Neophyte
ProfessorI haven't touch integrations in years and I no longer have access to the system with this setup, but if I recall correctly... What we did was have the ticketing/glue system immediately ACK all alerts once it generates a ticket. LM chains/rules were setup that if the alert was not ACK within a short time it will fall into another stage which would notify the team directly on alerts, which would let us know that the integration was broken (and still allow alerts to be worked). Kinda a indirect way to determine it.
Yes we do something similar. Our escalation chain posts first to the AI alert grouping tool (PagerDuty). If the grouping tool does not ACK all the alerts, the second stage sends directly to our ticketing system (no AI grouping of alerts). We have a 20 minute delay because we prefer grouping. That's also why we monitor the integration logs.
