So I am not sure if this belongs on the dev list as I am not really doing core dev work and I am not currently subscribed to it. I will definitely contribute my work back to you guys though (if you wanted it). Its for graphing md3000i performance data (stupid thing isnt accessible through snmp, so its basically data attached to the management host). Anyway, I have been following the following example: http://www.observium.org/wiki/Developing/NewGraph and I am not sure about few things.
1) What are the following values for? Specifically the Pages/pages part. I understand that it cycles through the stats that i have setup, but I am not sure about the words Pages/pages.
foreach ($stats as $stat) { $i++; $rrd_list[$i]['filename'] = $rrd_filename; $rrd_list[$i]['descr'] = "Pages " . ucfirst($stat); $rrd_list[$i]['ds'] = "pages" . ucfirst($stat); }
2) Now that I have the agent script created and the rest of the scripts in the tutorial, how am I supposed to activate it for a host? I am assuming it should probably be "activated" within the applications list? If so, I am guessing there is logic missing in that example tutorial on how to add it to the list of application options?
3) Is creating a diff and reapplying after every upgrade really the recommended method? Its to bad all the definitions are in a single file or else that might be a little easier to manage.
4) Any idea what type of logic i should be using to figuring out the proper numbers for the rrd graph settigns in this code section?
rrdtool_create($rrd_filename, "--step 300 \ DS:pagesShared:GAUGE:600:0:125000000000 \ DS:pagesSharing:GAUGE:600:0:125000000000 \ DS:pagesUnshared:GAUGE:600:0:125000000000 ".$config['rrd_rra']);
5) What format should the agent script output be in? I am guessing it should be one per line? Right now my output is like so:
<<md3000i>> total_io=3158.0 read_percentage=27.9 cache_hit_percentage=93.2 current_kbps=22475.2 current_iops=631.6
6) Last question: Is there any way to have certain agent scripts to be polled at slower intervals? This check takes about 5 seconds and is probably a resource hog on the management host since the script has to run a stupid java cli app. The management server in this case is the same host as the observium system, so I wouldnt have a problem not using the agent and just doing a cron if that was somehow an option.
Thanks for you help. I know I had a lot of detailed questions. I sincerely appreciate the help.
-Mark (MACScr)