From 8dd4c7991c53a21645709e79524de3f4ed5c0b49 Mon Sep 17 00:00:00 2001 From: Michael Paquier Date: Wed, 9 Dec 2015 15:31:03 +0900 Subject: [PATCH] Ensure consistent on-disk state of UNLOGGED indexes at recovery Unlogged relation indexes need to have a consistent initial state on-disk at the time of replay to ensure that their replayed pages are found on disk should end of recovery happen and subsequently reset those relations. This commit extends the XLOG record XLOG_FPI with a set of flags aimed at controlling flushing of the page replayed. All types of relation indexes whose persistence is unlogged are impacted by the bug this commit fixes, with various degrees of problems, most of them causing errors on promoted standbys when trying to INSERT new tuples to their parent relations. The worst problem found was with GIN indexes, where trying to insert a new tuple in it caused the system to remain stuck on a semaphore lock, making the system unresponsive. --- src/backend/access/brin/brin.c | 2 +- src/backend/access/brin/brin_pageops.c | 2 +- src/backend/access/gin/gininsert.c | 4 ++-- src/backend/access/gist/gist.c | 2 +- src/backend/access/heap/rewriteheap.c | 6 ++++-- src/backend/access/nbtree/nbtree.c | 2 +- src/backend/access/nbtree/nbtsort.c | 3 ++- src/backend/access/rmgrdesc/xlogdesc.c | 10 +++++++++- src/backend/access/spgist/spginsert.c | 6 +++--- src/backend/access/transam/xlog.c | 17 +++++++++++++++++ src/backend/access/transam/xloginsert.c | 26 ++++++++++++++++++-------- src/backend/commands/tablecmds.c | 15 +++++++++++---- src/backend/commands/vacuumlazy.c | 2 +- src/backend/storage/buffer/bufmgr.c | 23 +++++++++++++++++++++++ src/include/access/xlog_internal.h | 7 +++++++ src/include/access/xloginsert.h | 6 ++++-- src/include/storage/bufmgr.h | 1 + 17 files changed, 106 insertions(+), 28 deletions(-) diff --git a/src/backend/access/brin/brin.c b/src/backend/access/brin/brin.c index 99337b0..c00791f 100644 --- a/src/backend/access/brin/brin.c +++ b/src/backend/access/brin/brin.c @@ -682,7 +682,7 @@ brinbuildempty(PG_FUNCTION_ARGS) brin_metapage_init(BufferGetPage(metabuf), BrinGetPagesPerRange(index), BRIN_CURRENT_VERSION); MarkBufferDirty(metabuf); - log_newpage_buffer(metabuf, false); + log_newpage_buffer(metabuf, false, true); END_CRIT_SECTION(); UnlockReleaseBuffer(metabuf); diff --git a/src/backend/access/brin/brin_pageops.c b/src/backend/access/brin/brin_pageops.c index f876f62..572fe20 100644 --- a/src/backend/access/brin/brin_pageops.c +++ b/src/backend/access/brin/brin_pageops.c @@ -865,7 +865,7 @@ brin_initialize_empty_new_buffer(Relation idxrel, Buffer buffer) page = BufferGetPage(buffer); brin_page_init(page, BRIN_PAGETYPE_REGULAR); MarkBufferDirty(buffer); - log_newpage_buffer(buffer, true); + log_newpage_buffer(buffer, true, false); END_CRIT_SECTION(); /* diff --git a/src/backend/access/gin/gininsert.c b/src/backend/access/gin/gininsert.c index 49e9185..b2e6b29 100644 --- a/src/backend/access/gin/gininsert.c +++ b/src/backend/access/gin/gininsert.c @@ -454,10 +454,10 @@ ginbuildempty(PG_FUNCTION_ARGS) START_CRIT_SECTION(); GinInitMetabuffer(MetaBuffer); MarkBufferDirty(MetaBuffer); - log_newpage_buffer(MetaBuffer, false); + log_newpage_buffer(MetaBuffer, false, true); GinInitBuffer(RootBuffer, GIN_LEAF); MarkBufferDirty(RootBuffer); - log_newpage_buffer(RootBuffer, false); + log_newpage_buffer(RootBuffer, false, true); END_CRIT_SECTION(); /* Unlock and release the buffers. */ diff --git a/src/backend/access/gist/gist.c b/src/backend/access/gist/gist.c index 53bccf6..fd53268 100644 --- a/src/backend/access/gist/gist.c +++ b/src/backend/access/gist/gist.c @@ -84,7 +84,7 @@ gistbuildempty(PG_FUNCTION_ARGS) START_CRIT_SECTION(); GISTInitBuffer(buffer, F_LEAF); MarkBufferDirty(buffer); - log_newpage_buffer(buffer, true); + log_newpage_buffer(buffer, true, true); END_CRIT_SECTION(); /* Unlock