/* Returns the relation object for the index that we're going to use as key for a
* particular table. (Indexes are relations too!) Returns null if the table is unkeyed.
* The return value is opened with a shared lock; call relation_close() when finished. */
Relation table_key_index(Relation rel) {
char replident = rel->rd_rel->relreplident;
Oid repl_ident_oid;
List *indexes;
ListCell *index_oid;
if (replident == REPLICA_IDENTITY_NOTHING) {
return NULL;
}
if (replident == REPLICA_IDENTITY_INDEX) {
repl_ident_oid = RelationGetReplicaIndex(rel);
if (repl_ident_oid != InvalidOid) {
return relation_open(repl_ident_oid, AccessShareLock);
}
}
// There doesn't seem to be a convenient way of getting the primary key index for
// a table, so we have to iterate over all the table's indexes.
indexes = RelationGetIndexList(rel);
foreach(index_oid, indexes) {
Relation index_rel = relation_open(lfirst_oid(index_oid), AccessShareLock);
Form_pg_index index = index_rel->rd_index;
if (IndexIsValid(index) && IndexIsReady(index) && index->indisprimary) {
list_free(indexes);
return index_rel;
}
relation_close(index_rel, AccessShareLock);
}
/*
* Calculate total on-disk size of all indexes attached to the given table.
*
* Can be applied safely to an index, but you'll just get zero.
*/
static int64
calculate_indexes_size(Relation rel)
{
int64 size = 0;
/*
* Aggregate all indexes on the given relation
*/
if (rel->rd_rel->relhasindex)
{
List *index_oids = RelationGetIndexList(rel);
ListCell *cell;
foreach(cell, index_oids)
{
Oid idxOid = lfirst_oid(cell);
Relation idxRel;
ForkNumber forkNum;
idxRel = relation_open(idxOid, AccessShareLock);
for (forkNum = 0; forkNum <= MAX_FORKNUM; forkNum++)
size += calculate_relation_size(&(idxRel->rd_node),
idxRel->rd_backend,
forkNum);
relation_close(idxRel, AccessShareLock);
}
/*
* Calculate total on-disk size of a TOAST relation, including its indexes.
* Must not be applied to non-TOAST relations.
*/
static int64
calculate_toast_table_size(Oid toastrelid)
{
int64 size = 0;
Relation toastRel;
ForkNumber forkNum;
ListCell *lc;
List *indexlist;
toastRel = relation_open(toastrelid, AccessShareLock);
/* toast heap size, including FSM and VM size */
for (forkNum = 0; forkNum <= MAX_FORKNUM; forkNum++)
size += calculate_relation_size(&(toastRel->rd_node),
toastRel->rd_backend, forkNum);
/* toast index size, including FSM and VM size */
indexlist = RelationGetIndexList(toastRel);
/* Size is calculated using all the indexes available */
foreach(lc, indexlist)
{
Relation toastIdxRel;
toastIdxRel = relation_open(lfirst_oid(lc),
AccessShareLock);
for (forkNum = 0; forkNum <= MAX_FORKNUM; forkNum++)
size += calculate_relation_size(&(toastIdxRel->rd_node),
toastIdxRel->rd_backend, forkNum);
relation_close(toastIdxRel, AccessShareLock);
}
/*
* Compute the on-disk size of files for the relation according to the
* stat function, including heap data, index data, and toast data.
*/
static int64
calculate_total_relation_size(Oid Relid)
{
Relation heapRel;
Oid toastOid;
int64 size;
ListCell *cell;
heapRel = relation_open(Relid, AccessShareLock);
toastOid = heapRel->rd_rel->reltoastrelid;
/* Get the heap size */
size = calculate_relation_size(&(heapRel->rd_node));
/* Include any dependent indexes */
if (heapRel->rd_rel->relhasindex)
{
List *index_oids = RelationGetIndexList(heapRel);
foreach(cell, index_oids)
{
Oid idxOid = lfirst_oid(cell);
Relation iRel;
iRel = relation_open(idxOid, AccessShareLock);
size += calculate_relation_size(&(iRel->rd_node));
relation_close(iRel, AccessShareLock);
}
/*
* Generate a WHERE clause for UPDATE or DELETE.
*/
static void
print_where_clause(StringInfo s,
Relation relation,
HeapTuple oldtuple,
HeapTuple newtuple)
{
TupleDesc tupdesc = RelationGetDescr(relation);
int natt;
bool first_column = true;
Assert(relation->rd_rel->relreplident == REPLICA_IDENTITY_DEFAULT ||
relation->rd_rel->relreplident == REPLICA_IDENTITY_FULL ||
relation->rd_rel->relreplident == REPLICA_IDENTITY_INDEX);
/* Build the WHERE clause */
appendStringInfoString(s, " WHERE ");
RelationGetIndexList(relation);
/* Generate WHERE clause using new values of REPLICA IDENTITY */
if (OidIsValid(relation->rd_replidindex))
{
Relation indexRel;
int key;
/* Use all the values associated with the index */
indexRel = index_open(relation->rd_replidindex, AccessShareLock);
for (key = 0; key < indexRel->rd_index->indnatts; key++)
{
int relattr = indexRel->rd_index->indkey.values[key];
/*
* For a relation having REPLICA IDENTITY set at DEFAULT
* or INDEX, if one of the columns used for tuple selectivity
* is changed, the old tuple data is not NULL and need to
* be used for tuple selectivity. If no such columns are
* updated, old tuple data is NULL.
*/
print_where_clause_item(s, relation,
oldtuple ? oldtuple : newtuple,
relattr, &first_column);
}
index_close(indexRel, NoLock);
return;
}
/* We need absolutely some values for tuple selectivity now */
Assert(oldtuple != NULL &&
relation->rd_rel->relreplident == REPLICA_IDENTITY_FULL);
/*
* Fallback to default case, use of old values and print WHERE clause
* using all the columns. This is actually the code path for FULL.
*/
for (natt = 0; natt < tupdesc->natts; natt++)
print_where_clause_item(s, relation, oldtuple,
natt + 1, &first_column);
}
/*---------------------------------------------------------------------------
* This cluster code allows for clustering multiple tables at once. Because
* of this, we cannot just run everything on a single transaction, or we
* would be forced to acquire exclusive locks on all the tables being
* clustered, simultaneously --- very likely leading to deadlock.
*
* To solve this we follow a similar strategy to VACUUM code,
* clustering each relation in a separate transaction. For this to work,
* we need to:
* - provide a separate memory context so that we can pass information in
* a way that survives across transactions
* - start a new transaction every time a new relation is clustered
* - check for validity of the information on to-be-clustered relations,
* as someone might have deleted a relation behind our back, or
* clustered one on a different index
* - end the transaction
*
* The single-relation case does not have any such overhead.
*
* We also allow a relation to be specified without index. In that case,
* the indisclustered bit will be looked up, and an ERROR will be thrown
* if there is no index with the bit set.
*---------------------------------------------------------------------------
*/
void
cluster(ClusterStmt *stmt, bool isTopLevel)
{
if (stmt->relation != NULL)
{
/* This is the single-relation case. */
Oid tableOid,
indexOid = InvalidOid;
Relation rel;
RelToCluster rvtc;
/* Find and lock the table */
rel = heap_openrv(stmt->relation, AccessExclusiveLock);
tableOid = RelationGetRelid(rel);
/* Check permissions */
if (!pg_class_ownercheck(tableOid, GetUserId()))
aclcheck_error(ACLCHECK_NOT_OWNER, ACL_KIND_CLASS,
RelationGetRelationName(rel));
/*
* Reject clustering a remote temp table ... their local buffer
* manager is not going to cope.
