unignoring resources

John Mazzitelli mazz at redhat.com
Mon Apr 8 16:33:38 UTC 2013


would like opinions on this before I implement anything.

As part of BZ RHQ-1, I have successfully implemented the ability to not only ignore resource types, but resources as well.

If you recall, we already had the ability to ignore top-level servers that have not yet been committed to inventory (if you go to the discovery queue, you'll see the "ignore" button).

Now we have the ability to ignore any resource that has been committed to inventory. See https://bugzilla.redhat.com/show_bug.cgi?id=535289 for more.

But the question becomes, what happens if a person wants to "bring back" those ignored resources? That is "unignore" a resource.

Well, previously, you could have done this in the discovery queue. Had you ignored resources sitting in the discovery queue, you could have pressed the "unignore" button and that resource would be put back into the NEW state and thus you'd see it in the discovery queue again (allowing an admin user to commit it into inventory).

We now also have the ability to "unignore" any resource that is ignored (that is, even those that were committed before but then ignored).

How it works is identical to how it worked with resources in the discovery queue that were ignored - that is, the "unignored" resource is put into the NEW state (and, if its a top level server, it becomes visible again in the discovery queue).

NOTE THE DIFFERENCE! The new feature allows you to ignore COMMITTED resources (those resources that are in inventory and being managed) - when you unignore, THEY DO NOT GO BACK INTO THE COMMITTED STATE. They go back into the NEW state. This means they will show up again, but in the discovery queue - they need to be committed back into inventory.

What are the implications? It means TOP LEVEL server resources that were committed but ignored and then unignored will show up in the discovery queue and will need to be recommitted to inventory by an admin user. If the resource to be unignored is anything other than a top-level server, then it should eventually get automatically committed (I haven't tested this, but it is probably when the agent runs a server-scan discovery).

The alternative is to somehow remember that an ignored resource was committed (remember, we could have always ignored resources in the NEW state, not committed state) and if it was committed when it was ignored, put it back into the committed state.

The former is already done (because it just uses the same API that the discovery queue uses to ignore/unignore). The latter would require additional work (we'd have to somehow figure out a resource was committed before and if so, automatically commit it back into inventory).


More information about the rhq-devel mailing list