This is a simple script that will create gnuplot graphs out of any metric. It's not the smartest thing ever created, actually it is horrible in many ways but its a result of 1-2 days work. Its fast, has no dependencies and if you set it up it will work. Not sure how much more effort i will put on this but if it goes well for me in my production systems, i am definitely planning to improve.
To Do : move properties of the graph elsewhere , add many checks , change output.csv to a variable,make the plot command more simple
It requires gnuplot, stadard unix utilities and imagemagick if you want to add a custom logo. The script requires some configuration before its executed, like input / output dir (check the config for more detail and a brief sentence of what each variable means.) When its good to go then just run : ./sysgraph <date eg: 2014-12-29> <graph name, check help>
New graphs can be added relatively easy and that's what the rest of the documentation describes.
Current implemented graphs are :
cpu_load load_avg io_read io_write io_wait mem_used active_conn (active postgresql connections) pg_shared_buffers (it doesn't get it from pg_shared_buffers view, so its not very informative)
Other graphs can be easily added but you will need 3 things :
-
A Parser Parser can be any program or script, it has to have at least 2 variables for input and output dir, it can have many other variables defined in the configuration file (see bellow) parsers final file should be called output.csv (to be fixed and defined in the config file).
-
config file configuration file has options for the parsers and some global variables defined, it can hold all configuration parameters for the parsers, like :
read_disk="fioa" read_column="rkB/s"
- script changes
Adding a new parser needs some editing in the script (sysgraph.sh):
in function _property:
elif [ $graph = io_read ] then title="io read report for $date" xlabel="time" ylabel="KB" limit="10000" file_pattern="iostat" plot_command="plot '$tmpdir/output.csv' using 1:2 title 'Reads KB' with lines ls 1"
A new block similar to this has to be added for each new graph title = title of the graph x/ylabel = title of X or Y axis limit = limit that you want to set (useful for big spikes) file_pattern = the pattern of the file that the script to work with plot_command = the plot command that will run, "1:2" means plot field 1,2 , "lines ls 1" is the type and style of the graph
A new block similar to the following has to be added in function _run_parser
elif [ $graph = io_read ] then $parsedir/reads_writes.sh $tmpdir $tmpdir $read_disk $read_column $read_op
the variables should be defined in configuration file
to add a logo on the graph you need to have imagemagick and a logo logo's path can be defined in the configuration file.