[Xymon] Monitoring multiple values
David Welker
dewelker at gmail.com
Wed Aug 27 16:15:53 CEST 2014
Ralph and Jeremy,
I was able to get a graph, but the 5 figures for one column are in the
k/seconds range while the other are milliseconds, so while one shows up
rather well, the other hovers around the 0 line.
1.) Is there a way to change this - like the default time of a Xymon graph,
say last 2 hours, on a column status page?
2.) What might be even better, is there a way to display 2 graphs on a
column status page with each in their own graph?
3.) Also, is it possible to display the actual data point values captured,
on the GPRINT line as I mentioned earlier?
Thanks again!
David
On Thu, Aug 21, 2014 at 11:06 AM, David Welker <dewelker at gmail.com> wrote:
> Right on target! Just before I sent that last response, I figured out
> Xymon wouldn't let me use the N-300, etc. notations and decided to try it
> via Rrdtool directly, so at least I was thinking along the right lines!
> Only three problems...
>
> 1) Minor problem - just an fyi, but trying to update a non-existent rrd
> fails, so you have to add a bit more code to create.one first...no big
> deal, but brings me to....
>
> 2.) If I want to eventually show all 5 figures, plus an average, min, max,
> and last, how do I...
> a.) define the rrd? I have:
> DS:time:gauge:600:U:U
> RRA:AVERAGE:0.5:1:1200
> RRA:MIN:0.5:24:2400
> RRA:MAX:0.5:24:2400
> RRA:AVERAGE:0.5:24:2400
> b.) define the graph? I have:
> [tvalues]
> TITLE Time Values
> YAXIS Time (m/s)
> DEF:t1=time.rrd:t1:AVERAGE
> DEF:t2=time.rrd:t2:AVERAGE
> DEF:t3=time.rrd:t3:AVERAGE
> DEF:t4=time.rrd:t4:AVERAGE
> DEF:t5=time.rrd:t5:AVERAGE
> LINE2:
> GPRINT:t1:%6.11f%s (cur)
> GPRINT:t2::\: %6.11f%s (cur)
> GPRINT:t3::\: %6.11f%s (cur)
> GPRINT:t4::\: %6.11f%s (cur)
> GPRINT:t5::\: %6.11f%s (cur)
> GPRINT:t6: \: %6.11f%s (max)
> GPRINT:t6: \: %6.11f%s (min)
> GPRINT:t6: \: %6.11f%s (avg)
> * Note: Assuming that defining the RRAs was correct, I really want the
> RRA values for t6
>
> 3.) lastly, I set up the client-side script to collect the data via a
> clientlaunch.cfg task, the clientlaunch xymond_rrd status then runs my
> script to create the rrds (via the extra-script/extra-tests), so do I need
> extra-script/extra-tests on the xymond_rrd data line?
>
> Currently, both the rrds and the graph are being created, but empty.
>
> Thanks for your suggestions!
>
> Any and all help appreciated,
>
> David
>
>
>
> On Tue, Aug 19, 2014 at 11:29 PM, Ralph Mitchell <ralphmitchell at gmail.com>
> wrote:
>
>> Now that I've had a chance to read the man page, I see I wasn't
>> remembering it properly. I thought the last bit:
>>
>> # The data
>> echo "$TEMP:$WIND:$HMTY:$RAIN"
>>
>> included the timestamp for the RRD insert. So, here's another approach
>> that may work. Instead of echoing out the data, it could be directly
>> written to the rrd, like this:
>>
>> RRD=/home/xymon/data/rrd/<hostname>/thing.rrd
>>
>> TIME=`date +%s`
>> MINUS1=`expr $TIME - 60`
>> MINUS2=`expr $MINUS1 - 60`
>> MINUS3=`expr $MINUS2 - 60`
>> MINUS4=`expr $MINUS3 - 60`
>>
>> rrdtool update $RRD $MINUS4:$DATA4
>> rrdtool update $RRD $MINUS3:$DATA3
>> rrdtool update $RRD $MINUS2:$DATA2
>> rrdtool update $RRD $MINUS1:$DATA1
>> rrdtool update $RRD $TIME:$DATA0
>>
>>
>> You would need to break out the 5 data points from the original message
>> as DATA0 through DATA4. As long as you don't echo out the data values, I
>> *think* xymond_rrd won't update the rrd for itself. Even it it does enter
>> a NaN value, you've just fired off 5 real data points, which ought to be
>> more significant that the NaN. By updating the Xymon rrd for the server,
>> you still get the graphs in the trends column, with multiple time spreads,
>> zooming, etc.
