Skip to content

Commit debcec7

Browse files
robertmhaasrobertmhaas
committed
Include the backend ID in the relpath of temporary relations.
This allows us to reliably remove all leftover temporary relation files on cluster startup without reference to system catalogs or WAL; therefore, we no longer include temporary relations in XLOG_XACT_COMMIT and XLOG_XACT_ABORT WAL records. Since these changes require including a backend ID in each SharedInvalSmgrMsg, the size of the SharedInvalidationMessage.id field has been reduced from two bytes to one, and the maximum number of connections has been reduced from INT_MAX / 4 to 2^23-1. It would be possible to remove these restrictions by increasing the size of SharedInvalidationMessage by 4 bytes, but right now that doesn't seem like a good trade-off. Review by Jaime Casanova and Tom Lane.
1 parent 3f9479e commit debcec7

File tree

37 files changed

+668
-342
lines changed

37 files changed

+668
-342
lines changed

‎doc/src/sgml/storage.sgml‎

Lines changed: 15 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
<!-- $PostgreSQL: pgsql/doc/src/sgml/storage.sgml,v 1.32 2010/02/16 22:34:43 tgl Exp $ -->
1+
<!-- $PostgreSQL: pgsql/doc/src/sgml/storage.sgml,v 1.33 2010/08/13 20:10:50 rhaas Exp $ -->
22

33
<chapter id="storage">
44

@@ -133,16 +133,20 @@ there.
133133
</para>
134134

135135
<para>
136-
Each table and index is stored in a separate file, named after the table
137-
or index's <firstterm>filenode</> number, which can be found in
138-
<structname>pg_class</>.<structfield>relfilenode</>. In addition to the
139-
main file (a/k/a main fork), each table and index has a <firstterm>free space
140-
map</> (see <xref linkend="storage-fsm">), which stores information about free
141-
space available in the relation. The free space map is stored in a file named
142-
with the filenode number plus the suffix <literal>_fsm</>. Tables also have a
143-
<firstterm>visibility map</>, stored in a fork with the suffix
144-
<literal>_vm</>, to track which pages are known to have no dead tuples.
145-
The visibility map is described further in <xref linkend="storage-vm">.
136+
Each table and index is stored in a separate file. For ordinary relations,
137+
these files are named after the table or index's <firstterm>filenode</> number,
138+
which can be found in <structname>pg_class</>.<structfield>relfilenode</>. But
139+
for temporary relations, the file name is of the form
140+
<literal>t<replaceable>BBB</>_<replaceable>FFF</></>, where <replaceable>BBB</>
141+
is the backend ID of the backend which created the file, and <replaceable>FFF</>
142+
is the filenode number. In either case, in addition to the main file (a/k/a
143+
main fork), each table and index has a <firstterm>free space map</> (see <xref
144+
linkend="storage-fsm">), which stores information about free space available in
145+
the relation. The free space map is stored in a file named with the filenode
146+
number plus the suffix <literal>_fsm</>. Tables also have a
147+
<firstterm>visibility map</>, stored in a fork with the suffix <literal>_vm</>,
148+
to track which pages are known to have no dead tuples. The visibility map is
149+
described further in <xref linkend="storage-vm">.
146150
</para>
147151

148152
<caution>

‎src/backend/access/heap/visibilitymap.c‎

Lines changed: 2 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@
88
*
99
*
1010
* IDENTIFICATION
11-
* $PostgreSQL: pgsql/src/backend/access/heap/visibilitymap.c,v 1.10 2010/04/23 23:21:44 rhaas Exp $
11+
* $PostgreSQL: pgsql/src/backend/access/heap/visibilitymap.c,v 1.11 2010/08/13 20:10:50 rhaas Exp $
1212
*
1313
* INTERFACE ROUTINES
1414
* visibilitymap_clear - clear a bit in the visibility map
@@ -373,8 +373,7 @@ visibilitymap_truncate(Relation rel, BlockNumber nheapblocks)
373373
}
374374

375375
/* Truncate the unused VM pages, and send smgr inval message */
376-
smgrtruncate(rel->rd_smgr, VISIBILITYMAP_FORKNUM, newnblocks,
377-
rel->rd_istemp);
376+
smgrtruncate(rel->rd_smgr, VISIBILITYMAP_FORKNUM, newnblocks);
378377

