Ofcourse monitoring an apache webserver goes further than just monitoring the apache process, but in sake of this example let's just stick with the process monitoring.
In Zabbix we add the server, the item apache process counter and the trigger where we define the error situation to be less than 8.
Zabbix receives the data from the Linux server and starts storing the data for trend analyses and such. On the same time VCops starts to pull the stored data from the MySQL database. However, all VCops sees is the stored value (say 20) for this item. VCops doesn't know if this value is right or wrong, since that interpertating is done by Zabbix.
Our SQL statement needs tweaking, that is for sure, but I am wondering if we are not asking more from either Zabbix or VCops than we can expect. Is there a table in the Zabbix database where occured errors are stored? When Zabbix determines an item to be faulty, say less than required diskspace and such, is that stored somewhere in the database? In the Zabbix dashboard we can acknowledge an error, so I am wondering is that information can be retrieved from the MySQL database, or if that is just Zabbix internal technology.
We are using Zabbix 2.0 on RHEL 6.
Comment