*/
if (isOtherTempNamespace(RelationGetNamespace(rel)))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
if (stmt->indexname == NULL)
{
ListCell *index;
/* We need to find the index that has indisclustered set. */
foreach(index, RelationGetIndexList(rel))
{
HeapTuple idxtuple;
Form_pg_index indexForm;
indexOid = lfirst_oid(index);
idxtuple = SearchSysCache(INDEXRELID,
ObjectIdGetDatum(indexOid),
0, 0, 0);
if (!HeapTupleIsValid(idxtuple))
elog(ERROR, "cache lookup failed for index %u", indexOid);
indexForm = (Form_pg_index) GETSTRUCT(idxtuple);
if (indexForm->indisclustered)
{
ReleaseSysCache(idxtuple);
break;
}
ReleaseSysCache(idxtuple);
indexOid = InvalidOid;
}
/*---------------------------------------------------------------------------
* This cluster code allows for clustering multiple tables at once. Because
* of this, we cannot just run everything on a single transaction, or we
* would be forced to acquire exclusive locks on all the tables being
* clustered, simultaneously --- very likely leading to deadlock.
*
* To solve this we follow a similar strategy to VACUUM code,
* clustering each relation in a separate transaction. For this to work,
* we need to:
* - provide a separate memory context so that we can pass information in
* a way that survives across transactions
* - start a new transaction every time a new relation is clustered
* - check for validity of the information on to-be-clustered relations,
* as someone might have deleted a relation behind our back, or
* clustered one on a different index
* - end the transaction
*
* The single-relation case does not have any such overhead.
*
* We also allow a relation to be specified without index. In that case,
* the indisclustered bit will be looked up, and an ERROR will be thrown
* if there is no index with the bit set.
*---------------------------------------------------------------------------
*/
void
cluster(ClusterStmt *stmt, bool isTopLevel)
{
if (stmt->relation != NULL)
{
/* This is the single-relation case. */
Oid tableOid,
indexOid = InvalidOid;
Relation rel;
/* Find, lock, and check permissions on the table */
tableOid = RangeVarGetRelidExtended(stmt->relation,
AccessExclusiveLock,
false, false,
RangeVarCallbackOwnsTable, NULL);
rel = heap_open(tableOid, NoLock);
/*
* Reject clustering a remote temp table ... their local buffer
* manager is not going to cope.
*/
if (RELATION_IS_OTHER_TEMP(rel))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot cluster temporary tables of other sessions")));
if (stmt->indexname == NULL)
{
ListCell *index;
/* We need to find the index that has indisclustered set. */
foreach(index, RelationGetIndexList(rel))
{
HeapTuple idxtuple;
Form_pg_index indexForm;
indexOid = lfirst_oid(index);
idxtuple = SearchSysCache1(INDEXRELID,
ObjectIdGetDatum(indexOid));
if (!HeapTupleIsValid(idxtuple))
elog(ERROR, "cache lookup failed for index %u", indexOid);
indexForm = (Form_pg_index) GETSTRUCT(idxtuple);
if (indexForm->indisclustered)
{
ReleaseSysCache(idxtuple);
break;
}
ReleaseSysCache(idxtuple);
indexOid = InvalidOid;
}
/*---------------------------------------------------------------------------
* This cluster code allows for clustering multiple tables at once. Because
* of this, we cannot just run everything on a single transaction, or we
* would be forced to acquire exclusive locks on all the tables being
* clustered, simultaneously --- very likely leading to deadlock.
*
* To solve this we follow a similar strategy to VACUUM code,
* clustering each relation in a separate transaction. For this to work,
* we need to:
* - provide a separate memory context so that we can pass information in
* a way that survives across transactions
* - start a new transaction every time a new relation is clustered
* - check for validity of the information on to-be-clustered relations,
* as someone might have deleted a relation behind our back, or
* clustered one on a different index
* - end the transaction
*
* The single-relation case does not have any such overhead.
*
* We also allow a relation being specified without index. In that case,
* the indisclustered bit will be looked up, and an ERROR will be thrown
* if there is no index with the bit set.
*---------------------------------------------------------------------------
*/
void
cluster(ClusterStmt *stmt)
{
if (stmt->relation != NULL)
{
/* This is the single-relation case. */
Oid tableOid,
indexOid = InvalidOid;
Relation rel;
RelToCluster rvtc;
/* Find and lock the table */
rel = heap_openrv(stmt->relation, AccessExclusiveLock);
tableOid = RelationGetRelid(rel);
/* Check permissions */
if (!pg_class_ownercheck(tableOid, GetUserId()))
aclcheck_error(ACLCHECK_NOT_OWNER, ACL_KIND_CLASS,
RelationGetRelationName(rel));
if (stmt->indexname == NULL)
{
List *index;
/* We need to find the index that has indisclustered set. */
foreach(index, RelationGetIndexList(rel))
{
HeapTuple idxtuple;
Form_pg_index indexForm;
indexOid = lfirsto(index);
idxtuple = SearchSysCache(INDEXRELID,
ObjectIdGetDatum(indexOid),
0, 0, 0);
if (!HeapTupleIsValid(idxtuple))
elog(ERROR, "cache lookup failed for index %u", indexOid);
indexForm = (Form_pg_index) GETSTRUCT(idxtuple);
if (indexForm->indisclustered)
{
ReleaseSysCache(idxtuple);
break;
}
ReleaseSysCache(idxtuple);
indexOid = InvalidOid;
}
/*
* Compute the on-disk size of files for the relation according to the
* stat function, including heap data, index data, toast data, aoseg data,
* aoblkdir data, and aovisimap data.
*/
static int64
calculate_total_relation_size(Oid Relid)
{
Relation heapRel;
Oid toastOid;
int64 size;
ListCell *cell;
ForkNumber forkNum;
heapRel = try_relation_open(Relid, AccessShareLock, false);
if (!RelationIsValid(heapRel))
return 0;
toastOid = heapRel->rd_rel->reltoastrelid;
/* Get the heap size */
if (Relid == 0 || heapRel->rd_node.relNode == 0)
size = 0;
else
{
size = 0;
for (forkNum = 0; forkNum <= MAX_FORKNUM; forkNum++)
size += calculate_relation_size(heapRel, forkNum);
}
/* Include any dependent indexes */
if (heapRel->rd_rel->relhasindex)
{
List *index_oids = RelationGetIndexList(heapRel);
foreach(cell, index_oids)
{
Oid idxOid = lfirst_oid(cell);
Relation iRel;
iRel = try_relation_open(idxOid, AccessShareLock, false);
if (RelationIsValid(iRel))
{
for (forkNum = 0; forkNum <= MAX_FORKNUM; forkNum++)
size += calculate_relation_size(iRel, forkNum);
relation_close(iRel, AccessShareLock);
}
}
/*
* Compute the on-disk size of files for the relation according to the
* stat function, including heap data, index data, toast data, aoseg data,
* aoblkdir data, and aovisimap data.
*/
static int64
calculate_total_relation_size(Oid Relid)
{
Relation heapRel;
Oid toastOid;
AppendOnlyEntry *aoEntry = NULL;
int64 size;
ListCell *cell;
heapRel = try_relation_open(Relid, AccessShareLock, false);
if (!RelationIsValid(heapRel))
return 0;
toastOid = heapRel->rd_rel->reltoastrelid;
if (RelationIsAoRows(heapRel) || RelationIsAoCols(heapRel))
aoEntry = GetAppendOnlyEntry(Relid, SnapshotNow);
/* Get the heap size */
if (Relid == 0 || heapRel->rd_node.relNode == 0)
size = 0;
else
size = calculate_relation_size(heapRel);
/* Include any dependent indexes */
if (heapRel->rd_rel->relhasindex)
{
List *index_oids = RelationGetIndexList(heapRel);
foreach(cell, index_oids)
{
Oid idxOid = lfirst_oid(cell);
Relation iRel;
iRel = try_relation_open(idxOid, AccessShareLock, false);
if (RelationIsValid(iRel))
{
size += calculate_relation_size(iRel);
relation_close(iRel, AccessShareLock);
}
}
/* ----------------------------------------------------------------
* ExecOpenIndices
*
* Find the indices associated with a result relation, open them,
* and save information about them in the result ResultRelInfo.
