In the bucket of AWS S3, the host information of the clustering member is accumulated.
The clustering member host information is still not deleted even though added below S3_PING options and shutdown normally.

<property name="remove_old_coords_on_view_change">true</property>
<property name="remove_all_files_on_view_change">true</property>

S3_PING (and all other FILE_PING based implementations) write the full contents of the logical address cache, instead of just the current members. (1)

It's apparently causing an issue for a user because they're frequently changing members, and they're seeing timeout errors trying to connect to the old members that no longer exist.

It seems like JGroups upstream master code (2) already fixed.

(1) JGroups 3.6.12: https://github.com/belaban/JGroups/blob/3.6.12/src/org/jgroups/protocols/FILE_PING.java#L271-L284
(2) JGroups master: https://github.com/belaban/JGroups/blob/master/src/org/jgroups/protocols/FILE_PING.java#L283-L289