...
GUI Profiling
Enable profiling in Checkmk
Info SETUP→ General → Global Settings → USER INTERFACE
Don't use 'Enable profiling for all request'. This option will profile a Multisite request randomly and at the most of the time it will be the slowest one.
Install the Python module (for Python 2 or Python 3 respectively)
Code Block language bash root@cmk-VirtualBox:~# pip install snakeviz root@cmk-VirtualBox:~# pip3 install snakeviz
Modify the URL to profile by adding '&_profile=1' and fire the request in your browser
Code Block language bash http://localhost/nagnis_master/check_mk/view.py?view_name=allhosts&_profile=1
Now two files are created in ~/var/check_mk/
Code Block root@klappanas:/opt/omd/sites/nagnis_master/var/check_mk# ll |grep multisite -rw-rw---- 1 nagnis_master nagnis_master 100246 Nov 6 13:35 multisite.profile -rwxr-xr-x 1 nagnis_master nagnis_master 155 Nov 6 13:35 multisite.py*
By executing 'multisite.py' you can get runtime statistics about the last processed page
Code Block root@klappanas:/opt/omd/sites/nagnis_master/var/check_mk# ./multisite.py |more Fri Nov 6 13:35:33 2020 /omd/sites/nagnis_master/var/check_mk/multisite.profile 9585 function calls (9480 primitive calls) in 0.007 seconds Ordered by: internal time ncalls tottime percall cumtime percall filename:lineno(function) 1 0.001 0.001 0.002 0.002 /omd/sites/nagnis_master/lib/python/cmk/gui/modules.py:140(_cmk_gui_top_level_modules) 4652 0.000 0.000 0.000 0.000 {method 'startswith' of 'str' objects} ... ... ... ...
For viewing multisite.profile with snakeviz, move the file to another directory (e.g. /tmp) and change the owner to the owner of your Internet Browser
Code Block language bash anastasios@klappanas:/omd/sites/nagnis_master/var/check_mk$ sudo cp multisite.profile /tmp/ anastasios@klappanas:/tmp$ sudo chown anastasios:anastasios multisite.profile
Install the Python module (for Python 2 or Python 3 respectively)
Code Block language bash root@cmk-VirtualBox:~# pip install snakeviz root@cmk-VirtualBox:~# pip3 install snakeviz
Now you can use snakeviz for viewing the profiling file
Info Code Block anastasios@klappanas:/tmp$ snakeviz multisite.profile
Now you can analyze at which point of the request your view needs the most time!
Profiling via cli
The cmk commands have an option to profile them
Code Block OMD[at]:~$ cmk |grep profile --profile Enable profiling mode
The profile file will be created in the current directory. Let's go to tmp and run the cmk command with the profile option
Code Block ➜ ~ su - at OMD[at]:~$ cd tmp OMD[at]:~$ cmk -Afv --profile
Note cmk -Afv --profile is only one example. This command will trigger the agent baking via the cli. You may need to profile a different command
Now we will see two files:
Code Block OMD[at]:~/tmp$ ls |grep profile profile.out* show_profile.py*
To work with these files we need to copy them outside the site
Code Block OMD[at]:~/tmp$ cp show_profile.py profile.out /tmp/
- Now there are two ways to analyze these files:
Open the profile file via the commandline:
Code Block ➜ /tmp ./show_profile.py
- Analyze via snakevize like described here: Checkmk profiling
Profile a checkmk function
Warning Please note that we do not support any changes of our code. If you do this, please keep a backup of the file!
There is an option to profile a function inside checkmk. This can be either your own code e.g. plugin or a checkmk specific code:
Code Block from cmk.utils.profile import profile_call ..... @profile_call("/tmp/myprofile_dir") def my_func_to_profile(self): print("ding dong")
Analyze a profiling file with python3 but generated on python2
Note If you're running Checkmk <1.6 and >2.0 on your system, you will also have python2 und python3 installed.
If you generate a profile on Checkmk 1.6 (pyhon2) you won't be able to open this file, if python3 is your default python on your system.
There is a simple to achieve this anyway:
Info The solution is: Python virtualenv inspired by: https://computingforgeeks.com/how-to-install-python2-with-virtualenv-on-ubuntu/
Here is my short step-by-step guide:
Code Block # Install virtualenv sudo apt install virtualenv # Create a directory for virtualenv mkdir ~/.virtualenvs # Change to this directory and create a virtualenv with the python interpreter you want (we will use python2) cd ~/.virtualenvs virtualenv --python=python2 env # Confirm that the right python version is installed ls env/lib # Activate the new virtualenvironment source env/bin/activate # Now you can install all kind of modules inside this python2 context pip install snakeviz
In this env you can analyse all profile files generated with python2!
Network Analyze with the internet browser
Start the Network Monitor
Info Ctrl + Shift + I
or
For Firefox follow this manual: https://developer.mozilla.org/en-US/docs/Tools/Network_Monitor
For Google Chrome follow this manual: https://developer.chrome.com/docs/devtools/network/- Now reload the slow/ crashing view again and save/export this file as .har. Please submit this file to us
...