*
* At entry, caller has already opened and locked
* resultRelInfo->ri_RelationDesc.
* ----------------------------------------------------------------
*/
void
ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative)
{
Relation resultRelation = resultRelInfo->ri_RelationDesc;
List *indexoidlist;
ListCell *l;
int len,
i;
RelationPtr relationDescs;
IndexInfo **indexInfoArray;
resultRelInfo->ri_NumIndices = 0;
/* fast path if no indexes */
if (!RelationGetForm(resultRelation)->relhasindex)
return;
/*
* Get cached list of index OIDs
*/
indexoidlist = RelationGetIndexList(resultRelation);
len = list_length(indexoidlist);
if (len == 0)
return;
/*
* allocate space for result arrays
*/
relationDescs = (RelationPtr) palloc(len * sizeof(Relation));
indexInfoArray = (IndexInfo **) palloc(len * sizeof(IndexInfo *));
resultRelInfo->ri_NumIndices = len;
resultRelInfo->ri_IndexRelationDescs = relationDescs;
resultRelInfo->ri_IndexRelationInfo = indexInfoArray;
/*
* For each index, open the index relation and save pg_index info. We
* acquire RowExclusiveLock, signifying we will update the index.
*
* Note: we do this even if the index is not indisready; it's not worth
* the trouble to optimize for the case where it isn't.
*/
i = 0;
foreach(l, indexoidlist)
{
Oid indexOid = lfirst_oid(l);
Relation indexDesc;
IndexInfo *ii;
indexDesc = index_open(indexOid, RowExclusiveLock);
/* extract index key information from the index's pg_index info */
ii = BuildIndexInfo(indexDesc);
/*
* If the indexes are to be used for speculative insertion, add extra
* information required by unique index entries.
*/
if (speculative && ii->ii_Unique)
BuildSpeculativeIndexInfo(indexDesc, ii);
relationDescs[i] = indexDesc;
indexInfoArray[i] = ii;
i++;
}
/*
* get_relation_info -
* Retrieves catalog information for a given relation.
*
* Given the Oid of the relation, return the following info into fields
* of the RelOptInfo struct:
*
* min_attr lowest valid AttrNumber
* max_attr highest valid AttrNumber
* indexlist list of IndexOptInfos for relation's indexes
* pages number of pages
* tuples number of tuples
*
* Also, initialize the attr_needed[] and attr_widths[] arrays. In most
* cases these are left as zeroes, but sometimes we need to compute attr
* widths here, and we may as well cache the results for costsize.c.
*
* If inhparent is true, all we need to do is set up the attr arrays:
* the RelOptInfo actually represents the appendrel formed by an inheritance
* tree, and so the parent rel's physical size and index information isn't
* important for it.
*/
void
get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
RelOptInfo *rel)
{
Index varno = rel->relid;
Relation relation;
bool hasindex;
List *indexinfos = NIL;
/*
* We need not lock the relation since it was already locked, either by
* the rewriter or when expand_inherited_rtentry() added it to the query's
* rangetable.
*/
relation = heap_open(relationObjectId, NoLock);
/* Temporary and unlogged relations are inaccessible during recovery. */
if (!RelationNeedsWAL(relation) && RecoveryInProgress())
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot access temporary or unlogged relations during recovery")));
rel->min_attr = FirstLowInvalidHeapAttributeNumber + 1;
rel->max_attr = RelationGetNumberOfAttributes(relation);
rel->reltablespace = RelationGetForm(relation)->reltablespace;
Assert(rel->max_attr >= rel->min_attr);
rel->attr_needed = (Relids *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(Relids));
rel->attr_widths = (int32 *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(int32));
/*
* Estimate relation size --- unless it's an inheritance parent, in which
* case the size will be computed later in set_append_rel_pathlist, and we
* must leave it zero for now to avoid bollixing the total_table_pages
* calculation.
*/
if (!inhparent)
estimate_rel_size(relation, rel->attr_widths - rel->min_attr,
&rel->pages, &rel->tuples, &rel->allvisfrac);
/*
* Make list of indexes. Ignore indexes on system catalogs if told to.
* Don't bother with indexes for an inheritance parent, either.
*/
if (inhparent ||
(IgnoreSystemIndexes && IsSystemClass(relation->rd_rel)))
hasindex = false;
else
hasindex = relation->rd_rel->relhasindex;
if (hasindex)
{
List *indexoidlist;
ListCell *l;
LOCKMODE lmode;
indexoidlist = RelationGetIndexList(relation);
/*
* For each index, we get the same type of lock that the executor will
* need, and do not release it. This saves a couple of trips to the
* shared lock manager while not creating any real loss of
* concurrency, because no schema changes could be happening on the
* index while we hold lock on the parent rel, and neither lock type
* blocks any other kind of index operation.
*/
if (rel->relid == root->parse->resultRelation)
lmode = RowExclusiveLock;
else
lmode = AccessShareLock;
foreach(l, indexoidlist)
{
Oid indexoid = lfirst_oid(l);
Relation indexRelation;
Form_pg_index index;
IndexOptInfo *info;
int ncolumns;
int i;
/*
* Extract info from the relation descriptor for the index.
*/
indexRelation = index_open(indexoid, lmode);
index = indexRelation->rd_index;
/*
* Ignore invalid indexes, since they can't safely be used for
* queries. Note that this is OK because the data structure we
* are constructing is only used by the planner --- the executor
* still needs to insert into "invalid" indexes!
*/
if (!index->indisvalid)
{
index_close(indexRelation, NoLock);
continue;
}
/*
* If the index is valid, but cannot yet be used, ignore it; but
* mark the plan we are generating as transient. See
* src/backend/access/heap/README.HOT for discussion.
*/
if (index->indcheckxmin &&
!TransactionIdPrecedes(HeapTupleHeaderGetXmin(indexRelation->rd_indextuple->t_data),
TransactionXmin))
{
root->glob->transientPlan = true;
index_close(indexRelation, NoLock);
continue;
}
info = makeNode(IndexOptInfo);
info->indexoid = index->indexrelid;
info->reltablespace =
RelationGetForm(indexRelation)->reltablespace;
info->rel = rel;
info->ncolumns = ncolumns = index->indnatts;
info->indexkeys = (int *) palloc(sizeof(int) * ncolumns);
info->indexcollations = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->opfamily = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->opcintype = (Oid *) palloc(sizeof(Oid) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
info->indexkeys[i] = index->indkey.values[i];
info->indexcollations[i] = indexRelation->rd_indcollation[i];
info->opfamily[i] = indexRelation->rd_opfamily[i];
info->opcintype[i] = indexRelation->rd_opcintype[i];
}
info->relam = indexRelation->rd_rel->relam;
info->amcostestimate = indexRelation->rd_am->amcostestimate;
info->canreturn = index_can_return(indexRelation);
info->amcanorderbyop = indexRelation->rd_am->amcanorderbyop;
info->amoptionalkey = indexRelation->rd_am->amoptionalkey;
info->amsearcharray = indexRelation->rd_am->amsearcharray;
info->amsearchnulls = indexRelation->rd_am->amsearchnulls;
info->amhasgettuple = OidIsValid(indexRelation->rd_am->amgettuple);
info->amhasgetbitmap = OidIsValid(indexRelation->rd_am->amgetbitmap);
/*
* Fetch the ordering information for the index, if any.
