I am trying to determine the behavior of a system using ZFS to back a PostgreSQL instance as it relates to fillfactor and table clustering.
ZFS implements copy-on-write, so when PostgreSQL modifies a block on disk, the filesystem writes a new block rather than updating the existing block. Unless I'm missing something, this would seem to negate the purpose of using any fillfactor less than 100% and reduce any benefits of using CLUSTER. In fact, when using ZFS snapshots, it would seem that CLUSTER would actually be wasteful.
Therefore one might posit that PostgreSQL should be configured to use 100% fillfactor and avoid clustering on ZFS. Can anyone comment on this?
