The documentation for pg_replication_slots previously mentioned only
max_slot_wal_keep_size as a condition under which the wal_status column
could show unreserved or lost. However, since commit 
be87200,
replication slots can also be invalidated due to horizon or wal_level,
and since commit 
ac0e33136ab, idle_replication_slot_timeout can also
trigger this state.
This commit updates the description of the wal_status column to
reflect that max_slot_wal_keep_size is not the only cause of the lost state.
Back-patched to v16, where the additional invalidation cases were introduced.
Author: Fujii Masao <masao.fujii@gmail.com>
Reviewed-by: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Reviewed-by: Nisha Moond <nisha.moond412@gmail.com>
Discussion: https://postgr.es/m/
78b34e84-2195-4f28-a151-
5d204a382fdd@oss.nttdata.com
Backpatch-through: 16
          <para>
           <literal>unreserved</literal> means that the slot no longer
           retains the required WAL files and some of them are to be removed at
-          the next checkpoint.  This state can return
+          the next checkpoint.  This typically occurs when
+          <xref linkend="guc-max-slot-wal-keep-size"/> is set to
+          a non-negative value.  This state can return
           to <literal>reserved</literal> or <literal>extended</literal>.
          </para>
         </listitem>
         <listitem>
          <para>
-          <literal>lost</literal> means that some required WAL files have
-          been removed and this slot is no longer usable.
+          <literal>lost</literal> means that this slot is no longer usable.
          </para>
         </listitem>
        </itemizedlist>
-       The last two states are seen only when
-       <xref linkend="guc-max-slot-wal-keep-size"/> is
-       non-negative.
       </para></entry>
      </row>