*/
if (info->relam == BTREE_AM_OID)
{
/*
* If it's a btree index, we can use its opfamily OIDs
* directly as the sort ordering opfamily OIDs.
*/
Assert(indexRelation->rd_am->amcanorder);
info->sortopfamily = info->opfamily;
info->reverse_sort = (bool *) palloc(sizeof(bool) * ncolumns);
info->nulls_first = (bool *) palloc(sizeof(bool) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
int16 opt = indexRelation->rd_indoption[i];
info->reverse_sort[i] = (opt & INDOPTION_DESC) != 0;
info->nulls_first[i] = (opt & INDOPTION_NULLS_FIRST) != 0;
}
}
else if (indexRelation->rd_am->amcanorder)
{
/*
* Otherwise, identify the corresponding btree opfamilies by
* trying to map this index's "<" operators into btree. Since
* "<" uniquely defines the behavior of a sort order, this is
* a sufficient test.
*
* XXX This method is rather slow and also requires the
* undesirable assumption that the other index AM numbers its
* strategies the same as btree. It'd be better to have a way
* to explicitly declare the corresponding btree opfamily for
* each opfamily of the other index type. But given the lack
* of current or foreseeable amcanorder index types, it's not
* worth expending more effort on now.
*/
info->sortopfamily = (Oid *) palloc(sizeof(Oid) * ncolumns);
info->reverse_sort = (bool *) palloc(sizeof(bool) * ncolumns);
info->nulls_first = (bool *) palloc(sizeof(bool) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
int16 opt = indexRelation->rd_indoption[i];
Oid ltopr;
Oid btopfamily;
Oid btopcintype;
int16 btstrategy;
info->reverse_sort[i] = (opt & INDOPTION_DESC) != 0;
info->nulls_first[i] = (opt & INDOPTION_NULLS_FIRST) != 0;
ltopr = get_opfamily_member(info->opfamily[i],
info->opcintype[i],
info->opcintype[i],
BTLessStrategyNumber);
if (OidIsValid(ltopr) &&
get_ordering_op_properties(ltopr,
&btopfamily,
&btopcintype,
&btstrategy) &&
btopcintype == info->opcintype[i] &&
btstrategy == BTLessStrategyNumber)
{
/* Successful mapping */
info->sortopfamily[i] = btopfamily;
}
else
{
/* Fail ... quietly treat index as unordered */
info->sortopfamily = NULL;
info->reverse_sort = NULL;
info->nulls_first = NULL;
break;
}
}
}
else
{
info->sortopfamily = NULL;
info->reverse_sort = NULL;
info->nulls_first = NULL;
}
/*
* Fetch the index expressions and predicate, if any. We must
* modify the copies we obtain from the relcache to have the
* correct varno for the parent relation, so that they match up
* correctly against qual clauses.
*/
info->indexprs = RelationGetIndexExpressions(indexRelation);
info->indpred = RelationGetIndexPredicate(indexRelation);
if (info->indexprs && varno != 1)
ChangeVarNodes((Node *) info->indexprs, 1, varno, 0);
if (info->indpred && varno != 1)
ChangeVarNodes((Node *) info->indpred, 1, varno, 0);
/* Build targetlist using the completed indexprs data */
info->indextlist = build_index_tlist(root, info, relation);
info->predOK = false; /* set later in indxpath.c */
info->unique = index->indisunique;
info->immediate = index->indimmediate;
info->hypothetical = false;
/*
* Estimate the index size. If it's not a partial index, we lock
* the number-of-tuples estimate to equal the parent table; if it
* is partial then we have to use the same methods as we would for
* a table, except we can be sure that the index is not larger
* than the table.
*/
if (info->indpred == NIL)
{
info->pages = RelationGetNumberOfBlocks(indexRelation);
info->tuples = rel->tuples;
}
else
{
double allvisfrac; /* dummy */
estimate_rel_size(indexRelation, NULL,
&info->pages, &info->tuples, &allvisfrac);
if (info->tuples > rel->tuples)
info->tuples = rel->tuples;
}
index_close(indexRelation, NoLock);
indexinfos = lcons(info, indexinfos);
}
list_free(indexoidlist);
}
/*
* ExecRefreshMatView -- execute a REFRESH MATERIALIZED VIEW command
*
* This refreshes the materialized view by creating a new table and swapping
* the relfilenodes of the new table and the old materialized view, so the OID
* of the original materialized view is preserved. Thus we do not lose GRANT
* nor references to this materialized view.
*
* If WITH NO DATA was specified, this is effectively like a TRUNCATE;
* otherwise it is like a TRUNCATE followed by an INSERT using the SELECT
* statement associated with the materialized view. The statement node's
* skipData field shows whether the clause was used.
*
* Indexes are rebuilt too, via REINDEX. Since we are effectively bulk-loading
* the new heap, it's better to create the indexes afterwards than to fill them
* incrementally while we load.
*
* The matview's "populated" state is changed based on whether the contents
* reflect the result set of the materialized view's query.
*/
ObjectAddress
ExecRefreshMatView(RefreshMatViewStmt *stmt, const char *queryString,
ParamListInfo params, char *completionTag)
{
Oid matviewOid;
Relation matviewRel;
RewriteRule *rule;
List *actions;
Query *dataQuery;
Oid tableSpace;
Oid relowner;
Oid OIDNewHeap;
DestReceiver *dest;
uint64 processed = 0;
bool concurrent;
LOCKMODE lockmode;
char relpersistence;
Oid save_userid;
int save_sec_context;
int save_nestlevel;
ObjectAddress address;
/* Determine strength of lock needed. */
concurrent = stmt->concurrent;
lockmode = concurrent ? ExclusiveLock : AccessExclusiveLock;
/*
* Get a lock until end of transaction.
*/
matviewOid = RangeVarGetRelidExtended(stmt->relation,
lockmode, 0,
RangeVarCallbackOwnsTable, NULL);
matviewRel = table_open(matviewOid, NoLock);
/* Make sure it is a materialized view. */
if (matviewRel->rd_rel->relkind != RELKIND_MATVIEW)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("\"%s\" is not a materialized view",
RelationGetRelationName(matviewRel))));
/* Check that CONCURRENTLY is not specified if not populated. */
if (concurrent && !RelationIsPopulated(matviewRel))
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("CONCURRENTLY cannot be used when the materialized view is not populated")));
/* Check that conflicting options have not been specified. */
if (concurrent && stmt->skipData)
ereport(ERROR,
(errcode(ERRCODE_SYNTAX_ERROR),
errmsg("CONCURRENTLY and WITH NO DATA options cannot be used together")));
/*
* Check that everything is correct for a refresh. Problems at this point
* are internal errors, so elog is sufficient.
*/
if (matviewRel->rd_rel->relhasrules == false ||
matviewRel->rd_rules->numLocks < 1)
elog(ERROR,
"materialized view \"%s\" is missing rewrite information",
RelationGetRelationName(matviewRel));
if (matviewRel->rd_rules->numLocks > 1)
elog(ERROR,
"materialized view \"%s\" has too many rules",
RelationGetRelationName(matviewRel));
rule = matviewRel->rd_rules->rules[0];
if (rule->event != CMD_SELECT || !(rule->isInstead))
elog(ERROR,
"the rule for materialized view \"%s\" is not a SELECT INSTEAD OF rule",
RelationGetRelationName(matviewRel));
actions = rule->actions;
if (list_length(actions) != 1)
elog(ERROR,
"the rule for materialized view \"%s\" is not a single action",
RelationGetRelationName(matviewRel));
/*
* Check that there is a unique index with no WHERE clause on one or more
* columns of the materialized view if CONCURRENTLY is specified.
