Count in Event Correlation

[sbianco1978]

Dears,

I need to create an Event Correlation rule that performs a sequence that group by for a value and meantime count for another value.

My scope is:

sequence by user.name with maxspan=10m and count county >= 2
sequence 1
sequence 2

I couldn't find any way to do it. Could anyone help me?

Thank you.

Best regards

[justin_ibarra]

Hi there ,

If I understand your use case correctly, you are not able to due this in a single rule, but you should be able to achieve it leveraging building block rules (BBR) and a threshold rule.

You would need to do this in a few parts:

• ruleA: create a threshold rule where county > 2 with a lookback (to - from) of 10m (as a BBR)
• ruleB: create a sequence with maxspan=10 and sequence1, sequence2 (as a BBR)
• create a sequence rule alerting on the 2 previous rule alerts:

sequence by user.name with maxspan=10m
  [any where event.kind == "signal" and kibana.alert.rule.rule_id == "<ruleA.rule_id>"]
  [any where event.kind == "signal" and kibana.alert.rule.rule_id == "<ruleB.rule_id>"]

You may have to play with timing of the 2 maxspans and the lookback to get it right, but this should achieve what you are trying to do. You also don't necessarily need to use BBR's for the 2 interim rules, but if you don't, then you will get 3 alerts every time this fully completes.

Hope this helps