Hi,
On 3/29/23 2:09 AM, Michael Paquier wrote:
On Tue, Mar 28, 2023 at 05:43:26PM +0900, Kyotaro Horiguchi wrote:
No. Fine by me, except that "block read requests" seems to suggest
kernel read() calls, maybe because it's not clear whether "block"
refers to our buffer blocks or file blocks to me.. If it is generally
clear, I'm fine with the proposal.
Okay. Would somebody like to draft a patch?
Please find a draft attached.
Regards,
--
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com
diff --git a/doc/src/sgml/monitoring.sgml b/doc/src/sgml/monitoring.sgml
index c809ff1ba4..ae5d9a0226 100644
--- a/doc/src/sgml/monitoring.sgml
+++ b/doc/src/sgml/monitoring.sgml
@@ -5724,8 +5724,10 @@ SELECT pid, wait_event_type, wait_event FROM
pg_stat_activity WHERE wait_event i
<returnvalue>bigint</returnvalue>
</para>
<para>
- Returns the number of buffers fetched for table or index, in the
current
- transaction.
+ Returns the number of block read requests for table or index, in the
+ current transaction. This number minus <function>
+ pg_stat_get_xact_blocks_hit</function> gives the number of kernel
+ <function>read()</function> calls.
</para></entry>
</row>
@@ -5738,8 +5740,9 @@ SELECT pid, wait_event_type, wait_event FROM
pg_stat_activity WHERE wait_event i
<returnvalue>bigint</returnvalue>
</para>
<para>
- Returns the number of buffer hits for table or index, in the current
- transaction.
+ Returns the number of block read requests for table or index, in the
+ current transaction, found in cache (not triggering kernel
+ <function>read()</function> calls).
</para></entry>
</row>
