Hi Maryam, Just going to chime in here as I am want to do on occasion, more or less reiterating Alok’s answers but including long and tedious reasoning.
For events I tend to think the events should be reported at a relative level of importance from the reporting function. If my role in the system is to maintain a link and that link goes down this is a pretty big thing for me in my role. Aggregation/normalization of the severity of the event should then be handled by an aggregative function that understands the context of the notification in a broader context. I guess in my opinion if a link is down I would expect a switching function would be issuing a warning or error level event. The “collector” or host application of that function may adjust that level in the context of functional redundancy or aggregation. If the application is aware of the role of the switch in the overall system it may inform the switch of how to report these events, so configurability would seem a useful function to reduce event processing overhead. / Chris On 2016-11-08, 07:51, "[email protected] on behalf of Tahhan, Maryam" <[email protected] on behalf of [email protected]> wrote: Hi Folks For the OVS events collectd plugin we wrote as part of barometer, we are currently sending a notification message from collectd with a level that is set to WARNING for when the link status is detected to be down. So this means that when the status changes from "Unknown" to "down" when a new interface is added a warning message will be generated. So from a Doctor/VES perspective: 1. When do you expect to see a warning message vs an informational message? When the link status only changes from Up to down 2. Do you want to know when the status changes from unknown to down? 3. Should we allow the notification severity to be configurable for link status? Or should we from a collectd point of view provide you with informational messages only and let you decide a severity? BR Maryam _______________________________________________ opnfv-tech-discuss mailing list [email protected] https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss _______________________________________________ opnfv-tech-discuss mailing list [email protected] https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss
