From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Date: Mon, 3 Nov 2008 19:03:41 +0000 (+0000)
Subject: Reduce the acceptable staleness of pgstat data for autovacuum, per the
X-Git-Tag: recoveryinfrav9~431
X-Git-Url: http://git.postgresql.org/gitweb/irc:/static/gitweb.js?a=commitdiff_plain;h=52af005e8b029133455addd9a9d5ec85ddbad9c0;p=users%2Fsimon%2Fpostgres.git

Reduce the acceptable staleness of pgstat data for autovacuum, per the
longstanding note in the source that this patch removes.
---

diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c
index 5d1b80a86f..fee9d98158 100644
--- a/src/backend/postmaster/autovacuum.c
+++ b/src/backend/postmaster/autovacuum.c
@@ -2149,8 +2149,10 @@ do_autovacuum(void)
 		 * It could have changed if something else processed the table while
 		 * we weren't looking.
 		 *
-		 * FIXME we ignore the possibility that the table was finished being
-		 * vacuumed in the last 500ms (PGSTAT_STAT_INTERVAL).  This is a bug.
+		 * Note: we have a special case in pgstat code to ensure that the stats
+		 * we read are as up-to-date as possible, to avoid the problem that
+		 * somebody just finished vacuuming this table.  The window to the race
+		 * condition is not closed but it is very small.
 		 */
 		MemoryContextSwitchTo(AutovacMemCxt);
 		tab = table_recheck_autovac(relid, table_toast_map);
diff --git a/src/backend/postmaster/pgstat.c b/src/backend/postmaster/pgstat.c
index c3a767bd91..916a4c2e7d 100644
--- a/src/backend/postmaster/pgstat.c
+++ b/src/backend/postmaster/pgstat.c
@@ -3381,7 +3381,10 @@ backend_read_statsfile(void)
 	/*
 	 * We set the minimum acceptable timestamp to PGSTAT_STAT_INTERVAL msec
 	 * before now.  This indirectly ensures that the collector needn't write
-	 * the file more often than PGSTAT_STAT_INTERVAL.
+	 * the file more often than PGSTAT_STAT_INTERVAL.  In an autovacuum
+	 * worker, however, we want a lower delay to avoid using stale data, so we
+	 * use PGSTAT_RETRY_DELAY (since the number of worker is low, this
+	 * shouldn't be a problem).
 	 *
 	 * Note that we don't recompute min_ts after sleeping; so we might end up
 	 * accepting a file a bit older than PGSTAT_STAT_INTERVAL.  In practice
@@ -3389,8 +3392,12 @@ backend_read_statsfile(void)
 	 * PGSTAT_STAT_INTERVAL; and we don't want to lie to the collector about
 	 * what our cutoff time really is.
 	 */
-	min_ts = TimestampTzPlusMilliseconds(GetCurrentTimestamp(),
-										 -PGSTAT_STAT_INTERVAL);
+	if (IsAutoVacuumWorkerProcess())
+		min_ts = TimestampTzPlusMilliseconds(GetCurrentTimestamp(),
+											 -PGSTAT_RETRY_DELAY);
+	else
+		min_ts = TimestampTzPlusMilliseconds(GetCurrentTimestamp(),
+											 -PGSTAT_STAT_INTERVAL);
 
 	/*
 	 * Loop until fresh enough stats file is available or we ran out of time.