killing server_perf process causes issues

[GoogleCodeExporter]

When initially setting up the system it works great! But, if I kill the
server_perf process and the start it up again the graphs disappear and the
only way to get this up and running again is to wipe the cacti rra files
and remove the cacti database and create a new one.

Is there a special way to stop server_perf to prevent this? 

Original issue reported on code.google.com by ben.pric...@gmail.com on 11 Jan 2009 at 1:39

[GoogleCodeExporter]

also, what data is written to the cacti database if you start server_perf with 
the -m
switch?

Original comment by ben.pric...@gmail.com on 11 Jan 2009 at 1:41

[GoogleCodeExporter]

perf2rrd put the perfdata in :
- rrdfile
- mysql (if -m option) for use with sql query, he use id for host and service 
from
oreon/centreon database configuration. So, You need to use centreon to 
configure nagios.

When I stop and restart I use always the same option : 
./server_perf.pl -du

and I dont see why you have to reinitiliaze cacti configuration... :/ 

Original comment by olivier....@gmail.com on 11 Jan 2009 at 9:20

[GoogleCodeExporter]

I know what is happening now :)

example perf data 

load5=2;80;90;0;100

It looks like the first time n2rrd loads, it only processes the first part of 
the
performance data (2). When you shut it down and start back up it processes 
(80). shut
down, start again "90" etc, etc. this causes the graphs not to be displayed. I 
needed
to delete all data templates, graph templates, rrd's and devices to get working 
again. 

This is not a big issue. Just means I need to write a perl script to strip the 
data
so just the load5=2 remains. I have done this, but it doesn't seem to be 
working.
This is quite hard to troubleshoot as the logging for n2rrd won't work!

Do you have a sample script that would achieve this? I have tried the function 
below
with no luck.

/etc/n2rrd/templates/code/myfile.pl (This matches up fine to my template name)

# file contents 

my $tmp_pdata = substr($tmp_pdata,0,-12);
return $tmp_pdata;

Original comment by ben.pric...@gmail.com on 11 Jan 2009 at 10:54

[GoogleCodeExporter]

when you nagios script use the official perfdata output : 
datasource=value;warn;crit;min;max

like you're example :
load5=2;80;90;0;100

nagios2cacti traduct like this : 
load5=2
load5_warn=80
load5_crit=90
load5_min=0
load5_max=100

so if you want to display all this data, you must define in you're RRA file
(/etc/n2rrd/template/rra), 5 datasources : 
load5, load5_min, load5_max, load5_warn, load5_max

Original comment by olivier....@gmail.com on 11 Jan 2009 at 11:17

[GoogleCodeExporter]

thanks for that. Whats the easiest way to exclude the warn, crit, min and max 
from
the data?

Original comment by ben.pric...@gmail.com on 13 Jan 2009 at 1:16

[GoogleCodeExporter]

The safe way : 
in cacti, edit graph_template
remove three line of gprint foreach datasource to exclude
edit the main line, choose LINE or AREA and apply a CDEF function like that : 
custom string : 0
Special Data Source: CURRENT_DATA_SOURCE
Operator: *

So, you're graph'll be clean when displayed

for you're custom check command, use the simply notation : load5=2 without the
warn,crit,min,max value, they'll be correctly manage.

Original comment by olivier....@gmail.com on 13 Jan 2009 at 6:46

• Changed state: Accepted