When working with users new to ProSphere I often find there is a common misunderstanding of the function of Discovery Jobs. Those who have used EMC ControlCenter in the past, often think of Discovery Jobs as being synonymous with ControlCenter Discovery Policies. However, this is not quite the case. I prefer to explain ProSphere Discovery Jobs as "Detection Jobs".
When you first run a Discovery Job, ProSphere will run a detection using the access credentials and scope (IP addresses) you defined in the Discovery Job. So for an array, we would reach out to the SMI provider(s) specified in the Discovery Job and detect the arrays that are in the environment. When an array is first detected, we then also perform a discovery of the array. However, after this initial discovery, the Discovery Job will no longer be of value in discovering changes to the array.
After the initial discovery, ProSphere relies on three mechanisms for rediscovering the array.
So what then is the function of the Discovery Job Polling Interval? This interval defines how frequently ProSphere will detect new objects in the environment. So if you have specified an IP range in your Discovery Job that includes an entire subnet, and you are trying to detect physical hosts, it makes sense to set this to once a day. It is not uncommon for new hosts to come online in your environment. However, for arrays, chances are you will know when one is added to your data center, and it probably does not happen very often. So you could setup that Discovery Job without a schedule at all, and only run it when you have added a new array, or you could set it to once per month. The same holds true for ESX hosts, you could probably set those to be discovered once a week. Any new VMs created on your ESX hosts will be discovered by the event based rediscovery I mentioned above.
It is also import to understand the function of the Discovery Job when looking at the Discovery Log at the bottom of the Discovery Jobs page. Here is an example of a log that confuses new users.
The second row shows 34 Fully Discovered Objects, but the Status is Failed? And the first row shows zeroes across the board, but the Status is Success? This log is telling you that on April 15 at 8:17 AM, 34 new object were discovered. However, because there were some Topology Detection Failures and Performance Detection Failures, ProSphere marks the Status as Failed. These failures may not be much of a concern to you, it is best to click the number under each column to find out more details. Meanwhile, the first row is telling you that later on April 15, the discovery job was run again, but this time no new objects were detected. I also know that on this day, the Discovery Job IP range was modified to exclude the previously failed objects (they were not important to us), that is why at 8:33 AM, there are no failures reported. So despite nothing new being found, the Discovery Job is considered a Success.
I hope this helps clear up any confusion you may have had working with ProSphere Discovery Jobs. Please let me know if you have any further questions or comments.