*/
if (concurrent)
{
List *indexoidlist = RelationGetIndexList(matviewRel);
ListCell *indexoidscan;
bool hasUniqueIndex = false;
foreach(indexoidscan, indexoidlist)
{
Oid indexoid = lfirst_oid(indexoidscan);
Relation indexRel;
indexRel = index_open(indexoid, AccessShareLock);
hasUniqueIndex = is_usable_unique_index(indexRel);
index_close(indexRel, AccessShareLock);
if (hasUniqueIndex)
break;
}
list_free(indexoidlist);
if (!hasUniqueIndex)
ereport(ERROR,
(errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE),
errmsg("cannot refresh materialized view \"%s\" concurrently",
quote_qualified_identifier(get_namespace_name(RelationGetNamespace(matviewRel)),
RelationGetRelationName(matviewRel))),
errhint("Create a unique index with no WHERE clause on one or more columns of the materialized view.")));
}
static void pg_decode_change(LogicalDecodingContext* ctx, ReorderBufferTXN* txn, Relation relation, ReorderBufferChange* change) {
DecodingJsonData* data;
Form_pg_class class_form;
TupleDesc tupdesc;
HeapTuple tuple;
MemoryContext old;
data = ctx->output_plugin_private;
data->xact_wrote_changes = true;
class_form = RelationGetForm(relation);
tupdesc = RelationGetDescr(relation);
old = MemoryContextSwitchTo(data->context);
OutputPluginPrepareWrite(ctx, true);
appendStringInfoString(ctx->out, "{\"type\":\"table\"");
appendStringInfo(
ctx->out,
",\"schema\":\"%s\"",
get_namespace_name(
get_rel_namespace(
RelationGetRelid(relation)
)
)
);
appendStringInfo(ctx->out, ",\"name\":\"%s\"", NameStr(class_form->relname));
appendStringInfo(
ctx->out,
",\"change\":\"%s\"",
change->action == REORDER_BUFFER_CHANGE_INSERT
? "INSERT"
: change->action == REORDER_BUFFER_CHANGE_UPDATE
? "UPDATE"
: change->action == REORDER_BUFFER_CHANGE_DELETE
? "DELETE"
: "FIXME"
);
if (change->action == REORDER_BUFFER_CHANGE_UPDATE || change->action == REORDER_BUFFER_CHANGE_DELETE) {
appendStringInfoString(ctx->out, ",\"key\":{");
RelationGetIndexList(relation);
if (OidIsValid(relation->rd_replidindex)) {
int i;
Relation index = index_open(relation->rd_replidindex, ShareLock);
tuple =
change->data.tp.oldtuple
? &change->data.tp.oldtuple->tuple
: &change->data.tp.newtuple->tuple;
for (i = 0; i < index->rd_index->indnatts; i++) {
int j = index->rd_index->indkey.values[i];
Form_pg_attribute attr = tupdesc->attrs[j - 1];
if (i > 0) appendStringInfoChar(ctx->out, ',');
appendStringInfo(ctx->out, "\"%s\":", NameStr(attr->attname));
print_value(ctx->out, tupdesc, tuple, j - 1);
}
index_close(index, NoLock);
} else {
appendStringInfoString(ctx->out, "\"***FIXME***\"");
}
appendStringInfoChar(ctx->out, '}');
}
if (change->action == REORDER_BUFFER_CHANGE_UPDATE || change->action == REORDER_BUFFER_CHANGE_INSERT) {
appendStringInfoString(ctx->out, ",\"data\":{");
tuple_to_stringinfo(ctx->out, tupdesc, &change->data.tp.newtuple->tuple, false);
appendStringInfoChar(ctx->out, '}');
}
appendStringInfoChar(ctx->out, '}');
MemoryContextSwitchTo(old);
MemoryContextReset(data->context);
OutputPluginWrite(ctx, true);
}
/*
* refresh_by_match_merge
*
* Refresh a materialized view with transactional semantics, while allowing
* concurrent reads.
*
* This is called after a new version of the data has been created in a
* temporary table. It performs a full outer join against the old version of
* the data, producing "diff" results. This join cannot work if there are any
* duplicated rows in either the old or new versions, in the sense that every
* column would compare as equal between the two rows. It does work correctly
* in the face of rows which have at least one NULL value, with all non-NULL
* columns equal. The behavior of NULLs on equality tests and on UNIQUE
* indexes turns out to be quite convenient here; the tests we need to make
* are consistent with default behavior. If there is at least one UNIQUE
* index on the materialized view, we have exactly the guarantee we need. By
* joining based on equality on all columns which are part of any unique
* index, we identify the rows on which we can use UPDATE without any problem.
* If any column is NULL in either the old or new version of a row (or both),
* we must use DELETE and INSERT, since there could be multiple rows which are
* NOT DISTINCT FROM each other, and we could otherwise end up with the wrong
* number of occurrences in the updated relation. The temporary table used to
* hold the diff results contains just the TID of the old record (if matched)
* and the ROW from the new table as a single column of complex record type
* (if matched).
*
* Once we have the diff table, we perform set-based DELETE, UPDATE, and
* INSERT operations against the materialized view, and discard both temporary
* tables.
*
* Everything from the generation of the new data to applying the differences
* takes place under cover of an ExclusiveLock, since it seems as though we
* would want to prohibit not only concurrent REFRESH operations, but also
* incremental maintenance. It also doesn't seem reasonable or safe to allow
* SELECT FOR UPDATE or SELECT FOR SHARE on rows being updated or deleted by
* this command.
*/
static void
refresh_by_match_merge(Oid matviewOid, Oid tempOid)
{
StringInfoData querybuf;
Relation matviewRel;
Relation tempRel;
char *matviewname;
char *tempname;
char *diffname;
TupleDesc tupdesc;
bool foundUniqueIndex;
List *indexoidlist;
ListCell *indexoidscan;
int16 relnatts;
bool *usedForQual;
Oid save_userid;
int save_sec_context;
int save_nestlevel;
initStringInfo(&querybuf);
matviewRel = heap_open(matviewOid, NoLock);
matviewname = quote_qualified_identifier(get_namespace_name(RelationGetNamespace(matviewRel)),
RelationGetRelationName(matviewRel));
tempRel = heap_open(tempOid, NoLock);
tempname = quote_qualified_identifier(get_namespace_name(RelationGetNamespace(tempRel)),
RelationGetRelationName(tempRel));
diffname = make_temptable_name_n(tempname, 2);
relnatts = matviewRel->rd_rel->relnatts;
usedForQual = (bool *) palloc0(sizeof(bool) * relnatts);
/* Open SPI context. */
if (SPI_connect() != SPI_OK_CONNECT)
elog(ERROR, "SPI_connect failed");
/* Analyze the temp table with the new contents. */
appendStringInfo(&querybuf, "ANALYZE %s", tempname);
if (SPI_exec(querybuf.data, 0) != SPI_OK_UTILITY)
elog(ERROR, "SPI_exec failed: %s", querybuf.data);
/*
* We need to ensure that there are not duplicate rows without NULLs in
* the new data set before we can count on the "diff" results. Check for
* that in a way that allows showing the first duplicated row found. Even
* after we pass this test, a unique index on the materialized view may
* find a duplicate key problem.
