ZLM-Cython Build you first module
- 1. ZABBIX www.zabbix.com Building your first Python module for ZLM-Cython Using zlm-cython with ease...
- 2. ZABBIX www.zabbix.com AGENDA What you shall know before you begin. Locations Building a module Testing a module Using a module What's next ?
- 3. ZABBIX www.zabbix.com What you shall know before you begin. Apparently, you shall know the Python. You shall know it well enough, to be able to write a short and fast snippets of the code. You shall be proficient with all libraries tools and protocols involved in your project. For example: if you are monitoring REDIS(1) storage, you shall be familiar not only with Python itself, but with Python module which you will use to access REDIS, and with REDIS itself. (1) REDIS – high-performance in-memory data structure store, used as database, cache and message broker. http://redis.io While you beginning to developing you module, you shall be also familiar with all “common-sense” Python performance programming techniques. I am recommending to look at the book “High Performance Python” by Ian Ozsvald. You shall be also proficient in the area of System Administration and Systems Management well enough, so you can understand the impact of the execution of your code to the target hosts and applications. You must be comfortable when dealing with such hosts, operating systems and all applications involved.
- 4. ZABBIX www.zabbix.com What you shall know before you begin. And of course, you must have “hands-on” experience on how to use OS shell and OS commands, how to edit text files, how to use source-code version control tools used by developers in your organization, how your code will be integrating with configuration management, build and deployment tools. And this is not the end. You shall be able to create “secure” code. This means, that you code must pass check, performed by your security team. You must be familiar with all local guidelines and requirements related to application security. Systems and Application Architect shall know about your software development and deployment activities. You must know how to document your code. And actually do that Documentation And the last, but not least. You shall know how to test your code for a bugs and a performance issues.
- 5. ZABBIX www.zabbix.com What you shall know before you begin. Well … You are not just in IT-monitoring anymore. You are becoming Software Developer and must think and act like one.
- 6. ZABBIX www.zabbix.com Locations Open your Zabbix configuration file (1) and locate variable LoadModulePath Verify, that you are loading zlm_python.so module is loaded. Look for variable LoadModule=zlm_python.so Your module files and all other directory and files will be relative to the location, defined by this variable. Please locate directory pymodules, which shall be subdirectory of the directory defined in LoadModulePath . Verify location of your log files as defined by LogFile. (1) zabbix_server.conf or zabbix_agentd.conf. If this directory do not exists, please create it and assign ownersyhip to user cefined in variable User. It is probably a good idea to set user sticky bit for this directory.
- 7. ZABBIX www.zabbix.com Building a module With text editor of your choice, please create file called ZBX_test.py and define three simple functions. Please note: Default function of the module Is main() This function return Integer This function return Float This function return String
- 8. ZABBIX www.zabbix.com Building a module We support only those three data types: Integer, String and a Float. ZLM- Python will automatically convert Python value to a Zabbix value. What you return from the function, will be passed directly to Zabbix All exceptions will be properly handled. We will talk about exceptions in a little while. You can refer individual functions inside the module from the Zabbix as: Module name.Function name. Default function name is “main” The first parameter, passed to a function by ZLM-Python is “context”, usually referred by variable name “ctx”. This object is used to pass data between different process and will be explained in more details later on.
- 9. ZABBIX www.zabbix.com Testing a module Before you will try to query the metric through Zabbix Agent or Zabbix Server, you shall Test a metric through zabbix_agentd -t <metric name> Testing this metric ZLM-Python initializing ZLM-Python loading configuration ZLM-Python pre-load our test module ZBX_test.this_is_int() returns 42. It is “unsigned Integer”. SUCCESS !
- 10. ZABBIX www.zabbix.com Testing a module What if something goes wrong ? What yo will see from zabbix_agentd -t <metric name> ? Let's create a metric collection function, which will do nothing, but throw a Python exception. When called, this function will throw an AttributeError exception and pass the parameter string. Raise an Exception AttributeError … and here is parameter
- 11. ZABBIX www.zabbix.com Testing a module What if something goes wrong ? What yo will see from zabbix_agentd -t <metric name> ? Call that metric Looks okay, so far Oy ! Our metric threw a traceback and become NOTSUPPORTED And here goes Python traceback, to help us to troubleshoot the problem.
- 12. ZABBIX www.zabbix.com Testing a module When you metric functions exit successfully and generate proper output, you can move to the next test: collecting data through zabbix_agentd (if your metric collector designed to run from zabbix_agentd) If your zabbix_agentd -t <metricname> is core-dumped, CONGRATULATIONS ! You've found a bug in ZLM-python. Yes, there are bugs and I will try to fix them as soon as I can. Please report this bug to https://github.com/vulogov/zlm-cython/issues Please include as much information as possible.
- 13. ZABBIX www.zabbix.com Testing a module Try to query your new metric keys (if you are loading the module inside Zabbix Agent) Here, we are calling default function Calling function which returns Integer, getting Integer Calling function which returns Float, getting Float Calling function which returns String, getting String
- 14. ZABBIX www.zabbix.com Testing a module The previous attempt was successful, but what if we get an error ? Collect this metric Item will become a NOTSUPPORTED And here goes Python traceback, to help us to troubleshoot the problem.
- 15. ZABBIX www.zabbix.com Using a module Now, let's create a Zabbix Item, for the function that returns Integer Name for the metric Collecting from zabbix_agendCalling ZLM-python Module: ZBX_test Function: this_is_int() Type of the Zabbix Item shall match to the type of the data returned from metric function
- 16. ZABBIX www.zabbix.com Using a module Now, let's create a Zabbix Item, for the function that returns float Name for the metric Collecting from zabbix_agendCalling ZLM-python Module: ZBX_test Function: this_is_float() Type of the Zabbix Item shall match to the type of the data returned from metric function
- 17. ZABBIX www.zabbix.com Using a module Now, let's create a Zabbix Item, for the function that returns string Name for the metric Collecting from zabbix_agendCalling ZLM-python Module: ZBX_test Function: this_is_float() Type of the Zabbix Item shall match to the type of the data returned from metric function
- 18. ZABBIX www.zabbix.com Using a module Now, let's create a Zabbix Item, for the function that returns float and ZLM-Python is executed on Zabbix Server Name for the metric Collecting on zabbix_server Calling ZLM-python Module: ZBX_test Function: this_is_float() Type of the Zabbix Item shall match to the type of the data returned from metric function
- 19. ZABBIX www.zabbix.com Using a module ZLM-python will help you to troubleshoot you metric collectors by properly passing information about exceptions back to Zabbix. If your metric throw an exception, the Item will become “Not supported” and you can get the information about traceback right from the frontend by clicking on the red icon. Display the information about an exception
- 20. ZABBIX www.zabbix.com What's next ? Passing parameters from Zabbix to Python functions Non-positional parameters You can use ether positional or non-positional parameters. Please note, all parameters are passed as strings. You shall do the proper type sanitation and conversion inside your module.
- 21. ZABBIX www.zabbix.com What's next ? Using the ZLM-python Context object Proxy for the context object of the ZLM-Cython core ZLM-python context object allow you to save some data to the global dictionary and pass it between Zabbix threads. If you assign some variable to the context in one thread, like ctx.name = “Value”, you can refer to it later on using this notation: ctx.name . Context is global and we do support all Python picklable data types.
- 22. ZABBIX www.zabbix.com Q/A ? Вопросы ? 質問 Jautāt ? Fragen ? Interroger ?