average or maximum value over 24h?

Support questions about the Network Weather Map plugin

Moderators: Developers, Moderators

Post Reply
trogs
Posts: 14
Joined: Mon Apr 20, 2009 5:31 pm

average or maximum value over 24h?

Post by trogs »

Hi,

Just wondering if it's possible to get the average value (or preferably the max value) over the last 24 hours and graph that? I think I read a thread a while ago that this was going to be possible in 0.96 ? Not sure if this made it in or not.

Basically I'm trying to create a graph to check link errors and turn red if there were any errors over the last 24 hours for a bunch of links. This would probably work better if I could check the max_value over the past 24h, as I'm not sure that the errors last long enough to show up in an average.

I found the SET rrd_start -1d command but this only gives me an instantaneous value for 1 day ago, not an average since then.

Anyone have any ideas?
User avatar
Howie
Cacti Guru User
Posts: 5508
Joined: Thu Sep 16, 2004 5:53 am
Location: United Kingdom
Contact:

Post by Howie »

http://www.network-weathermap.com/artic ... dtool.html

should have the information you need. However, I broke something in 0.96 and 0.96a, so you might need to get the WeathermapDatasource_rrd.php from subversion to get this working. That's available here.

I'll hang on a bit longer and then release 0.96b. There are a few more unusual outstanding bugs to resolve first though.
Weathermap 0.98a is out! & QuickTree 1.0. Superlinks is over there now (and built-in to Cacti 1.x).
Some Other Cacti tweaks, including strip-graphs, icons and snmp/netflow stuff.
(Let me know if you have UK DevOps or Network Ops opportunities, too!)
trogs
Posts: 14
Joined: Mon Apr 20, 2009 5:31 pm

Post by trogs »

Thanks, got it going.

Code: Select all

SET rrd_start -2days   
SET rrd_period 172800
SET rrd_aggregate_function MAXIMUM
Needed to use the max value across a couple of days, and add a scale:300000 to the target in order to get the map to display a good number. As the value in the RRD is a 5 minute average of errors - * that by 300 to get an actual number of errors, and 1000 to get it to an actual number to display on the graph (pretty much doing no label on this, but want to catch any errors and turn the link red)

Also need to use multiple targets, so I get a view from both ends of the link, as errors will often only show up unidirectionally on one port.

Works pretty good to show up faults that have been in the network across the last couple of days :)
Post Reply

Who is online

Users browsing this forum: No registered users and 0 guests