From: Todd, C. J <cha...@in...> - 2004-08-23 17:10:45
|
All: I have started my review of the HPI B.01.01 MIB with the Notifications and Eventing parts. Given that I am by no means an SNMP expert, I figured I'd start a quick dialog on the intended usage model, and then supply my HPI-specific comments from there. =20 I see that there are Notifications defined for: Sensor Notifications, SensorEventChange Notifications, Resource Notifications, Domain Notifications, Watchdog Notifications, HotSwap Notifications, Software Notifications, OEM Notifications, and User Notifications. These map nicely to the types of events generated by an HPI implementation. But, in looking at the Notification definitions, I see that they do not contain all of the information that an HPI event would have. (Refer to line number 7076 and beyond.) =20 There are also Event Tables defined. Refer to line number 2436 and beyond. It appears that there is an overall "Event Master table", and individual tables defined for: Resource Events, Domain Events, Sensor Events, Sensor Enable Change Events, Hot Swap Events, Watchdog Events, Software Events, OEM Events, and User Events. Again, these map nicely to the various HPI Event Types. After a quick glance, it appears that the information that is missing from the Notifications is contained in the event entries in these tables. (I do owe a more thorough review of this, however.) =20 So, I am concluding that the intended usage model is for an SNMP manager to get a Notification Trap asynchronously, and then to read (poll) the associated entry in the appropriate Event Table to find out more information. (I've seen this notify/poll model used before in SNMP, so I think this is a normal practice.) Am I correct in this assumption? =20 If this is the case, I have a few further questions: (1) Is there a notion that the SNMP Agent will read the HPI Event Queue via saHpiEventGet() and "store" those events that it reads in these event tables? If so, how long are these event entries kept around? If not, then how does the agent get the event entry information when queried for it at a later time (perhaps via the Event Log)? =20 (2) What happens if, say, two Sensor Notifications are sent to a manager? How can it get more information on a particular notification? Does it have to read both entries in the table, and do some sort of correlation of the information that it reads with the information that it received in the Trap? The information sent in the notification does not seem to match the INDEX for the sensor event table. (This holds for the other types of notifications, as well.) (3) What happens if there are multiple Trap receivers? =20 =20 =20 Thanks, Charlene =20 =20 =20 The content of this message is my personal opinion only and although I am an employee of Intel, the statements I make here in no way represent Intel's position on the issue, nor am I authorized to speak on behalf of Intel on this matter.=20 =20 |