Can I pass custom variables to a Cacti script data query?

[MisterJingo]

Hello,

I am creating a Cacti script data query and I would like to pass some custom host specific (i.e. per host) variables to the script, e.g. username and password. They are separate to the SNMP variables.

I have followed the excellent Script Data Query walkthrough, and successfully created and tested a script and XML file. I now just want to know if I can pass a couple of extra variables to the script.

These variables are used by the script to authenticate to the host during its indexed GET queries. Ideally I would like the username and password to be saved along with the other host properties on the Cacti Devices screen, but I have seen no way of adding custom fields to this screen so I don't hold much hope for that possibility.

I have considered (and remain open to) alternative methods of getting the information to the script, e.g.
- I cannot hijack (i.e. re-purpose) the |snmp_variables| or |host_variables|, they are all needed.
- I could get the administrator to enter the username and password for each host into a file on the Cacti server, and have the script read them from there, but this is inconvenient and I really would rather maintain this information through the Cacti WebUI.

Can anyone help please?

[BSOD2600]

You can pass custom variables, for example look at any one of the Data Input Methods scripts. specify the variables by brackets. The Data Template will then have a Custom Data section at the bottom, to which you check the "Use Per-Data Source Value". These topics are covered in http://docs.cacti.net/

[MisterJingo]

Thanks, I saw the section on Creating a Data Input Method, but thought that the part about specifying per data source variables inside <>'s in the input string applied only to Script/Command type Data Input Methods, and not to Get Script Data (Indexed) Data Queries (which is the sort of thing I think I need to be doing). Perhaps I misunderstood something.

The Perl script I wrote uses a mixture of SNMP and HTTP-requests to query the host and retrieve per interface statistics from the host's GUI, e.g.
Interface 1 unicast rx:100 tx:5,
Interface 2 unicast rx:250 tx:45,
Interface 3 dropped rx:50 tx:0,
Interface 4 utilization rx:60 tx:40,
etc

Most of the interface activity types that my script queries are counters, but some return a percentage value, e.g. utilization.

Here is some sample output:

Code: Select all

> query.pl 10.0.0.5 admin mypass HTTP public 2c 161 2500 10 "" "" "" "" "" "" get dropped 11A
rx:0 tx:2975618

It is just a script however, and I can always change it.

For each interface I want to create a separate graph for the different activity types (i.e. counters), e.g. unicast, multicast, dropped, utilization, other, etc. Each graph will show rx and tx data. For example, host-1 will have unicast, multicast and dropped graphs (each showing rx and tx data) for interfaces 1A to 12B.

It seems to me like I need to use a Get Script Data (Indexed) Data Query for something like this, and now I just want to know if I can pass a couple of extra variables to the script somehow. Again, perhaps I've missed something.

I have tried setting up an experimental Data Input Method with the input string shown below and the associated input fields.

Code: Select all

perl <path_cacti>/scripts/query.pl <host> <gui_u> <gui_p> <proto> <community> <version> <port> <timeout> <oids> "<username>" "<password>" "<auth_proto>" "<priv_pass>" "<priv_proto>" get unicast <int>

Using this method, I think that for each interface on the host I would need to create a new Data Input Method for each of the 12 different activity types and then an accompanying data template and graph template. If I understand this correctly, a Get Script Data (Indexed) Data Query seems a more appropriate way of getting the data.

Any other advice you can provide will be greatly appreciated.

Thank you.

[BSOD2600]

Using this method, I think that for each interface on the host I would need to create a new Data Input Method for each of the 12 different activity types and then an accompanying data template and graph template. If I understand this correctly, a Get Script Data (Indexed) Data Query seems a more appropriate way of getting the data..

Exactly, as it automates a lot of the otherwise manual parameter filling for each data source. Of course, an indexed script query takes a lot more time to develop but is worth it in the end, especially if you want to automate its deployment to numerous devices.