Thold - Suggest to monitor DS from 2 devices (HA system)

[alexf]

Dear All,

Requirements:
a) 2 F5 load balancers running as HA system (Active/Passive Mode)
b) Want to monitor the server connection of the HA as a whole
c) Provide Threshold monitoring of "sum of 2 datasource" from 2 devices

Software Installed
a) Cacti 0.8.7b (in Ubuntu 8.0.4 LTS)
b) Aggregate 0.66
c) Thold 0.3.9
d) PA 2.1

All the Plugins are working fine.

Procedure of my Setup
I am using the F5 server connections as the monitoring example
a) Create Data template for Server Connection of F5 using SNMP query
d) Create Graph Template for F5 Server Connection
c) Create 2 Devices - the Primary F5 and Secondary F5
d) Create Graph using the Graph Template for the Primary
and Secondary F5

e) Use Aggregate plugin to Create Graph for Server Connection on 2 Devices
- First DS refer to the Primary F5 server connection
- Second DS refer to the Secondary F5 server connection
- Third CDEF (Sum of Data Source, Don't Duplicate) of the above DS

f) Convert the above "Aggregate graph" to "Graph Template"



Problem in Thold Plugin
a) As the enabling of Thold threshold is in the definition (graph creation) page, I need to add the "Aggregate Graph" in the device definition of F5 device (Primary and Secondary)

b) The monitoring parameter is a CDEF of Datasource from 2 devices, I cannot make the "Threshold" setup. Why? See point c below:

c) For example, I have tried to add the graph template to primary device, but the definition of graph will remove the DS from the secondary device. All DS not belong to this device will be removed.

Thus, the threshold will only monitor either the primary or secondary device. Not the "SUM" of both devices.


d) I know the Thold plugin can handle the CDEF (with patch), but it work for DS from SINGLE device, not from 2 devices.


This prevents me from monitoring the HA as a whole.

For example, if the primary device is fail-over to the secondary device successfully, it will not cause any Interruption to the service as a whole.

I can monitor the "SUM OF Connection" to show this status and this will NOT trigger Alert (as the service is still available).
Of course, the alert for the Primary failure should be fired, but of lower severity.



Suggestion
My Suggestion is to allow THOLD to monitor any DS from Aggregated Graph rather than bound it only to a DEVICE definition page.

As the DS in a Aggregated graph has already been bounded to a device, the DS will be updated by its corresponding device. If Thold can monitor the DS (either native or CDEF DS) of a aggregated graph, then it is very easy to monitor any HA system


Conclusion:
With THOLD plugin capable to monitor DS from more than 1 device, this makes it suitable to monitor any HA system.


Or is there any method/suggestion/experience that can monitor a HA system, please kindly to help me.


Thanks in advance for any advice.


Best Regards,

Alex