Xymon Mailing List Archive search

Metrics Report for other tests

4 messages in this thread

list Charles Jones · Thu, 10 Aug 2006 14:12:40 -0700 ·
I only see options to create metrics reports for CPU, Disk, RAM, and 
Ping. Is it possible to make metrics reports for other defined services 
as well?

Ones I would be interested in are:
* Clock offset
* Network Traffic
* Processes (including TRACKed procs)
* Any custom NCV graphs

I guess I was expecting to be able to do a multi-host metric comparison 
of any available graphs.

-Charles
list Henrik Størner · Thu, 10 Aug 2006 23:15:23 +0200 ·
quoted from Charles Jones
On Thu, Aug 10, 2006 at 02:12:40PM -0700, Charles Jones wrote:
I only see options to create metrics reports for CPU, Disk, RAM, and 
Ping. Is it possible to make metrics reports for other defined services 
as well?

Ones I would be interested in are:
* Clock offset
* Network Traffic
* Processes (including TRACKed procs)
* Any custom NCV graphs
It is, but right now it can be rather tricky to get the hobbitgraph
definitions correct for it to work.

The list of graphs that you can generate metrics for is customizable;
it is in the ~hobbit/server/web/hostgraphs_form webpage template.
The option values there correspond to items in the hobbitgraphs.cfg
configuration, but must be "multi" graph definitions.


Regards,
Henrik
list Charles Jones · Thu, 10 Aug 2006 16:47:06 -0700 ·
quoted from Henrik Størner
Henrik Stoerner wrote:
On Thu, Aug 10, 2006 at 02:12:40PM -0700, Charles Jones wrote:
  
I only see options to create metrics reports for CPU, Disk, RAM, and 
Ping. Is it possible to make metrics reports for other defined services 
as well?

Ones I would be interested in are:
* Clock offset
* Network Traffic
* Processes (including TRACKed procs)
* Any custom NCV graphs
    
It is, but right now it can be rather tricky to get the hobbitgraph
definitions correct for it to work.

The list of graphs that you can generate metrics for is customizable;
it is in the ~hobbit/server/web/hostgraphs_form webpage template.
The option values there correspond to items in the hobbitgraphs.cfg
configuration, but must be "multi" graph definitions.
  
Hmm there don't seem to be many "multi" graphs:
]$ grep multi]  hobbitgraph.cfg
[la-multi]
[la1-multi]
[ram-multi]
[swap-multi]
[mem-multi]
[users-multi]
[conn-multi]
[netstat3-multi]
[apache3-multi]

Is there a procedure for creating "multi" versions of non-multi graphs?

Thanks,
-Charles
list Tom Georgoulias · Tue, 15 Aug 2006 17:02:57 -0400 ·
quoted from Charles Jones
Charles Jones wrote:
Ones I would be interested in are:
* Clock offset
* Network Traffic
* Processes (including TRACKed procs)
* Any custom NCV graphs
Hmm there don't seem to be many "multi" graphs:
Is there a procedure for creating "multi" versions of non-multi graphs?
I made several of them for our environment, both from the default Hobbit provided tests (apache %CPU) and from our custom NCV data (mysql queries/sec, memcached %hits, etc), simply by copying one of the multi graph defs and adjusting the variable extracted from the appropriate rrd file.  I've found that the multi graphs only work when you are graphing one variable per host, so you can't do something like RAM & Swap for 2 hosts in a single chart.

For example, this def should allow you to see the %CPU apache is consuming on your web servers, if you are using the apache test in bb-hosts.  Hope it helps.

Tom

[apache2-multi]
         TITLE Multi-Host Apache CPU Utlization
         FNPATTERN apache.rrd
         YAXIS %CPU
         DEF:C at RRDIDX@=@RRDFN@:CPU:AVERAGE
         LINE2:C at RRDIDX@#@COLOR@:@RRDPARAM@
         GPRINT:RPS at RRDIDX@:LAST: \: %5.1lf (cur)
         GPRINT:RPS at RRDIDX@:MAX: \: %5.1lf (max)
         GPRINT:RPS at RRDIDX@:MIN: \: %5.1lf (min)
         GPRINT:RPS at RRDIDX@:AVERAGE: \: %5.1lf (avg)\n