379378
/*
380379
* We might as well update the local smgr_vm_nblocks setting. smgrtruncate

‎src/backend/access/nbtree/nbtsort.c‎

Lines changed: 3 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -59,7 +59,7 @@
5959
* Portions Copyright (c) 1994, Regents of the University of California
6060
*
6161
* IDENTIFICATION
62-
* $PostgreSQL: pgsql/src/backend/access/nbtree/nbtsort.c,v 1.125 2010/04/28 16:10:40 heikki Exp $
62+
* $PostgreSQL: pgsql/src/backend/access/nbtree/nbtsort.c,v 1.126 2010/08/13 20:10:50 rhaas Exp $
6363
*
6464
*-------------------------------------------------------------------------
6565
*/
@@ -295,9 +295,8 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno)
295295
}
296296

297297
/*
298-
* Now write the page. We say isTemp = true even if it's not a temp
299-
* index, because there's no need for smgr to schedule an fsync for this
300-
* write; we'll do it ourselves before ending the build.
298+
* Now write the page. There's no need for smgr to schedule an fsync for
299+
* this write; we'll do it ourselves before ending the build.
301300
*/
302301
if (blkno == wstate->btws_pages_written)
303302
{

‎src/backend/access/transam/twophase.c‎

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -7,7 +7,7 @@
77
* Portions Copyright (c) 1994, Regents of the University of California
88
*
99
* IDENTIFICATION
10-
* $PostgreSQL: pgsql/src/backend/access/transam/twophase.c,v 1.62 2010/07/06 19:18:55 momjian Exp $
10+
* $PostgreSQL: pgsql/src/backend/access/transam/twophase.c,v 1.63 2010/08/13 20:10:50 rhaas Exp $
1111
*
1212
* NOTES
1313
* Each global transaction is associated with a global transaction
@@ -865,8 +865,8 @@ StartPrepare(GlobalTransaction gxact)
865865
hdr.prepared_at = gxact->prepared_at;
866866
hdr.owner = gxact->owner;
867867
hdr.nsubxacts = xactGetCommittedChildren(&children);
868-
hdr.ncommitrels = smgrGetPendingDeletes(true, &commitrels, NULL);
869-
hdr.nabortrels = smgrGetPendingDeletes(false, &abortrels, NULL);
868+
hdr.ncommitrels = smgrGetPendingDeletes(true, &commitrels);
869+
hdr.nabortrels = smgrGetPendingDeletes(false, &abortrels);
870870
hdr.ninvalmsgs = xactGetCommittedInvalidationMessages(&invalmsgs,
871871
&hdr.initfileinval);
872872
StrNCpy(hdr.gid, gxact->gid, GIDSIZE);
@@ -1320,13 +1320,13 @@ FinishPreparedTransaction(const char *gid, bool isCommit)
13201320
}
13211321
for (i = 0; i < ndelrels; i++)
13221322
{
1323-
SMgrRelation srel = smgropen(delrels[i]);
1323+
SMgrRelation srel = smgropen(delrels[i], InvalidBackendId);
13241324
ForkNumber fork;
13251325

13261326
for (fork = 0; fork <= MAX_FORKNUM; fork++)
13271327
{
13281328
if (smgrexists(srel, fork))
1329-
smgrdounlink(srel, fork, false, false);
1329+
smgrdounlink(srel, fork, false);
13301330
}
13311331
smgrclose(srel);
13321332
}

‎src/backend/access/transam/xact.c‎

Lines changed: 10 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -10,7 +10,7 @@
1010
*
1111
*
1212
* IDENTIFICATION
13-
* $PostgreSQL: pgsql/src/backend/access/transam/xact.c,v 1.297 2010/08/13 15:42:21 rhaas Exp $
13+
* $PostgreSQL: pgsql/src/backend/access/transam/xact.c,v 1.298 2010/08/13 20:10:50 rhaas Exp $
1414
*
1515
*-------------------------------------------------------------------------
1616
*/
@@ -912,15 +912,14 @@ RecordTransactionCommit(void)
912912
TransactionId latestXid = InvalidTransactionId;
913913
int nrels;
914914
RelFileNode *rels;
915-
bool haveNonTemp;
916915
int nchildren;
917916
TransactionId *children;
918917
int nmsgs = 0;
919918
SharedInvalidationMessage *invalMessages = NULL;
920919
bool RelcacheInitFileInval = false;
921920

