now TID at RT and PQ index got managed by binlog, ie when daemon got started wo binlog TID and other index members related to it stay same. That is why such indexes got flushed on every invocation
SphinxQL FLUSH RTINDEX
rt_flush_period searchd section option
replication send index to cluster
even no transaction to the index being made.
With binlog enabled daemon TID got correctly increased and all such flushes passed quick to indexes there were no transactions.
now TID at RT and PQ index got managed by binlog, ie when daemon got started wo binlog TID and other index members related to it stay same. That is why such indexes got flushed on every invocation
FLUSH RTINDEXeven no transaction to the index being made.
With binlog enabled daemon TID got correctly increased and all such flushes passed quick to indexes there were no transactions.