Auto-discovery of Postgres table resources

Thomas Segismont tsegismo at redhat.com
Wed Apr 9 09:08:28 UTC 2014


Le 09/04/2014 10:08, Libor Zoubek a écrit :
> Dne Tue, 08 Apr 2014 16:03:40 +0200 Thomas Segismont
> <tsegismo at redhat.com> napsal(a):
>
>> Agreed. I know we already discussed this but I'm not really satisfied
>> with the "Ignore resource" solution. So far, I see two possibilities:
>>  1. Very quick and dirty: add some properties in resource plugin
>> config to narrow down auto-discovery results. For example, we could
>> add a "tablesDiscoveryPattern" property in the database resource type,
>> which we would read in the table resource discovery component.
> I'd be for this type of solution, but as a general property of every
> resourceType. Something like ignoreChildPattern where user could set
> wildcard or strict match to resource type and resource name or key. This
> feature could go even further. Imagine "platform profile" be it a
> resource type hierarchy with overidden 'ignoreChildPattern's. As user
> assings his platform to profile he'll see exactly what he wants. For
> example we'd assign platform hosting RHQ server to a profile which would
> ignore lots of EJBs or PUs, because most users are not interested in
> those resources.

I like your idea of platform profiles. We need to keep in mind that a 
resource can have children of different types, so it's not really at the 
parent resource level that we should define what needs to be ignored.

>
>>  2. Introduce a new Facet "DiscoveryFilter" or similar.
>
> I think it also would help if rhq-plugin.xml could set resource types as
> ignored. User who wants to monitor tables, then manually enables that
> type and possibly ignores some tables on resource level.
>
>

If we follow the platform profiles path I think the agent plugin schema 
does not need to be updated.


More information about the rhq-devel mailing list