Steps to reproduce:

1. Configure hosts to be monitored by proxy groups.
2. Do nothing.

Result:
Host interfaces randomly go to UNKNOWN state when Zabbix moves them between proxies.
Existing acknowledged/supressed problems are closed (see https://support.zabbix.com/browse/ZBX-24516) and new ones opened few seconds later, causing unwanted alerts.
Information about duration of "interface unreachable" problems is lost.

Expected:
Host interfaces stay green.

It is really annoying. Best workaround I know would require reworking templates to have different tag for each interface and then use tag-based maintenance.