During https://wikitech.wikimedia.org/wiki/Incident_documentation/2021-10-22_eqiad_return_path_timeouts an alert that pages fired, but did not actually make it to VictorOps in time because of the partial networking outage that triggered the alert in the first place. The alert did make it to IRC, which was thankfully enough to get SRE eyes on it.
One idea that @CDanis suggested is to use the OOB network as a backup for sending these alerts to VO.
There was also some weirdness with the "BGP status on cr2-eqiad" alert in icinga, that issued a recovery, but never recorded it being down/in problem state. The tentative theory is that was affected by the same networking issues.