database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-08-28 18:48:04 +03:00
Code Activity

Compute XID horizon for page level index vacuum on primary.

Browse Source 
Previously the xid horizon was only computed during WAL replay. That
had two major problems:
1) It relied on knowing what the table pointed to looks like. That was
   easy enough before the introducing of tableam (we knew it had to be
   heap, although some trickery around logging the heap relfilenodes
   was required). But to properly handle table AMs we need
   per-database catalog access to look up the AM handler, which
   recovery doesn't allow.
2) Not knowing the xid horizon also makes it hard to support logical
   decoding on standbys. When on a catalog table, we need to be able
   to conflict with slots that have an xid horizon that's too old. But
   computing the horizon by visiting the heap only works once
   consistency is reached, but we always need to be able to detect
   conflicts.

There's also a secondary problem, in that the current method performs
redundant work on every standby. But that's counterbalanced by
potentially computing the value when not necessary (either because
there's no standby, or because there's no connected backends).

Solve 1) and 2) by moving computation of the xid horizon to the
primary and by involving tableam in the computation of the horizon.

To address the potentially increased overhead, increase the efficiency
of the xid horizon computation for heap by sorting the tids, and
eliminating redundant buffer accesses. When prefetching is available,
additionally perform prefetching of buffers.  As this is more of a
maintenance task, rather than something routinely done in every read
only query, we add an arbitrary 10 to the effective concurrency -
thereby using IO concurrency, when not globally enabled.  That's
possibly not the perfect formula, but seems good enough for now.

Bumps WAL format, as latestRemovedXid is now part of the records, and
the heap's relfilenode isn't anymore.

Author: Andres Freund, Amit Khandekar, Robert Haas
Reviewed-By: Robert Haas
Discussion:
    https://postgr.es/m/20181212204154.nsxf3gzqv3gesl32@alap3.anarazel.de
    https://postgr.es/m/20181214014235.dal5ogljs3bmlq44@alap3.anarazel.de
    https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
This commit is contained in:
Andres Freund
2019-03-26 14:41:46 -07:00
parent 126d631222
commit 558a9165e0
16 changed files with 316 additions and 321 deletions
Download Patch File Download Diff File Expand all files Collapse all files

