I thought that would prevent search & replication messages from cluttering splunkd.log while site1 was unavailable, but I am getting this combination of warnings about 1500 times an hour:
WARN CMMessages - got genid thats invalid or out of range, setting to INVALID_GENID, jn=18446744073709551616.000000
WARN CMSlave - handleHeartbeatDone: successful heartbeat. Forcing a re-add, Reason="Proxy is in disconnect state."
The system also seems to think that the indexes are not fully searchable, but I think that there is a primary bucket for everything the remaining site2.