We are seeing inconsistent among clients ones one client update a file(by scp), some of the node see new contents but some of the nodes don't.   The inconsistent can last 30mins to a few hours and fix by its own.  I think it should because some of the node not dropping the page cache properly.

     Looking into the code I see when Fc cap revoke,  fuse client drop objectcache , queue a task to finisher thread to do fuse_lowlevel_notify_inval_inode,  then ack the cap revoke.  So seems there is a window between the cap-revoke-ack , and the final fuse_lowlevel_notify_inval_inode finished, in this window page cache still valid and user can read stale data.    Though it is strange that the window can be that large(no pg issue during the window).

     Could you please confirm if this is the real problems and why it is implemented in this way?