153
src/backend/access/hash/hash_xlog.c
View File

@@ -969,155 +969,6 @@ hash_xlog_update_meta_page(XLogReaderState *record)
UnlockReleaseBuffer(metabuf);
}
/*
* Get the latestRemovedXid from the heap pages pointed at by the index
* tuples being deleted. See also btree_xlog_delete_get_latestRemovedXid,
* on which this function is based.
*/
static TransactionId
hash_xlog_vacuum_get_latestRemovedXid(XLogReaderState *record)
{
xl_hash_vacuum_one_page *xlrec;
OffsetNumber *unused;
Buffer ibuffer,
hbuffer;
Page ipage,
hpage;
RelFileNode rnode;
BlockNumber blkno;
ItemId iitemid,
hitemid;
IndexTuple itup;
HeapTupleHeader htuphdr;
BlockNumber hblkno;
OffsetNumber hoffnum;
TransactionId latestRemovedXid = InvalidTransactionId;
int i;
xlrec = (xl_hash_vacuum_one_page *) XLogRecGetData(record);
/*
* If there's nothing running on the standby we don't need to derive a
* full latestRemovedXid value, so use a fast path out of here. This
* returns InvalidTransactionId, and so will conflict with all HS
* transactions; but since we just worked out that that's zero people,
* it's OK.
*
* XXX There is a race condition here, which is that a new backend might
* start just after we look. If so, it cannot need to conflict, but this
* coding will result in throwing a conflict anyway.
*/
if (CountDBBackends(InvalidOid) == 0)
return latestRemovedXid;
/*
* Check if WAL replay has reached a consistent database state. If not, we
* must PANIC. See the definition of
* btree_xlog_delete_get_latestRemovedXid for more details.
*/
if (!reachedConsistency)
elog(PANIC, "hash_xlog_vacuum_get_latestRemovedXid: cannot operate with inconsistent data");
/*
* Get index page. If the DB is consistent, this should not fail, nor
* should any of the heap page fetches below. If one does, we return
* InvalidTransactionId to cancel all HS transactions. That's probably
* overkill, but it's safe, and certainly better than panicking here.
*/
XLogRecGetBlockTag(record, 0, &rnode, NULL, &blkno);
ibuffer = XLogReadBufferExtended(rnode, MAIN_FORKNUM, blkno, RBM_NORMAL);
if (!BufferIsValid(ibuffer))
return InvalidTransactionId;
LockBuffer(ibuffer, HASH_READ);
ipage = (Page) BufferGetPage(ibuffer);
/*
* Loop through the deleted index items to obtain the TransactionId from
* the heap items they point to.
*/
unused = (OffsetNumber *) ((char *) xlrec + SizeOfHashVacuumOnePage);
for (i = 0; i < xlrec->ntuples; i++)
{
/*
* Identify the index tuple about to be deleted.
*/
iitemid = PageGetItemId(ipage, unused[i]);
itup = (IndexTuple) PageGetItem(ipage, iitemid);
/*
* Locate the heap page that the index tuple points at
*/
hblkno = ItemPointerGetBlockNumber(&(itup->t_tid));
hbuffer = XLogReadBufferExtended(xlrec->hnode, MAIN_FORKNUM,
hblkno, RBM_NORMAL);
if (!BufferIsValid(hbuffer))
{
UnlockReleaseBuffer(ibuffer);
return InvalidTransactionId;
}
LockBuffer(hbuffer, HASH_READ);
hpage = (Page) BufferGetPage(hbuffer);
/*
* Look up the heap tuple header that the index tuple points at by
* using the heap node supplied with the xlrec. We can't use
* heap_fetch, since it uses ReadBuffer rather than XLogReadBuffer.
* Note that we are not looking at tuple data here, just headers.
*/
hoffnum = ItemPointerGetOffsetNumber(&(itup->t_tid));
hitemid = PageGetItemId(hpage, hoffnum);
/*
* Follow any redirections until we find something useful.
*/
while (ItemIdIsRedirected(hitemid))
{
hoffnum = ItemIdGetRedirect(hitemid);
hitemid = PageGetItemId(hpage, hoffnum);
CHECK_FOR_INTERRUPTS();
}
/*
* If the heap item has storage, then read the header and use that to
* set latestRemovedXid.
*
* Some LP_DEAD items may not be accessible, so we ignore them.
*/
if (ItemIdHasStorage(hitemid))
{
htuphdr = (HeapTupleHeader) PageGetItem(hpage, hitemid);
HeapTupleHeaderAdvanceLatestRemovedXid(htuphdr, &latestRemovedXid);
}
else if (ItemIdIsDead(hitemid))
{
/*
* Conjecture: if hitemid is dead then it had xids before the xids
* marked on LP_NORMAL items. So we just ignore this item and move
* onto the next, for the purposes of calculating
* latestRemovedxids.
*/
}
else
Assert(!ItemIdIsUsed(hitemid));
UnlockReleaseBuffer(hbuffer);
}
UnlockReleaseBuffer(ibuffer);
/*
* If all heap tuples were LP_DEAD then we will be returning
* InvalidTransactionId here, which avoids conflicts. This matches
* existing logic which assumes that LP_DEAD tuples must already be older
* than the latestRemovedXid on the cleanup record that set them as
* LP_DEAD, hence must already have generated a conflict.
*/
return latestRemovedXid;
}
/*
* replay delete operation in hash index to remove
* tuples marked as DEAD during index tuple insertion.
@@ -1149,12 +1000,10 @@ hash_xlog_vacuum_one_page(XLogReaderState *record)
*/
if (InHotStandby)
{
TransactionId latestRemovedXid =
hash_xlog_vacuum_get_latestRemovedXid(record);
RelFileNode rnode;
XLogRecGetBlockTag(record, 0, &rnode, NULL, NULL);
ResolveRecoveryConflictWithSnapshot(latestRemovedXid, rnode);
ResolveRecoveryConflictWithSnapshot(xldata->latestRemovedXid, rnode);
}
action = XLogReadBufferForRedoExtended(record, 0, RBM_NORMAL, true, &buffer);

17
src/backend/access/hash/hashinsert.c
View File

@@ -23,8 +23,8 @@
#include "storage/buf_internals.h"
#include "storage/predicate.h"
static void _hash_vacuum_one_page(Relation rel, Buffer metabuf, Buffer buf,
RelFileNode hnode);
static void _hash_vacuum_one_page(Relation rel, Relation hrel,
Buffer metabuf, Buffer buf);
/*
* _hash_doinsert() -- Handle insertion of a single index tuple.
@@ -137,7 +137,7 @@ restart_insert:
if (IsBufferCleanupOK(buf))
{
_hash_vacuum_one_page(rel, metabuf, buf, heapRel->rd_node);
_hash_vacuum_one_page(rel, heapRel, metabuf, buf);
if (PageGetFreeSpace(page) >= itemsz)
break; /* OK, now we have enough space */
@@ -335,8 +335,7 @@ _hash_pgaddmultitup(Relation rel, Buffer buf, IndexTuple *itups,
*/
static void
_hash_vacuum_one_page(Relation rel, Buffer metabuf, Buffer buf,
RelFileNode hnode)
_hash_vacuum_one_page(Relation rel, Relation hrel, Buffer metabuf, Buffer buf)
{
OffsetNumber deletable[MaxOffsetNumber];
int ndeletable = 0;
@@ -360,6 +359,12 @@ _hash_vacuum_one_page(Relation rel, Buffer metabuf, Buffer buf,
if (ndeletable > 0)
{
TransactionId latestRemovedXid;
latestRemovedXid =
index_compute_xid_horizon_for_tuples(rel, hrel, buf,
deletable, ndeletable);
/*
* Write-lock the meta page so that we can decrement tuple count.
*/
@@ -393,7 +398,7 @@ _hash_vacuum_one_page(Relation rel, Buffer metabuf, Buffer buf,
xl_hash_vacuum_one_page xlrec;
XLogRecPtr recptr;
xlrec.hnode = hnode;
xlrec.latestRemovedXid = latestRemovedXid;
xlrec.ntuples = ndeletable;
XLogBeginInsert();