*/
resetStringInfo(&querybuf);
appendStringInfo(&querybuf,
"SELECT x FROM %s x WHERE x IS NOT NULL AND EXISTS "
"(SELECT * FROM %s y WHERE y IS NOT NULL "
"AND (y.*) = (x.*) AND y.ctid <> x.ctid) LIMIT 1",
tempname, tempname);
if (SPI_execute(querybuf.data, false, 1) != SPI_OK_SELECT)
elog(ERROR, "SPI_exec failed: %s", querybuf.data);
if (SPI_processed > 0)
{
ereport(ERROR,
(errcode(ERRCODE_CARDINALITY_VIOLATION),
errmsg("new data for \"%s\" contains duplicate rows without any NULL columns",
RelationGetRelationName(matviewRel)),
errdetail("Row: %s",
SPI_getvalue(SPI_tuptable->vals[0], SPI_tuptable->tupdesc, 1))));
}
/* Start building the query for creating the diff table. */
resetStringInfo(&querybuf);
appendStringInfo(&querybuf,
"CREATE TEMP TABLE %s AS "
"SELECT x.ctid AS tid, y FROM %s x FULL JOIN %s y ON (",
diffname, matviewname, tempname);
/*
* Get the list of index OIDs for the table from the relcache, and look up
* each one in the pg_index syscache. We will test for equality on all
* columns present in all unique indexes which only reference columns and
* include all rows.
*/
tupdesc = matviewRel->rd_att;
foundUniqueIndex = false;
indexoidlist = RelationGetIndexList(matviewRel);
foreach(indexoidscan, indexoidlist)
{
Oid indexoid = lfirst_oid(indexoidscan);
HeapTuple indexTuple;
Form_pg_index index;
indexTuple = SearchSysCache1(INDEXRELID, ObjectIdGetDatum(indexoid));
if (!HeapTupleIsValid(indexTuple)) /* should not happen */
elog(ERROR, "cache lookup failed for index %u", indexoid);
index = (Form_pg_index) GETSTRUCT(indexTuple);
/* We're only interested if it is unique and valid. */
if (index->indisunique && IndexIsValid(index))
{
int numatts = index->indnatts;
int i;
bool expr = false;
Relation indexRel;
/* Skip any index on an expression. */
for (i = 0; i < numatts; i++)
{
if (index->indkey.values[i] == 0)
{
expr = true;
break;
}
}
if (expr)
{
ReleaseSysCache(indexTuple);
continue;
}
/* Skip partial indexes. */
indexRel = index_open(index->indexrelid, RowExclusiveLock);
if (RelationGetIndexPredicate(indexRel) != NIL)
{
index_close(indexRel, NoLock);
ReleaseSysCache(indexTuple);
continue;
}
/* Hold the locks, since we're about to run DML which needs them. */
index_close(indexRel, NoLock);
/* Add quals for all columns from this index. */
for (i = 0; i < numatts; i++)
{
int attnum = index->indkey.values[i];
Oid type;
Oid op;
const char *colname;
/*
* Only include the column once regardless of how many times
* it shows up in how many indexes.
*
* This is also useful later to omit columns which can not
* have changed from the SET clause of the UPDATE statement.
*/
if (usedForQual[attnum - 1])
continue;
usedForQual[attnum - 1] = true;
/*
* Actually add the qual, ANDed with any others.
*/
if (foundUniqueIndex)
appendStringInfoString(&querybuf, " AND ");
colname = quote_identifier(NameStr((tupdesc->attrs[attnum - 1])->attname));
appendStringInfo(&querybuf, "y.%s ", colname);
type = attnumTypeId(matviewRel, attnum);
op = lookup_type_cache(type, TYPECACHE_EQ_OPR)->eq_opr;
mv_GenerateOper(&querybuf, op);
appendStringInfo(&querybuf, " x.%s", colname);
foundUniqueIndex = true;
}
}
ReleaseSysCache(indexTuple);
}
/*
* Callback for individual changed tuples
*/
static void
decoder_raw_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
Relation relation, ReorderBufferChange *change)
{
DecoderRawData *data;
MemoryContext old;
char replident = relation->rd_rel->relreplident;
bool is_rel_non_selective;
data = ctx->output_plugin_private;
if (data->isLocal) {
XTM_INFO("Skip action %d in transaction %u\n", change->action, txn->xid);
return;
}
XTM_INFO("Send action %d in transaction %u to replica\n", change->action, txn->xid);
/* Avoid leaking memory by using and resetting our own context */
old = MemoryContextSwitchTo(data->context);
/*
* Determine if relation is selective enough for WHERE clause generation
* in UPDATE and DELETE cases. A non-selective relation uses REPLICA
* IDENTITY set as NOTHING, or DEFAULT without an available replica
* identity index.
*/
RelationGetIndexList(relation);
is_rel_non_selective = (replident == REPLICA_IDENTITY_NOTHING ||
(replident == REPLICA_IDENTITY_DEFAULT &&
!OidIsValid(relation->rd_replidindex)));
/* Decode entry depending on its type */
switch (change->action)
{
case REORDER_BUFFER_CHANGE_INSERT:
if (change->data.tp.newtuple != NULL)
{
OutputPluginPrepareWrite(ctx, true);
decoder_raw_insert(ctx->out,
relation,
&change->data.tp.newtuple->tuple);
OutputPluginWrite(ctx, true);
}
break;
case REORDER_BUFFER_CHANGE_UPDATE:
if (!is_rel_non_selective)
{
HeapTuple oldtuple = change->data.tp.oldtuple != NULL ?
&change->data.tp.oldtuple->tuple : NULL;
HeapTuple newtuple = change->data.tp.newtuple != NULL ?
&change->data.tp.newtuple->tuple : NULL;
OutputPluginPrepareWrite(ctx, true);
decoder_raw_update(ctx->out,
relation,
oldtuple,
newtuple);
OutputPluginWrite(ctx, true);
}
break;
case REORDER_BUFFER_CHANGE_DELETE:
if (!is_rel_non_selective)
{
OutputPluginPrepareWrite(ctx, true);
decoder_raw_delete(ctx->out,
relation,
&change->data.tp.oldtuple->tuple);
OutputPluginWrite(ctx, true);
}
break;
default:
/* Should not come here */
Assert(0);
break;
}
MemoryContextSwitchTo(old);
MemoryContextReset(data->context);
}
/*
* get_relation_info -
* Retrieves catalog information for a given relation.
*
* Given the Oid of the relation, return the following info into fields
* of the RelOptInfo struct:
*
* min_attr lowest valid AttrNumber
* max_attr highest valid AttrNumber
* indexlist list of IndexOptInfos for relation's indexes
* pages number of pages
* tuples number of tuples
*
* Also, initialize the attr_needed[] and attr_widths[] arrays. In most
* cases these are left as zeroes, but sometimes we need to compute attr
* widths here, and we may as well cache the results for costsize.c.
*
* If inhparent is true, all we need to do is set up the attr arrays:
* the RelOptInfo actually represents the appendrel formed by an inheritance
* tree, and so the parent rel's physical size and index information isn't
* important for it.
*/
void
get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
RelOptInfo *rel)
{
Index varno = rel->relid;
Relation relation;
bool hasindex;
List *indexinfos = NIL;
bool needs_longlock;
/*
* We need not lock the relation since it was already locked, either by
* the rewriter or when expand_inherited_rtentry() added it to the query's
* rangetable.
*/
relation = heap_open(relationObjectId, NoLock);
needs_longlock = rel_needs_long_lock(relationObjectId);
rel->min_attr = FirstLowInvalidHeapAttributeNumber + 1;
rel->max_attr = RelationGetNumberOfAttributes(relation);
Assert(rel->max_attr >= rel->min_attr);
rel->attr_needed = (Relids *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(Relids));
rel->attr_widths = (int32 *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(int32));
/*
* CDB: Get partitioning key info for distributed relation.
*/
rel->cdbpolicy = RelationGetPartitioningKey(relation);
/*
* Estimate relation size --- unless it's an inheritance parent, in which
* case the size will be computed later in set_append_rel_pathlist, and we
* must leave it zero for now to avoid bollixing the total_table_pages
* calculation.
