Probable cause and affected incidents
How ARGUNIS groups incidents that go down together and shows which one is the probable cause and which are affected.
On this page
When several related items fail at the same time, ARGUNIS doesn't notify you about each one: it marks one incident as the probable cause and the others as affected, and notifies you about the cause with the affected incidents in the same notification.
How it decides the probable cause
By inventory dependencies: if the items are related ("runs on" or "depends on") and go down within the correlation window, the lowest one in the chain is the probable cause. For example, the server under the database under the WebService. It works even if they are in different groups.
By inventory group: if there are no relationships, incidents that open in the same group close to when the group's server stopped responding hang from that server.
Relationships are added on each item's detail page (Relationships section, Add relationship button), ARGUNIS discovers them from the connections reported by agents or from the DNS of WebServices and TCP services, or learned correlation proposes them in Accept or dismiss proposed relationships.
Where you see it
In the incident list, an affected incident shows "derived from" with the cause's ticket, and the cause shows "probable cause of N incidents".
In an affected incident's detail, the Probable cause field and a box explain how it was reached (by dependencies or by group) and whether its notification is grouped into the cause's notification.
In the cause's detail, the Affected section lists the incidents that depend on it.
In an outage's detail, the Scenario before the outage (Cerebro) section shows the state of the previous hours saved by the Cerebro. If that scenario repeats, the Cerebro warns you before the next outage. See Precursor scenarios.
What happens when the cause is resolved
If the probable cause is resolved and an affected incident is still open, ARGUNIS notifies about it on its own (only once). If the chain has another open cause, the affected incident then hangs from that one.
On an item's detail page, Simulate outage shows you what would happen if that item goes down: which incidents would open, which would be the probable cause, and who would be notified. It doesn't create anything or notify anyone.