Aggregate V0.71 - Released

juliangomezr · Post by **juliangomezr** » Wed Jul 01, 2009 11:10 am

Hi, i hav ethe same problem as bigchoice...

I have installed the plugin, enabled and gave the realm permissions to the prlugin aggregate-->create color templates, but when I went to graph management, and selected four cpu graph and choosed create aggregate graphs, it question to create or to not create that graphs, i selected the yes option, and nothing happened.

It happends to me witeh version 0.70B2 and 0.71, i had disabled, unistalled, an installed with a fresh instaltion and nothing works.

The folder name I had ussed on the plugins folder is aggregate, the same name i used on the global config file.

What can I do for resolve this isue?

Thanks for your help

juliangomezr · Post by **juliangomezr** » Wed Jul 01, 2009 3:02 pm

I found a solution on:

http://forums.cacti.net/about31069.html ... =aggregate

That mension a link to:

http://bugs.cacti.net/view.php?id=1435

It solves my problem.

juliangomezr · Post by **juliangomezr** » Wed Jul 01, 2009 3:33 pm

pthaynes wrote:

Gandalf,

Sorry for wasting your time. I was looking under the host expecting to see the aggregate graph. Really I should have been searching through all graphs containing the name Aggregate. I did that search and there was my graph!!

Im so sorry for making probably a waste of time, but i just be able to create a graph, but i cant find it anywhere, i cant find any graph starting with aggregate name or somthing like that. Where I can find the aggregate graphs created

.

I feel so but asking that but i spended so much time looking for it, and no luck!!!

Thanks for your help!!!!

Post by **gandalf** » Mon Jul 06, 2009 8:42 am

juliangomezr wrote:Where I can find the aggregate graphs created .

How did you name them? You may simply filter for that name.
Or filter for all Graphs with a graph template of "NONE"
Reinhard

juliangomezr · Post by **juliangomezr** » Mon Jul 06, 2009 9:41 am

Hi,Thanks for your answer. What happened to me is that never appear the aggregate graphics, i had disable the thold 4.0.1 plugin, and try to crete them and the appeared with the aggregate name... I'm not sure if there is a later version of the thold plugin, cause I fix a code problem using the link poted previusly, but the created graphic did not appeared until i disabled thold plugin... I have a problem, i can associate the aggregate graph to a host, so it appear sociated to that host, but i cant associte it to a template, because cacti says a message that it will erase the data sources... so i had to leave it with none template. I was creating an aggregte garph with many cpu measure of a same server...
thanks for your help.

juliangomezr · Post by **juliangomezr** » Mon Jul 06, 2009 1:12 pm

Hi.

I have a question, I'm using aggregate to graph multiple CPU, and I graph the total value of CPU use, but it show the sum of all values,so it can show 144% of cPu use, and this is not possible, how I make that the real value or an average value of all cpu, appear in the total value?

Thnaks for your help

juliangomezr · Post by **juliangomezr** » Tue Jul 07, 2009 3:51 pm

Hi, i think i solve my last question, and I want to share it:

1. Copy the CDEF called _AGGREGATE SIMILAR Items and create _AGGREGATE SIMILAR Items, Average. See CDEFs.jpg.

2. Edit your aggregate CPU graph, and chage the CDEF on each item of totals like on total.jpg

3. The graph shold see like on grafica.jpg.

I hope this could help anyone that has the same situation like me...

five0.4tluv · Post by **five0.4tluv** » Thu Jul 30, 2009 12:09 pm

I've got the plugin showing up in PIA mgmt. I go to install and I get this

Notice: Undefined index: keys in /var/www/cacti/lib/plugins.php on line 128

Warning: Invalid argument supplied for foreach() in /var/www/cacti/lib/plugins.php on line 128

Notice: Undefined index: keys in /var/www/cacti/lib/plugins.php on line 128

Warning: Invalid argument supplied for foreach() in /var/www/cacti/lib/plugins.php on line 128

V 0.8.7d PIA 2.4

five0.4tluv · Post by **five0.4tluv** » Thu Jul 30, 2009 12:59 pm

I was using the other package not 0.71

works good now, just need to play with each option to get what I want working.

phandung · Post by **phandung** » Mon Sep 14, 2009 10:53 pm

I done!

DigiSage · Post by **DigiSage** » Mon Oct 05, 2009 6:53 pm

I'm trying to make a single graph that shows the average CPU utilization for my entire cluster of 32 nodes. I figured I could use Aggregate to make a graph using the "Total" value shown on the standard ucd/net CPU Usage graph from each node.

Eventually I realized that Aggregate can only use actual data sources and "Total" is not a data source. Unfortunately, I was unable to find a single data source that showed total CPU usage. The best I was able to find was a source for System CPU Usage and User CPU Usage. However, I'd need the graph to treat the sum of them as a single data source, because System+User = actual CPU utilization.

Suppose I have 2 hosts and I have System-Host1, User-Host1, System-Host2, User-Host2 in one graph, and that on each system, User CPU usage is 100% and System is 0%. The displayed overall CPU average would be 50% when in actuality it's 100%. So as you can see, this won't work. As I was unable to literally combine two data sources into a single "virtual" data source, I gave up on this path.

Next, I also graph each CPU individually. For each host, I have a single graph that shows CPU usage for each of the 8 CPUs per host. So, in effect, I have a single data source per CPU in each host.

But this is 8 data sources per host, and the aggregate contains 32 hosts... Graphing this actually works, however it results in a PNG that takes a full 15 seconds to render, contains nearly 1000 items and is over 4000 pixels long! Obviously, this also isn't a viable solution.

So, what options are there to do this? I don't even need every single host listed in the graph legend... simply a single plot that shows average CPU usage across all those nodes. What's the best way to go about doing this?

Post by **gandalf** » Tue Oct 06, 2009 3:25 pm

This was already requested and is on the feature list.
But please verify, if totalling of that many datasources will perform faster: I'm not sure that will work.
Reinhard

DigiSage · Post by **DigiSage** » Tue Oct 06, 2009 3:33 pm

I just gave up and hacked together my own way of doing it... I made a python script that queries usage (via snmpv3 bulk walking) for each CPU of each node and writes the mean to stdout. Script runtime (over 31 hosts) is around 4 to 6 seconds total. Just got Cacti working with it and it seems go be working great.

It's kind of lame, because it's querying data that cacti is already aware of somewhere, I just don't know how to go about accessing it. But, until then, this works.

dononeil · Post by **dononeil** » Tue Oct 06, 2009 6:29 pm

I'm having problems with the stacks/lines for the individual items in the aggregate not showing up properly as + and - around 0.... The total graphs fine (inbound above 0, outbound below 0), but not the individual items.

See the attached graph. My original graphs work fine showing + and - for in / out.

Any ideas what I did wrong here, or what needs to be fixed? Each output item has the inversion cdef applied like the manual states.

Post by **gandalf** » Sun Oct 25, 2009 1:57 pm

I just received a different report of a possible bug that looks quite similar. It's on my list, now (was on vacation)
R.

Cacti

Aggregate V0.71 - Released

solved

error on install

fixed it

Re: fixed it

Who is online