many backends hang on MultiXactOffsetSLRU

    we have a Postgresql v14.8 database, almost thousands of backends hang on MultiXactOffsetSLRU at the same time, all of these sessions running same query "SELECT ....", from OS and postgresql slow log, we found all of these query on "BIND" stage.

 LOG:  duration: 36631.688 ms  bind S_813: SELECT
LOG:  duration: 36859.786 ms  bind S_1111: SELECT
LOG:  duration: 35868.148 ms  bind <unnamed>: SELECT
LOG:  duration: 36906.471 ms  bind <unnamed>: SELECT
LOG:  duration: 35955.489 ms  bind <unnamed>: SELECT
LOG:  duration: 36833.510 ms  bind <unnamed>: SELECT
LOG:  duration: 36839.535 ms  bind S_1219: SELECT

...

this database hang on MultiXactOffsetSLRU and MultiXactOffsetBuffer long time.  

could you direct me why they are hanging on 'BIND‘ stage with MultiXactOffsetSLRU ? 

Thanks,

James