922921
/* Get data needed for commit record */
923-
nrels = smgrGetPendingDeletes(true, &rels, &haveNonTemp);
922+
nrels = smgrGetPendingDeletes(true, &rels);
924923
nchildren = xactGetCommittedChildren(&children);
925924
if (XLogStandbyInfoActive())
926925
nmsgs = xactGetCommittedInvalidationMessages(&invalMessages,
@@ -1048,7 +1047,7 @@ RecordTransactionCommit(void)
10481047
* asynchronous commit if all to-be-deleted tables are temporary though,
10491048
* since they are lost anyway if we crash.)
10501049
*/
1051-
if (XactSyncCommit || forceSyncCommit || haveNonTemp)
1050+
if (XactSyncCommit || forceSyncCommit || nrels > 0)
10521051
{
10531052
/*
10541053
* Synchronous commit case:
@@ -1334,7 +1333,7 @@ RecordTransactionAbort(bool isSubXact)
13341333
xid);
13351334

13361335
/* Fetch the data we need for the abort record */
1337-
nrels = smgrGetPendingDeletes(false, &rels, NULL);
1336+
nrels = smgrGetPendingDeletes(false, &rels);
13381337
nchildren = xactGetCommittedChildren(&children);
13391338

13401339
/* XXX do we really need a critical section here? */
@@ -4474,15 +4473,15 @@ xact_redo_commit(xl_xact_commit *xlrec, TransactionId xid, XLogRecPtr lsn)
44744473
/* Make sure files supposed to be dropped are dropped */
44754474
for (i = 0; i < xlrec->nrels; i++)
44764475
{
4477-
SMgrRelation srel = smgropen(xlrec->xnodes[i]);
4476+
SMgrRelation srel = smgropen(xlrec->xnodes[i], InvalidBackendId);
44784477
ForkNumber fork;
44794478

44804479
for (fork = 0; fork <= MAX_FORKNUM; fork++)
44814480
{
44824481
if (smgrexists(srel, fork))
44834482
{
44844483
XLogDropRelation(xlrec->xnodes[i], fork);
4485-
smgrdounlink(srel, fork, false, true);
4484+
smgrdounlink(srel, fork, true);
44864485
}
44874486
}
44884487
smgrclose(srel);
@@ -4579,15 +4578,15 @@ xact_redo_abort(xl_xact_abort *xlrec, TransactionId xid)
45794578
/* Make sure files supposed to be dropped are dropped */
45804579
for (i = 0; i < xlrec->nrels; i++)
45814580
{
4582-
SMgrRelation srel = smgropen(xlrec->xnodes[i]);
4581+
SMgrRelation srel = smgropen(xlrec->xnodes[i], InvalidBackendId);
45834582
ForkNumber fork;
45844583

45854584
for (fork = 0; fork <= MAX_FORKNUM; fork++)
45864585
{
45874586
if (smgrexists(srel, fork))
45884587
{
45894588
XLogDropRelation(xlrec->xnodes[i], fork);
4590-
smgrdounlink(srel, fork, false, true);
4589+
smgrdounlink(srel, fork, true);
45914590
}
45924591
}
45934592
smgrclose(srel);
@@ -4661,7 +4660,7 @@ xact_desc_commit(StringInfo buf, xl_xact_commit *xlrec)
46614660
appendStringInfo(buf, "; rels:");
46624661
for (i = 0; i < xlrec->nrels; i++)
46634662
{
4664-
char *path = relpath(xlrec->xnodes[i], MAIN_FORKNUM);
4663+
char *path = relpathperm(xlrec->xnodes[i], MAIN_FORKNUM);
46654664

46664665
appendStringInfo(buf, " %s", path);
46674666
pfree(path);
@@ -4716,7 +4715,7 @@ xact_desc_abort(StringInfo buf, xl_xact_abort *xlrec)
47164715
appendStringInfo(buf, "; rels:");
47174716
for (i = 0; i < xlrec->nrels; i++)
47184717
{
4719-
char *path = relpath(xlrec->xnodes[i], MAIN_FORKNUM);
4718+
char *path = relpathperm(xlrec->xnodes[i], MAIN_FORKNUM);
47204719

