[hobbit] Custom graphs, passing data

[Henrik Stoerner]

The way i'd do that is to split the data you want graphed into a
separate message using the "data" protocol instead of a "status".
E.g. in your custom test generating the data, you'd have 

$BB $BBDISP "data $MACHINE.mydata
item1: value1
item2: value2
"

which feeds the lines into Hobbit, and into the NCV module using
"mydata" as the ID. So you'd need to setup NCV_mydata in
hobbitserver.cfg as described in the docs.

Having sent a "data" message doesn't prevent you from generating a
"status" message in the same script. And then you can have whatever 
formatting you like for the status since it won't go through the NCV
module. Just dont call your status column the same as the ID you used
for the "data" message. So ...

$BB $BBDISP "status $MACHINE.mytest green `date` OK
... all of the fancy stuff here ..."

To get the graph to show up on the status page, make sure you have a
TEST2RRD entry with "mystatus=mydata".


Regards,
Henrik


On Tue, Jun 05, 2007 at 12:49:04PM -0600, Haertig, David F (Dave) wrote:
> There appear to be two ways to get data into a custom graph.  (1) Use
> the built-in NCV module to parse it out of a Hobbit message, or (2) use
> the --extra-script and --extra-test options to hobbitd_rrd.  I have
> tried (1) but not (2) yet.
>  
> I want to use (1) for it's efficient operation, but the Hobbit message
> format it requires for parsing looks ugly when displayed on the webpage
> (at least I think it's ugly).  So I was looking for ways to "hide" the
> data from the webpage while still allowing the NCV module to find it.
> My first thought was to send the Hobbit message with the graphing data
> hidden inside a <div style="display:invisible">...</div> block.  But
> this doesn't work because when generating the webpage, Hobbit puts the
> incoming message inside a <table>, and you can't hide <div>'s when
> they're inside a table cell.
>  
> Does anyone have other solutions to this, while still using method (1)
> so the NCV module can efficientlhy parse things?  Or am I stuck with
> learning and moving to method (2)?