1
0
mirror of https://github.com/postgres/postgres.git synced 2025-09-02 04:21:28 +03:00

Adjust btree index build to not use shared buffers, thereby avoiding the

locking conflict against concurrent CHECKPOINT that was discussed a few
weeks ago.  Also, if not using WAL archiving (which is always true ATM
but won't be if PITR makes it into this release), there's no need to
WAL-log the index build process; it's sufficient to force-fsync the
completed index before commit.  This seems to gain about a factor of 2
in my tests, which is consistent with writing half as much data.  I did
not try it with WAL on a separate drive though --- probably the gain would
be a lot less in that scenario.
This commit is contained in:
Tom Lane
2004-06-02 17:28:18 +00:00
parent 4d0e47d5a9
commit 2095206de1
8 changed files with 304 additions and 214 deletions

View File

@@ -11,7 +11,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/storage/smgr/smgr.c,v 1.72 2004/05/31 20:31:33 tgl Exp $
* $PostgreSQL: pgsql/src/backend/storage/smgr/smgr.c,v 1.73 2004/06/02 17:28:18 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@@ -48,6 +48,7 @@ typedef struct f_smgr
BlockNumber (*smgr_nblocks) (SMgrRelation reln);
BlockNumber (*smgr_truncate) (SMgrRelation reln, BlockNumber nblocks,
bool isTemp);
bool (*smgr_immedsync) (SMgrRelation reln);
bool (*smgr_commit) (void); /* may be NULL */
bool (*smgr_abort) (void); /* may be NULL */
bool (*smgr_sync) (void); /* may be NULL */
@@ -57,7 +58,8 @@ typedef struct f_smgr
static const f_smgr smgrsw[] = {
/* magnetic disk */
{mdinit, NULL, mdclose, mdcreate, mdunlink, mdextend,
mdread, mdwrite, mdnblocks, mdtruncate, NULL, NULL, mdsync
mdread, mdwrite, mdnblocks, mdtruncate, mdimmedsync,
NULL, NULL, mdsync
}
};
@@ -582,6 +584,34 @@ smgrtruncate(SMgrRelation reln, BlockNumber nblocks, bool isTemp)
return newblks;
}
/*
* smgrimmedsync() -- Force the specified relation to stable storage.
*
* Synchronously force all of the specified relation down to disk.
*
* This is really only useful for non-WAL-logged index building:
* instead of incrementally WAL-logging the index build steps,
* we can just write completed index pages to disk with smgrwrite
* or smgrextend, and then fsync the completed index file before
* committing the transaction. (This is sufficient for purposes of
* crash recovery, since it effectively duplicates forcing a checkpoint
* for the completed index. But it is *not* workable if one wishes
* to use the WAL log for PITR or replication purposes.)
*
* The preceding writes should specify isTemp = true to avoid
* duplicative fsyncs.
*/
void
smgrimmedsync(SMgrRelation reln)
{
if (! (*(smgrsw[reln->smgr_which].smgr_immedsync)) (reln))
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not sync relation %u/%u: %m",
reln->smgr_rnode.tblNode,
reln->smgr_rnode.relNode)));
}
/*
* smgrDoPendingDeletes() -- Take care of relation deletes at end of xact.
*/