Q&A about how Discovery ages device nodes.
 

1) The Discovery datastore disk usage is too high. How to control when nodes are removed?

There are several configurable parameters in Administration -> Model Maintenance. These values can be reduced to make various nodes age more quickly. See https://docs.bmc.com/docs/display/DISCO121/Configuring+model+maintenance+settings for details. Note that lowering the aging parameters will reduce the data kept in the datastore, but only gradually, and will not reclaim filesystem space.

2) What are the options for reclaiming disk space allocated to the datastore?

The recommended approach is to compact the datastore on a regular basis. See https://docs.bmc.com/docs/display/DISCO121/tw_ds_offline_compact for details. Note that there needs to be enough free space in the datastore partition to compact the largest file, so this should be done before the datastore usage becomes too high. In certain cases, it may be necessary to reinitialize the datastore, but since this involves losing data and discovery task history, it should not be done without consulting support.  

3) The DDD (Directly Discovered Data) removal threshold was set to 7 days more than two weeks ago, but the number of hosts is not decreasing. Why?

Changing the DDD limit will reduce space used in the datastore by removing DDD, but it will not lower the number of hosts. The Host node (and other root nodes) are inferred nodes rather than DDD. The removal of Host nodes is controlled by the "device aging" parameters.

4) How does age_count get increased or decreased?

With each successful scan, age_count is increased by 1. When a scan fails, age_count is set to -1, and with every subsequent failed scan, age_count is decremented by 1.

When a decreasing age count reaches the "Device aging access failures" limit (-7 by default), Discovery will destroy the device node on the next scan, assuming the Device aging time limit (10 days by default) has also been reached. Both these limits must be exceeded for a device to be destroyed.  

If a device with a negative age_count is successfully scanned, then age_count is reset to +1.

5) How can a host be forced to age out?

To "force" the aging of a host, the options are to:
- Select the host and choose Actions > Destroy
OR
- Set the following aging parameters in Model Maintenance
(NOTE: This is a global setting and will impact aging of all devices):
Device aging time limit: 2 days
Device aging access failures: 1 failure
Number of credential failures that count as one failed access: 1 failure

After making these changes, scan the host without credentials (or with incorrect credentials) to produce a "No Access" result. This will decrement the age count and the host will be destroyed after 2 days.

6) How to change the Model Maintenance device aging time limits from the command line?  

It is possible to change these values of "Device aging access failures" and "Device aging time limit"  with the commands below:

MIN_FAILED_ACCESSES_BEFORE_DESTROY -> Device aging access failures
MIN_SECONDS_SINCE_ACCESS_SUCCESS_BEFORE_DESTROY -> Device aging time limit (must be set to a higher value than the Scan optimization setting)    

command: tw_options -u system MIN_FAILED_ACCESSES_BEFORE_DESTROY=<value of choice>

command: tw_options -u system MIN_SECONDS_SINCE_ACCESS_SUCCESS_BEFORE_DESTROY=<value of choice>

Please note that changing these values via the UI is the recommended option.

7) The Host node  Documentation says:

By default, a Host node will be automatically destroyed, if BOTH of the following circumstances are fulfilled:
* No update has been made in ten days from any of the IP addresses known for this host.
* No update has been made in the last seven scans on the IP address currently being scanned.

So what happens if the host is simply not scanned any longer?

It will stay in the datastore forever (i.e. age_count is never decremented).

8) The Host node  Documentation says "In this context, "seen" means anything other than no update". What is meant by "update"?

An 'update' occurs anytime there is a successful discovery access, even if none of the host attributes have changed.

9) Do Packages and Patches age like hosts?

No, Packages and Patches are globally shared across multiple hosts. See Package node and Patch node.

9) What happens with decommissioned or unreachable Hosts?

The designed behavior is that Discovery will create a NoResponse Discovery Access (DA) for a decommissioned or unreachable host. NoResponse DAs are counted as discovery failures (i.e. have a negative age count), and when the aging threshold is reached Discovery will age out and destroy the hosts automatically. 

10) How the aging happens for ESXi host which are implicitly scanned via vCenter IP ?

 The ESXi host will not be aged out, as they are not scanned explicitly. They are scanned via vCenter credentials implicitly, hence only option to remove the ESXi hosts which are residing incorrectly, is to manually destroy these host.