Issue Summary

This can be an Improvement Request, but due to the Discovery Import behavior - this actually becomes a bug:
Insight discovery import is being triggered by an external application: Discovery Tool or Discovery Collector.

The collector may have n files to export at one time, and m at another, depending on the time of its polling...
If I have a large number of Scan Settings executed / collected from various Discovery Instances and agents - Setting the Threshold by number of executions / synchronizations does not make sense, and it is better to set the Threshold by a number of days, so an object will be considered as Missing if not synchronized in x amount of days.

In DC Insight Import configuration one can select between Days and Synchronizations for the Threshold count.

This is reproducible on Data Center: NO

Steps to Reproduce

1. configure Discovery to scan an IP  in one scan setting, transferring the results to the ../discovery/import directory (or a "Dropbox" shared folder for testing)
2. Configure a one Discovery agents to send the Scans to the same directory (e.g on a Laptop, and you can use SFTP or File Copy with a "Dropbox" shared location for testing)
3. Set the Discovery Collector to fetch files from the ../discovery/import directory and upload to Cloud every 10m
4. In Discovery Import (Jira Cloud) - set Missing Objects (Update) with threshold set to 2 Executions, as we are expecting 2 files to be imported in total: 1st Disco Scan, and 2nd the Agent's.
5. If the Laptop (agent) is offline in time for its next scan - the Collector may upload / export the next Disco Scan before getting the Agent's file to export

Expected Results

As long as both scans are uploaded within a given time - there is no effect on Hosts / objects not included in each of the imported file

Actual Results

The second import of the Discovery tool will render the Agent as not found.

This seems manageable with only 2 files to upload, but not if you have multiple scan settings' results to upload, as well as multiple Agents sending their XMLs to the Collector...

...

Workaround

Set the missing Objects Threshold to a very high number of executions, but this may still result in fluctuating statuses / objects - as the files to import are not being queued in order...
Once a full cycle of Scanning>Importing is done - the Status will no longer fluctuate, and will only change after some time if a Host is not found consecutively.