Chained probes (calling probes from within other probes)

[[email protected]…]

I'd like to be able to call a probe from another probe. The driving factor for this is to reduce code duplication in my probes. An example:

I have a probe that uses dmidecode to determine what hardware manufacturer I'm currently running on. This sets a dynamic group that I use within my configuration (manuf-dell, for example).

I have a second probe that queries my RPM database looking for a specific rpm gpg key and sets a dynamic group if I don't find it (rpm_needs_dell_gpg). The problem is, I only want to set this group if I know I'm running on dell hardware.

I have three options for doing this:

1. One monolithic probe that combines these two options so I reduce code duplication. This is unappealing because this can get large quick (for one script).

2. Have my rpm_gpg probe issue the same dmidecode command as my manufacturer probe. This is unappealing because that means I have to update two places if I ever need to change the dmidecode call.

3. Move the manufacturer specific rpm gpg probes into the manufacturer probe. This is unappealing because that means I have to update two places if I ever need to change the rpm calls or methods I use to probe for gpg keys.

[desai]

I think that this could be better factored as a series of boolean flags (determined by probes) with a server side template that combines the conditions to achieve the desired result. This would result in probes that set groups like is_dell_hardware, and has_dell_gpgkey, which could be combined in an SGenshi template to have the desired result.