Cisco ACE Load Balancer

[tcolgate]

Hi,

I made some updates to the ace templates here to support Real Server monitoring, and to convert them to work as a script server (this was a massive performance boost for me).

It's the first time I've exported a template and scripts, but I hope it's of some use.

[Crazyachmed]

@tcolgate: thanks, great work! the templates seem to work fine. i'll take a closer look at them in some days, when the graphs contain some data.



i have an issue with the connection graphs (see screenshot). how do i get rid of the m,k etc. units? i already changed all values in the graph template to "exact numbers" which works only for the numeric values in the legend. same problem before updating to tcolgate's templates.

[tcolgate]

i have an issue with the connection graphs (see screenshot). how do i get rid of the m,k etc. units? i already changed all values in the graph template to "exact numbers" which works only for the numeric values in the legend. same problem before updating to tcolgate's templates.

Setting "Unit Exponent Value" to zero on the graph template should do what you want.

[Crazyachmed]

ah, thanks again

[Crazyachmed]

we have this application running over a bazillion ports on four ace clusters. i needed a better way to monitor the load than the ~400 vserver graphs per cluster, so i created a new graph: concurrent connections per system (context)

the value is actually a bit more than concurrent client connections, i think the ace also counts the probe connections onto this value. you can get the same value on the cli with "sh stats conn" or "sh conn count". in my case i have about 80k connections that include ~3k other connections

at the moment i do not remember the command my colleague used to display the counter for the different connections types. i'll ask him later.

this template only includes the new graph

[mxwood]

Has anyone seen this issue with this template? I am able to get the graphs for the interfaces and vlans for the ACE but not the vserver connections. when I debug the graph I get the following

RRDTool Command:

/usr/bin/rrdtool graph - \
--imgformat=PNG \
--start=-86400 \
--end=-300 \
--title="*************************" \
--base=1000 \
--height=120 \
--width=500 \
--alt-autoscale-max \
--lower-limit=0 \
--vertical-label="Connections" \
--slope-mode \
--font TITLE:12: \
--font AXIS:8: \
--font LEGEND:10: \
--font UNIT:8: \
DEF:a="/var/lib/cacti/rra/phx-ace_hctotal_228.rrd":NumberOfC:AVERAGE \
DEF:b="/var/lib/cacti/rra/phx-ace_hctotal_228.rrd":NumberOfC:MAX \
CDEF:cdefe=a,UN,INF,UNKN,IF \
AREA:a#157419FF:"Current Connections" \
GPRINTLAST:"Current\:%8.2lf %s" \
GPRINTAVERAGE:"Average\:%8.2lf %s" \
GPRINTMAX:"Maximum\:%8.2lf %s\n" \
AREA:cdefe#8F9286FF:""
RRDTool Says:

OK

[ybkevin]

if so, via oid? template?

[mxwood]

when I run the following command from the command line I do get an output
/usr/bin/php -q /usr/share/cacti/site/scripts/ss_ace_vserver.php xx.xxx.xxx.1 33 2:161:500:snmpstring:::MD5::[None] NumberOfC 1.44


this command returns a value for 1109 which somewhat compairs to the value of the total number of the rserver connections, is there a way to see what the graph data really is or has been put into the history?

Thanks for your help

Update---
the log file shows the following
2009 12:50:08 PM - CMDPHP: Poller[0] DEBUG: SQL Exec: "insert into poller_output (local_data_id, rrd_name, time, output) values (227, 'NumberOfC', '2009-10-02 12:50:07', 'U')"
10/02/2009 12:50:08 PM - CMDPHP: Poller[0] Host[33] DS[227] CMD: /usr/share/cacti/site/scripts/ss_ace_vserver.php xx.xx.xx.xxx 2:161:500:snmpstring:::MD5::[None] get NumberOfC 1.43, output: U
10/02/2009 12:50:08 PM - CMDPHP: Poller[0] Host[33] DS[227] WARNING: Result from CMD not valid. Partial Result:




Update: Well looks like I got the data graphing now... however, can anyone explain why I have values such as 491.87 as a current connection value? how do I have a fraction of a connection?

Thanks again

[tcolgate]

I had a problem with the original template due to the way the indexes work out. A VServer name (or rserver) name alone can't make up the index as it might not be unique, so I added the actual full (oid style) index, so that gets picked as it will always be unique. It looks ugly in the list, but it works.

I'm not sure about your floating point values for the server connections. The stuff you get out of rrd is averaged, so that woud make sense, but you should get a float value returned from the snmp query (that wouldn't make any sense at all).

[ScOp3]

First of all thanks for the host template. Unfortunately i have a slight problem with the two php scripts being executed. We are using SNMP v3 and the password passed on to the script contains special characters. Executing the script therefore fails from within cacti.

The debug query output looks like this.

Code: Select all

Executing script query '/usr/bin/php5 -q /srv/www/htdocs/cacti/scripts/ss_ace_rserver.php 1.2.3.4 65 3:161:500::User:PassB%fhG8&tword:MD5::[None] query RealName'

As you might see the sample password contains % and & characters. If i change the password to something like "passw0rd" the script runs fine.

Does anyone have a clue how to adjust the underlying ss_ace_rserver.php and ss_ace_vserver.php scripts to escape the special characters or print the password quoted?

Thanks for reading

ScOp3

[tcolgate]

That looks like it could possibly be a slightly wider issue in cacti. You coudl try quoting |host_snmp_password| in the XML templates, perhaps that might help? It's not a problem with the scripts (the issue occurs before the script runs), it's how cacti goes about launching it.
The XMl tell cacti how to launch it, but I'm not sure how much control you really have.

Good lcuk

[ScOp3]

That looks like it could possibly be a slightly wider issue in cacti. You coudl try quoting |host_snmp_password| in the XML templates, perhaps that might help? It's not a problem with the scripts (the issue occurs before the script runs), it's how cacti goes about launching it.
The XMl tell cacti how to launch it, but I'm not sure how much control you really have.

Good lcuk

Hey,

thanks for a reply. I created a workaround using a simple password but i agree that this is possibly related to the way cacti hands over the arguments/variables.

Anyhow i just noticed, that the latest ACE update broke the templates (A2(2.3)). You had any luck in fixing this issue so far?

Regards

ScOp3

[tcolgate]

I am running 2.3 here and it's still working for me (A2_2_3)

[DervMan]

I've successfully imported the 'cacti_host_template_cisco_ace_load_balancer.xml' but when trying to import either the 'ace_rserver.xml' or 'ace_rserver.xml' templates I don't get the confirmation screen after the import.

I can see the data templates under 'Data Templates' but they don't appear in the 'Associated Graph Templates' when I try to attach them to a host. I do see the 'Cisco ACE CPU' but nothing else, I'm guessing this is because the templates aren't importing correctly.

Any ideas how I can see what's going wrong?

[ringram]

Hi Folks-

I'm afraid I'm not having any luck with this template. Does anyone have this running against an ACE 4710, Version A3(2.4)?

When I run the data query for vservers, I only get one of my server farms. And the rserver query doesn't return anything. Did Cisco change their OIDs? I'm running Cacti 0.8.7e

Best regards,
Rich