I think this also reiterates the very important need for some sort of trace system for event processing / mapping, so we can "step through" a test event and our mappings / transforms. I imagine something very like a programming debugger or the like. -- James Pulver Information Technology Area Supervisor LEPP Computer Group Cornell University
jcurry wrote, On 9/4/2009 12:13 PM: > I have just proved that cascading event processing did indeed arrive with 2.4 > (strictly it's 2.4.1 that I am using). However I cannot find any > documentation for how it works and certainly no examples as suggested by Matt! > > There are 2 different aspects to consider; event class transforms and event > class mapping transforms. > > Pre 2.4, I believe it was relatively simple. Either a single class transform > was used (if no event mapping took place ie event generated by Zenoss > daemon), or an event mapping transform took place, in which case any class > transform for the resulting class was ignored. If a class transform was > used, it was simply a single transform - no cascading hierarchy of transforms. > > Now life is much more complicated. Ignore class mapping transforms for a > moment. If you have a class hierarchy /Skills/S1/S2 and each of these has a > transform, then, for an event of class /Skills/S1/S2 they are all applied, as > Matt says, starting with the Skills transform, then the S1, then the S2. If > a different private field is defined in each, then an event of class > /Skills/S1/S2 will end up with all the private fields, a /Skills/S1 event > will end up with those fields from /Skills and S1 but not S2, etc. If the > same private field is defined in all the transforms, then the most specific > wins ie. S2. > > Now introduce event class mapping transforms. In the past, if a mapping > transform was used, any class transform was ignored. Now, BOTH the event > class mapping and the event class transforms seem to be applied. Any private > field that appears in both, the mapping transform takes precedence - OK so > far. > > However, I find that a mapping transform may use a private field from the > class transform - so the order is class then mapping??? > > But I also find that a class transform can make use of a private field from a > mapping transform - so the order is mapping then class??? > > So, questions: > 1) Please where is there some good documentation on the cascading event > transforms that arrived in 2.4 but didn't even get a mention in the Release > Notes? > 2) What is the order of processing when you have both mapping transforms and > class transforms? > 3) Or is there something really clever goes on at database insertions time > where all this gets sorted out? > > Cheers, > Jane > > >> Matt Ray wrote, On 4/10/2009 4:11 PM: >> Quote: >> There are some changes being made to Event Transforms for 2.4. We're >> working on getting some examples added to the documentation and I >> don't think they're all in the beta yet. One of the changes will be >> cascading event transforms. Event transforms are applied down the >> event hierarchy, so that on an incoming /Business/Service/Bus event, >> the /Business transform would be applied, then the /Service >> transform, then the /Bus transform. This sounds like what you want. >> >> Thanks, >> Matt Ray >> Zenoss Community Manager >> community.zenoss.com >> [email protected] >> > > > > > > -------------------- m2f -------------------- > > Read this topic online here: > http://forums.zenoss.com/viewtopic.php?p=38958#38958 > > -------------------- m2f -------------------- > > > > _______________________________________________ > zenoss-users mailing list > [email protected] > http://lists.zenoss.org/mailman/listinfo/zenoss-users _______________________________________________ zenoss-users mailing list [email protected] http://lists.zenoss.org/mailman/listinfo/zenoss-users