>>
>> I hope that makes sense. I don't have a handy test server to try it out
>> on.
>>
>> Another possibility, if you (or someone nearby) are handy with C. The
>> xymond_sample.c program is included in the Xymon source. It demonstrates
>> how to hook into the message flow and process incoming data. It ought to
>> be possible to achieve the above rrd updates by receiving the message and
>> dropping the data points back into the message stream at 60 second
>> intervals. Maybe. You *really* need a spare test server to try that
>> out... :-)
>>
>> Ralph Mitchell
>>
>>
>>
>>
>> On Tue, Aug 19, 2014 at 11:22 AM, David Welker <dewelker at gmail.com>
>> wrote:
>>
>>> This sounded like a great idea, so I got the status showing up via the
>>> extra-scripts option, but the info on the data message is a little vague.
>>> According to the example, I just echo the DS statements, then the dataset
>>> name on a line by itself, followed by the data for one dataset on a line by
>>> itself, which doesn't seem to do anything in the way of creating any of the
>>> multiple rrds I need, nor does it address the time consideration for
>>> multple values (I need to insert 5 values a minute apart from 5 mins.
>>> before putting them into the rrd - I know this can be done with the rrdtool
>>> update command, but can it be done here somehow?) Anybody have an example
>>> of a multi-dataset case being run via the extra-scripts option of the
>>> xymond_rrd they'd be willing to share?
>>>
>>> Thanks!
>>> David
>>>
>>>
>>> On Fri, Aug 15, 2014 at 10:09 AM, Ralph Mitchell <
>>> ralphmitchell at gmail.com> wrote:
>>>
>>>> You could do that server-side. Look at the man page for xymond_rrd.
>>>> The --extra-script option allows you to pipe incoming data through a script
>>>> to process. You could take the report time and subtract 1 min, 2 min, etc,
>>>> then emit rrd updates as required. The client would still send all 5
>>>> values at the same time, but they'd be recorded as if received over the
>>>> preceding 5 minute block.
>>>>
>>>> Ralph Mitchell
>>>> On Aug 15, 2014 9:55 AM, "David Welker" <dewelker at gmail.com> wrote:
>>>>
>>>>> I've been looking, but haven't seen anything in the documentation, on
>>>>> the website, or in the mailing list that addresses this:
>>>>>
>>>>> I need to monitor 5 values that are received all at the same time.
>>>>> I'd like to insert the values at one minute intervals getting a
>>>>> consolidated data point (CDP) every 5 minutes that I can graph.
>>>>>
>>>>> For example, I ping a host via client script, 5 times, returning all
>>>>> of the data at one time. I want to put all 5 values in the database (rrd),
>>>>> but spaced apart every minute so I get a decent CDP. I could probably do
>>>>> this if this was on the server (could I?), but the script HAS to be on the
>>>>> client in this case. Is this even doable?
>>>>>
>>>>> Thanks, in advance, for any suggestions!
>>>>>
>>>>> David
>>>>>
>>>>> _______________________________________________
>>>>> Xymon mailing list
>>>>> Xymon at xymon.com
>>>>> http://lists.xymon.com/mailman/listinfo/xymon
>>>>>
>>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.xymon.com/pipermail/xymon/attachments/20140827/47091a11/attachment.html>
More information about the Xymon
mailing list