and release the buffer */ diff --git a/src/backend/access/heap/rewriteheap.c b/src/backend/access/heap/rewriteheap.c index 6a6fc3b..e9a9a8f 100644 --- a/src/backend/access/heap/rewriteheap.c +++ b/src/backend/access/heap/rewriteheap.c @@ -336,7 +336,8 @@ end_heap_rewrite(RewriteState state) MAIN_FORKNUM, state->rs_blockno, state->rs_buffer, - true); + true, + false); RelationOpenSmgr(state->rs_new_rel); PageSetChecksumInplace(state->rs_buffer, state->rs_blockno); @@ -685,7 +686,8 @@ raw_heap_insert(RewriteState state, HeapTuple tup) MAIN_FORKNUM, state->rs_blockno, page, - true); + true, + false); /* * Now write the page. We say isTemp = true even if it's not a diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index cf4a6dc..d211a98 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -206,7 +206,7 @@ btbuildempty(PG_FUNCTION_ARGS) (char *) metapage, true); if (XLogIsNeeded()) log_newpage(&index->rd_smgr->smgr_rnode.node, INIT_FORKNUM, - BTREE_METAPAGE, metapage, false); + BTREE_METAPAGE, metapage, false, true); /* * An immediate sync is required even if we xlog'd the page, because the diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c index f95f67a..faf611c 100644 --- a/src/backend/access/nbtree/nbtsort.c +++ b/src/backend/access/nbtree/nbtsort.c @@ -277,7 +277,8 @@ _bt_blwritepage(BTWriteState *wstate, Page page, BlockNumber blkno) if (wstate->btws_use_wal) { /* We use the heap NEWPAGE record type for this */ - log_newpage(&wstate->index->rd_node, MAIN_FORKNUM, blkno, page, true); + log_newpage(&wstate->index->rd_node, MAIN_FORKNUM, blkno, page, + true, false); } /* diff --git a/src/backend/access/rmgrdesc/xlogdesc.c b/src/backend/access/rmgrdesc/xlogdesc.c index 83cc9e8..389341e 100644 --- a/src/backend/access/rmgrdesc/xlogdesc.c +++ b/src/backend/access/rmgrdesc/xlogdesc.c @@ -77,10 +77,18 @@ xlog_desc(StringInfo buf, XLogReaderState *record) appendStringInfoString(buf, xlrec->rp_name); } - else if (info == XLOG_FPI || info == XLOG_FPI_FOR_HINT) + else if (info == XLOG_FPI_FOR_HINT) { /* no further information to print */ } + else if (info == XLOG_FPI) + { + xl_restore_fpi xlrec; + + memcpy(&xlrec, rec, sizeof(xl_restore_fpi)); + appendStringInfo(buf, "flush=%s", + xlrec.is_flush ? "true" : "false"); + } else if (info == XLOG_BACKUP_END) { XLogRecPtr startpoint; diff --git a/src/backend/access/spgist/spginsert.c b/src/backend/access/spgist/spginsert.c index bceee8d..0758bfd 100644 --- a/src/backend/access/spgist/spginsert.c +++ b/src/backend/access/spgist/spginsert.c @@ -173,7 +173,7 @@ spgbuildempty(PG_FUNCTION_ARGS) (char *) page, true); if (XLogIsNeeded()) log_newpage(&index->rd_smgr->smgr_rnode.node, INIT_FORKNUM, - SPGIST_METAPAGE_BLKNO, page, false); + SPGIST_METAPAGE_BLKNO, page, false, true); /* Likewise for the root page. */ SpGistInitPage(page, SPGIST_LEAF); @@ -183,7 +183,7 @@ spgbuildempty(PG_FUNCTION_ARGS) (char *) page, true); if (XLogIsNeeded()) log_newpage(&index->rd_smgr->smgr_rnode.node, INIT_FORKNUM, - SPGIST_ROOT_BLKNO, page, true); + SPGIST_ROOT_BLKNO, page, true, true); /* Likewise for the null-tuples root page. */ SpGistInitPage(page, SPGIST_LEAF | SPGIST_NULLS); @@ -193,7 +193,7 @@ spgbuildempty(PG_FUNCTION_ARGS) (char *) page, true); if (XLogIsNeeded()) log_newpage(&index->rd_smgr->smgr_rnode.node, INIT_FORKNUM, - SPGIST_NULL_BLKNO, page, true); + SPGIST_NULL_BLKNO, page, true, true); /* * An immediate sync is required even if we xlog'd the pages, because the diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 86debf4..819c6f3 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -9396,9 +9396,26 @@ xlog_redo(XLogReaderState *record) * when checksums are enabled. There is no difference in handling * XLOG_FPI and XLOG_FPI_FOR_HINT records, they use a different info * code just to distinguish them for statistics purposes. + * + * XLOG_FPI Records flagged