47214720
appendStringInfo(buf, " %s", path);
47224721
pfree(path);

‎src/backend/access/transam/xlogutils.c‎

Lines changed: 8 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -11,7 +11,7 @@
1111
* Portions Copyright (c) 1996-2010, PostgreSQL Global Development Group
1212
* Portions Copyright (c) 1994, Regents of the University of California
1313
*
14-
* $PostgreSQL: pgsql/src/backend/access/transam/xlogutils.c,v 1.71 2010/07/08 16:08:30 tgl Exp $
14+
* $PostgreSQL: pgsql/src/backend/access/transam/xlogutils.c,v 1.72 2010/08/13 20:10:50 rhaas Exp $
1515
*
1616
*-------------------------------------------------------------------------
1717
*/
@@ -68,7 +68,7 @@ log_invalid_page(RelFileNode node, ForkNumber forkno, BlockNumber blkno,
6868
*/
6969
if (log_min_messages <= DEBUG1 || client_min_messages <= DEBUG1)
7070
{
71-
char *path = relpath(node, forkno);
71+
char *path = relpathperm(node, forkno);
7272

7373
if (present)
7474
elog(DEBUG1, "page %u of relation %s is uninitialized",
@@ -133,7 +133,7 @@ forget_invalid_pages(RelFileNode node, ForkNumber forkno, BlockNumber minblkno)
133133
{
134134
if (log_min_messages <= DEBUG2 || client_min_messages <= DEBUG2)
135135
{
136-
char *path = relpath(hentry->key.node, forkno);
136+
char *path = relpathperm(hentry->key.node, forkno);
137137

138138
elog(DEBUG2, "page %u of relation %s has been dropped",
139139
hentry->key.blkno, path);
@@ -166,7 +166,7 @@ forget_invalid_pages_db(Oid dbid)
166166
{
167167
if (log_min_messages <= DEBUG2 || client_min_messages <= DEBUG2)
168168
{
169-
char *path = relpath(hentry->key.node, hentry->key.forkno);
169+
char *path = relpathperm(hentry->key.node, hentry->key.forkno);
170170

171171
elog(DEBUG2, "page %u of relation %s has been dropped",
172172
hentry->key.blkno, path);
@@ -200,7 +200,7 @@ XLogCheckInvalidPages(void)
200200
*/
201201
while ((hentry = (xl_invalid_page *) hash_seq_search(&status)) != NULL)
202202
{
203-
char *path = relpath(hentry->key.node, hentry->key.forkno);
203+
char *path = relpathperm(hentry->key.node, hentry->key.forkno);
204204

205205
if (hentry->present)
206206
elog(WARNING, "page %u of relation %s was uninitialized",
@@ -276,7 +276,7 @@ XLogReadBufferExtended(RelFileNode rnode, ForkNumber forknum,
276276
Assert(blkno != P_NEW);
277277

278278
/* Open the relation at smgr level */
279-
smgr = smgropen(rnode);
279+
smgr = smgropen(rnode, InvalidBackendId);
280280

281281
/*
282282
* Create the target file if it doesn't already exist. This lets us cope
@@ -293,7 +293,7 @@ XLogReadBufferExtended(RelFileNode rnode, ForkNumber forknum,
293293
if (blkno < lastblock)
294294
{
295295
/* page exists in file */
296-
buffer = ReadBufferWithoutRelcache(rnode, false, forknum, blkno,
296+
buffer = ReadBufferWithoutRelcache(rnode, forknum, blkno,
297297
mode, NULL);
298298
}
299299
else
@@ -312,7 +312,7 @@ XLogReadBufferExtended(RelFileNode rnode, ForkNumber forknum,
312312
{
313313
if (buffer != InvalidBuffer)
314314
ReleaseBuffer(buffer);
315-
buffer = ReadBufferWithoutRelcache(rnode, false, forknum,
315+
buffer = ReadBufferWithoutRelcache(rnode, forknum,
316316
P_NEW, mode, NULL);
317317
lastblock++;
318318
}

0 commit comments

Comments
 (0)