*/
if (!inhparent)
{
cdb_estimate_rel_size
(
rel,
relation,
relation,
rel->attr_widths - rel->min_attr,
&rel->pages,
&rel->tuples,
&rel->cdb_default_stats_used
);
}
/*
* Make list of indexes. Ignore indexes on system catalogs if told to.
* Don't bother with indexes for an inheritance parent, either.
*/
if (inhparent ||
(IgnoreSystemIndexes && IsSystemClass(relation->rd_rel)))
hasindex = false;
else
hasindex = relation->rd_rel->relhasindex;
if (hasindex)
{
List *indexoidlist;
ListCell *l;
LOCKMODE lmode;
/* Warn if indexed table needs ANALYZE. */
if (rel->cdb_default_stats_used)
cdb_default_stats_warning_for_table(relation->rd_id);
indexoidlist = RelationGetIndexList(relation);
/*
* For each index, we get the same type of lock that the executor will
* need, and do not release it. This saves a couple of trips to the
* shared lock manager while not creating any real loss of
* concurrency, because no schema changes could be happening on the
* index while we hold lock on the parent rel, and neither lock type
* blocks any other kind of index operation.
*/
if (rel->relid == root->parse->resultRelation)
lmode = RowExclusiveLock;
else
lmode = AccessShareLock;
foreach(l, indexoidlist)
{
Oid indexoid = lfirst_oid(l);
Relation indexRelation;
Form_pg_index index;
IndexOptInfo *info;
int ncolumns;
int i;
int16 amorderstrategy;
/*
* Extract info from the relation descriptor for the index.
*/
indexRelation = index_open(indexoid, lmode);
index = indexRelation->rd_index;
/*
* Ignore invalid indexes, since they can't safely be used for
* queries. Note that this is OK because the data structure we
* are constructing is only used by the planner --- the executor
* still needs to insert into "invalid" indexes!
*/
if (!index->indisvalid)
{
index_close(indexRelation, NoLock);
continue;
}
info = makeNode(IndexOptInfo);
info->indexoid = index->indexrelid;
info->rel = rel;
info->ncolumns = ncolumns = index->indnatts;
/*
* Need to make classlist and ordering arrays large enough to put
* a terminating 0 at the end of each one.
*/
info->indexkeys = (int *) palloc(sizeof(int) * ncolumns);
info->classlist = (Oid *) palloc0(sizeof(Oid) * (ncolumns + 1));
info->ordering = (Oid *) palloc0(sizeof(Oid) * (ncolumns + 1));
for (i = 0; i < ncolumns; i++)
{
info->classlist[i] = indexRelation->rd_indclass->values[i];
info->indexkeys[i] = index->indkey.values[i];
}
info->relam = indexRelation->rd_rel->relam;
info->amcostestimate = indexRelation->rd_am->amcostestimate;
info->amoptionalkey = indexRelation->rd_am->amoptionalkey;
/*
* Fetch the ordering operators associated with the index, if any.
*/
amorderstrategy = indexRelation->rd_am->amorderstrategy;
if (amorderstrategy != 0)
{
int oprindex = amorderstrategy - 1;
for (i = 0; i < ncolumns; i++)
{
info->ordering[i] = indexRelation->rd_operator[oprindex];
oprindex += indexRelation->rd_am->amstrategies;
}
}
/*
* Fetch the index expressions and predicate, if any. We must
* modify the copies we obtain from the relcache to have the
* correct varno for the parent relation, so that they match up
* correctly against qual clauses.
*/
info->indexprs = RelationGetIndexExpressions(indexRelation);
info->indpred = RelationGetIndexPredicate(indexRelation);
if (info->indexprs && varno != 1)
ChangeVarNodes((Node *) info->indexprs, 1, varno, 0);
if (info->indpred && varno != 1)
ChangeVarNodes((Node *) info->indpred, 1, varno, 0);
info->predOK = false; /* set later in indxpath.c */
info->unique = index->indisunique;
/*
* Estimate the index size. If it's not a partial index, we lock
* the number-of-tuples estimate to equal the parent table; if it
* is partial then we have to use the same methods as we would for
* a table, except we can be sure that the index is not larger
* than the table.
*/
cdb_estimate_rel_size(rel,
relation,
indexRelation,
NULL,
&info->pages,
&info->tuples,
&info->cdb_default_stats_used);
if (!info->indpred ||
info->tuples > rel->tuples)
info->tuples = rel->tuples;
if (info->cdb_default_stats_used &&
!rel->cdb_default_stats_used)
cdb_default_stats_warning_for_index(relation->rd_id, indexoid);
index_close(indexRelation, needs_longlock ? NoLock : lmode);
indexinfos = lcons(info, indexinfos);
}
list_free(indexoidlist);
}
/*
* refresh_by_match_merge
*
* Refresh a materialized view with transactional semantics, while allowing
* concurrent reads.
*
* This is called after a new version of the data has been created in a
* temporary table. It performs a full outer join against the old version of
* the data, producing "diff" results. This join cannot work if there are any
* duplicated rows in either the old or new versions, in the sense that every
* column would compare as equal between the two rows. It does work correctly
* in the face of rows which have at least one NULL value, with all non-NULL
* columns equal. The behavior of NULLs on equality tests and on UNIQUE
* indexes turns out to be quite convenient here; the tests we need to make
* are consistent with default behavior. If there is at least one UNIQUE
* index on the materialized view, we have exactly the guarantee we need.
*
* The temporary table used to hold the diff results contains just the TID of
* the old record (if matched) and the ROW from the new table as a single
* column of complex record type (if matched).
*
* Once we have the diff table, we perform set-based DELETE and INSERT
* operations against the materialized view, and discard both temporary
* tables.
*
* Everything from the generation of the new data to applying the differences
* takes place under cover of an ExclusiveLock, since it seems as though we
* would want to prohibit not only concurrent REFRESH operations, but also
* incremental maintenance. It also doesn't seem reasonable or safe to allow
* SELECT FOR UPDATE or SELECT FOR SHARE on rows being updated or deleted by
* this command.
*/
static void
refresh_by_match_merge(Oid matviewOid, Oid tempOid, Oid relowner,
int save_sec_context)
{
StringInfoData querybuf;
Relation matviewRel;
Relation tempRel;
char *matviewname;
char *tempname;
char *diffname;
TupleDesc tupdesc;
bool foundUniqueIndex;
List *indexoidlist;
ListCell *indexoidscan;
int16 relnatts;
bool *usedForQual;
initStringInfo(&querybuf);
matviewRel = heap_open(matviewOid, NoLock);
matviewname = quote_qualified_identifier(get_namespace_name(RelationGetNamespace(matviewRel)),
RelationGetRelationName(matviewRel));
tempRel = heap_open(tempOid, NoLock);
tempname = quote_qualified_identifier(get_namespace_name(RelationGetNamespace(tempRel)),
RelationGetRelationName(tempRel));
diffname = make_temptable_name_n(tempname, 2);
relnatts = matviewRel->rd_rel->relnatts;
usedForQual = (bool *) palloc0(sizeof(bool) * relnatts);
/* Open SPI context. */
if (SPI_connect() != SPI_OK_CONNECT)
elog(ERROR, "SPI_connect failed");
/* Analyze the temp table with the new contents. */
appendStringInfo(&querybuf, "ANALYZE %s", tempname);
if (SPI_exec(querybuf.data, 0) != SPI_OK_UTILITY)
elog(ERROR, "SPI_exec failed: %s", querybuf.data);
/*
* We need to ensure that there are not duplicate rows without NULLs in
* the new data set before we can count on the "diff" results. Check for
* that in a way that allows showing the first duplicated row found. Even
* after we pass this test, a unique index on the materialized view may
* find a duplicate key problem.