with 'is_flush' indicate that the page + * immediately needs to be flushed to disk, not just to shared buffers. + * This is important if the on-disk state is to be the authoritative, + * not the state in shared buffers. E.g. because on-disk files may later + * be copied directly. */ if (XLogReadBufferForRedo(record, 0, &buffer) != BLK_RESTORED) elog(ERROR, "unexpected XLogReadBufferForRedo result when restoring backup block"); + + if (info == XLOG_FPI) + { + xl_restore_fpi xlrec; + + memcpy(&xlrec, XLogRecGetData(record), sizeof(xl_restore_fpi)); + + if (xlrec.is_flush) + FlushSingleBuffer(buffer); + } + UnlockReleaseBuffer(buffer); } else if (info == XLOG_BACKUP_END) diff --git a/src/backend/access/transam/xloginsert.c b/src/backend/access/transam/xloginsert.c index 925255f..1689329 100644 --- a/src/backend/access/transam/xloginsert.c +++ b/src/backend/access/transam/xloginsert.c @@ -884,9 +884,9 @@ XLogSaveBufferForHint(Buffer buffer, bool buffer_std) int flags; char copied_buffer[BLCKSZ]; char *origdata = (char *) BufferGetBlock(buffer); - RelFileNode rnode; - ForkNumber forkno; - BlockNumber blkno; + RelFileNode rnode; + ForkNumber forkno; + BlockNumber blkno; /* * Copy buffer so we don't have to worry about concurrent hint bit or @@ -932,19 +932,26 @@ XLogSaveBufferForHint(Buffer buffer, bool buffer_std) * If the page follows the standard page layout, with a PageHeader and unused * space between pd_lower and pd_upper, set 'page_std' to TRUE. That allows * the unused space to be left out from the WAL record, making it smaller. + * + * If 'is_flush' is set to TRUE, relation will be flushed to the OS + * immediately after replaying the record. */ XLogRecPtr log_newpage(RelFileNode *rnode, ForkNumber forkNum, BlockNumber blkno, - Page page, bool page_std) + Page page, bool page_std, bool is_flush) { - int flags; - XLogRecPtr recptr; + int flags; + XLogRecPtr recptr; + xl_restore_fpi xlrec; flags = REGBUF_FORCE_IMAGE; if (page_std) flags |= REGBUF_STANDARD; + xlrec.is_flush = is_flush; + XLogBeginInsert(); + XLogRegisterData((char *) &xlrec, sizeof(xlrec)); XLogRegisterBlock(0, rnode, forkNum, blkno, page, flags); recptr = XLogInsert(RM_XLOG_ID, XLOG_FPI); @@ -969,9 +976,12 @@ log_newpage(RelFileNode *rnode, ForkNumber forkNum, BlockNumber blkno, * If the page follows the standard page layout, with a PageHeader and unused * space between pd_lower and pd_upper, set 'page_std' to TRUE. That allows * the unused space to be left out from the WAL record, making it smaller. + * + * If 'is_flush' is set to TRUE, relation will be requested to flush + * immediately its data at replay after replaying this full page image. */ XLogRecPtr -log_newpage_buffer(Buffer buffer, bool page_std) +log_newpage_buffer(Buffer buffer, bool page_std, bool is_flush) { Page page = BufferGetPage(buffer); RelFileNode rnode; @@ -983,7 +993,7 @@ log_newpage_buffer(Buffer buffer, bool page_std) BufferGetTag(buffer, &rnode, &forkNum, &blkno); - return log_newpage(&rnode, forkNum, blkno, page, page_std); + return log_newpage(&rnode, forkNum, blkno, page, page_std, is_flush); } /* diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 0ddde72..dbe7ed9 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -9892,9 +9892,14 @@ copy_relation_data(SMgrRelation src, SMgrRelation dst, /* * We need to log the copied data in WAL iff WAL archiving/streaming is - * enabled AND it's a permanent relation. + * enabled AND it's a permanent relation. Unlogged relations need to have + * their INIT_FORKNUM logged as well, and flushed at replay to ensure a + * consistent on-disk state when reset at the end of recovery. */ - use_wal = XLogIsNeeded() && relpersistence == RELPERSISTENCE_PERMANENT; + use_wal = XLogIsNeeded() && + (relpersistence == RELPERSISTENCE_PERMANENT || + (relpersistence == RELPERSISTENCE_UNLOGGED && + forkNum == INIT_FORKNUM)); nblocks = smgrnblocks(src, forkNum); @@ -9917,10 +9922,12 @@ copy_relation_data(SMgrRelation src, SMgrRelation dst, /* * WAL-log the copied page. Unfortunately we don't know what kind of a * page this is, so we have to log the full page including any unused - * space. + * space. For the same reason, pages part of INIT_FORKNUM are always + * forcibly flushed at replay. */ if (use_wal) - log_newpage(&dst->smgr_rnode.node, forkNum, blkno, page, false); + log_newpage(&dst->smgr_rnode.node, forkNum, blkno, page, + false, forkNum == INIT_FORKNUM); PageSetChecksumInplace(page, blkno); diff --git a/src/backend/commands/vacuumlazy.c b/src/backend/commands/vacuumlazy.c index 2429889..b0e3901 100644 --- a/src/backend/commands/vacuumlazy.c +++ b/src/backend/commands/vacuumlazy.c @@ -736,7 +736,7 @@ lazy_scan_heap(Relation onerel, LVRelStats *vacrelstats, */ if (RelationNeedsWAL(onerel) && PageGetLSN(page) == InvalidXLogRecPtr) - log_newpage_buffer(buf, true); + log_newpage_buffer(buf, true, false); PageSetAllVisible(page); visibilitymap_set(onerel, blkno, buf, InvalidXLogRecPtr, diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c index 63188a3..73ddaff 100644 --- a/src/backend/storage/buffer/bufmgr.c +++ b/src/backend/storage/buffer/bufmgr.c @@ -2988,6 +2988,29 @@ FlushDatabaseBuffers(Oid dbid) } } +/* --------------------------------------------------------------------- + * FlushSingleBuffer + * + * This function flushes buffer already pinned and locked, be it + * in shared or exclusive mode to the OS. + * -------------------------------------------------------------------- + */ +void +FlushSingleBuffer(Buffer buffer) +{ + BufferDesc *bufHdr; + + /* currently not needed, but no fundamental reason not to support */ + Assert(!BufferIsLocal(buffer)); + Assert(BufferIsPinned(buffer)); + + bufHdr = GetBufferDescriptor(buffer - 1); + + LWLockHeldByMe(bufHdr->content_lock); + + FlushBuffer(bufHdr, NULL); +} + /* * ReleaseBuffer -- release the pin on a buffer */ diff --git a/src/include/access/xlog_internal.h b/src/include/access/xlog_internal.h index 86b532d..aa574a8 100644 --- a/src/include/access/xlog_internal.h +++ b/src/include/access/xlog_internal.h @@ -226,6 +226,13 @@ typedef struct xl_restore_point char rp_name[MAXFNAMELEN]; } xl_restore_point; +/* log of full-page write */ +typedef struct xl_restore_fpi +{ + bool is_flush; /* image to be flushed immediately to disk + * after replay */ +} xl_restore_fpi; + /* End of recovery mark, when we don't do an END_OF_RECOVERY checkpoint */ typedef struct xl_end_of_recovery { diff --git a/src/include/access/xloginsert.h b/src/include/access/xloginsert.h index 31b45ba..491caa5 100644 --- a/src/include/access/xloginsert.h +++ b/src/include/access/xloginsert.h @@ -53,8 +53,10 @@ extern void XLogResetInsertion(void); extern bool XLogCheckBufferNeedsBackup(Buffer buffer); extern XLogRecPtr log_newpage(RelFileNode *rnode, ForkNumber forkNum, - BlockNumber blk, char *page, bool page_std); -extern XLogRecPtr log_newpage_buffer(Buffer buffer, bool page_std); + BlockNumber blk, char *page, bool page_std, + bool is_flush); +extern XLogRecPtr log_newpage_buffer(Buffer buffer, bool page_std, + bool is_flush); extern XLogRecPtr XLogSaveBufferForHint(Buffer buffer, bool buffer_std); extern void InitXLogInsert(void); diff --git a/src/include/storage/bufmgr.h b/src/include/storage/bufmgr.h index 0f59201..a823508 100644 --- a/src/include/storage/bufmgr.h +++ b/src/include/storage/bufmgr.h @@ -176,6 +176,7 @@ extern void CheckPointBuffers(int flags); extern BlockNumber BufferGetBlockNumber(Buffer buffer); extern BlockNumber RelationGetNumberOfBlocksInFork(Relation relation, ForkNumber forkNum); +extern void FlushSingleBuffer(Buffer buffer); extern void FlushRelationBuffers(Relation rel); extern void FlushDatabaseBuffers(Oid dbid); extern void DropRelFileNodeBuffers(RelFileNodeBackend rnode, -- 2.6.3