*/
resetStringInfo(&querybuf);
appendStringInfo(&querybuf,
"SELECT newdata FROM %s newdata "
"WHERE newdata IS NOT NULL AND EXISTS "
"(SELECT * FROM %s newdata2 WHERE newdata2 IS NOT NULL "
"AND newdata2 OPERATOR(pg_catalog.*=) newdata "
"AND newdata2.ctid OPERATOR(pg_catalog.<>) "
"newdata.ctid) LIMIT 1",
tempname, tempname);
if (SPI_execute(querybuf.data, false, 1) != SPI_OK_SELECT)
elog(ERROR, "SPI_exec failed: %s", querybuf.data);
if (SPI_processed > 0)
{
ereport(ERROR,
(errcode(ERRCODE_CARDINALITY_VIOLATION),
errmsg("new data for \"%s\" contains duplicate rows without any null columns",
RelationGetRelationName(matviewRel)),
errdetail("Row: %s",
SPI_getvalue(SPI_tuptable->vals[0], SPI_tuptable->tupdesc, 1))));
}
SetUserIdAndSecContext(relowner,
save_sec_context | SECURITY_LOCAL_USERID_CHANGE);
/* Start building the query for creating the diff table. */
resetStringInfo(&querybuf);
appendStringInfo(&querybuf,
"CREATE TEMP TABLE %s AS "
"SELECT mv.ctid AS tid, newdata "
"FROM %s mv FULL JOIN %s newdata ON (",
diffname, matviewname, tempname);
/*
* Get the list of index OIDs for the table from the relcache, and look up
* each one in the pg_index syscache. We will test for equality on all
* columns present in all unique indexes which only reference columns and
* include all rows.
*/
tupdesc = matviewRel->rd_att;
foundUniqueIndex = false;
indexoidlist = RelationGetIndexList(matviewRel);
foreach(indexoidscan, indexoidlist)
{
Oid indexoid = lfirst_oid(indexoidscan);
Relation indexRel;
Form_pg_index indexStruct;
indexRel = index_open(indexoid, RowExclusiveLock);
indexStruct = indexRel->rd_index;
/*
* We're only interested if it is unique, valid, contains no
* expressions, and is not partial.
*/
if (indexStruct->indisunique &&
IndexIsValid(indexStruct) &&
RelationGetIndexExpressions(indexRel) == NIL &&
RelationGetIndexPredicate(indexRel) == NIL)
{
int numatts = indexStruct->indnatts;
int i;
/* Add quals for all columns from this index. */
for (i = 0; i < numatts; i++)
{
int attnum = indexStruct->indkey.values[i];
Oid type;
Oid op;
const char *colname;
/*
* Only include the column once regardless of how many times
* it shows up in how many indexes.
*/
if (usedForQual[attnum - 1])
continue;
usedForQual[attnum - 1] = true;
/*
* Actually add the qual, ANDed with any others.
*/
if (foundUniqueIndex)
appendStringInfoString(&querybuf, " AND ");
colname = quote_identifier(NameStr((tupdesc->attrs[attnum - 1])->attname));
appendStringInfo(&querybuf, "newdata.%s ", colname);
type = attnumTypeId(matviewRel, attnum);
op = lookup_type_cache(type, TYPECACHE_EQ_OPR)->eq_opr;
mv_GenerateOper(&querybuf, op);
appendStringInfo(&querybuf, " mv.%s", colname);
foundUniqueIndex = true;
}
}
/* Keep the locks, since we're about to run DML which needs them. */
index_close(indexRel, NoLock);
}
/*
* reindex_relation - This routine is used to recreate all indexes
* of a relation (and its toast relation too, if any).
*
* Returns true if any indexes were rebuilt.
*/
bool
reindex_relation(Oid relid)
{
Relation rel;
Oid toast_relid;
bool is_pg_class;
bool result;
List *indexIds,
*doneIndexes,
*indexId;
/*
* Ensure to hold an exclusive lock throughout the transaction. The
* lock could perhaps be less intensive (in the non-overwrite case)
* but for now it's AccessExclusiveLock for simplicity.
*/
rel = heap_open(relid, AccessExclusiveLock);
toast_relid = rel->rd_rel->reltoastrelid;
/*
* Get the list of index OIDs for this relation. (We trust to the
* relcache to get this with a sequential scan if ignoring system
* indexes.)
*/
indexIds = RelationGetIndexList(rel);
/*
* reindex_index will attempt to update the pg_class rows for the
* relation and index. If we are processing pg_class itself, we
* want to make sure that the updates do not try to insert index
* entries into indexes we have not processed yet. (When we are
* trying to recover from corrupted indexes, that could easily
* cause a crash.) We can accomplish this because CatalogUpdateIndexes
* will use the relcache's index list to know which indexes to update.
* We just force the index list to be only the stuff we've processed.
*
* It is okay to not insert entries into the indexes we have not
* processed yet because all of this is transaction-safe. If we fail
* partway through, the updated rows are dead and it doesn't matter
* whether they have index entries. Also, a new pg_class index will
* be created with an entry for its own pg_class row because we do
* setNewRelfilenode() before we do index_build().
*/
is_pg_class = (RelationGetRelid(rel) == RelOid_pg_class);
doneIndexes = NIL;
/* Reindex all the indexes. */
foreach(indexId, indexIds)
{
Oid indexOid = lfirsto(indexId);
if (is_pg_class)
RelationSetIndexList(rel, doneIndexes);
reindex_index(indexOid);
CommandCounterIncrement();
if (is_pg_class)
doneIndexes = lappendo(doneIndexes, indexOid);
}
/*
* get_relation_info -
* Retrieves catalog information for a given relation.
*
* Given the Oid of the relation, return the following info into fields
* of the RelOptInfo struct:
*
* min_attr lowest valid AttrNumber
* max_attr highest valid AttrNumber
* indexlist list of IndexOptInfos for relation's indexes
* pages number of pages
* tuples number of tuples
*
* Also, initialize the attr_needed[] and attr_widths[] arrays. In most
* cases these are left as zeroes, but sometimes we need to compute attr
* widths here, and we may as well cache the results for costsize.c.
*
* If inhparent is true, all we need to do is set up the attr arrays:
* the RelOptInfo actually represents the appendrel formed by an inheritance
* tree, and so the parent rel's physical size and index information isn't
* important for it.
*/
void
get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent,
RelOptInfo *rel)
{
Index varno = rel->relid;
Relation relation;
bool hasindex;
List *indexinfos = NIL;
/*
* We need not lock the relation since it was already locked, either by
* the rewriter or when expand_inherited_rtentry() added it to the query's
* rangetable.
*/
relation = heap_open(relationObjectId, NoLock);
rel->min_attr = FirstLowInvalidHeapAttributeNumber + 1;
rel->max_attr = RelationGetNumberOfAttributes(relation);
rel->reltablespace = RelationGetForm(relation)->reltablespace;
Assert(rel->max_attr >= rel->min_attr);
rel->attr_needed = (Relids *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(Relids));
rel->attr_widths = (int32 *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(int32));
/*
* Estimate relation size --- unless it's an inheritance parent, in which
* case the size will be computed later in set_append_rel_pathlist, and we
* must leave it zero for now to avoid bollixing the total_table_pages
* calculation.
*/
if (!inhparent)
estimate_rel_size(relation, rel->attr_widths - rel->min_attr,
&rel->pages, &rel->tuples);
/*
* Make list of indexes. Ignore indexes on system catalogs if told to.
* Don't bother with indexes for an inheritance parent, either.
*/
if (inhparent ||
(IgnoreSystemIndexes && IsSystemClass(relation->rd_rel)))
hasindex = false;
else
hasindex = relation->rd_rel->relhasindex;
if (hasindex)
{
List *indexoidlist;
ListCell *l;
LOCKMODE lmode;
indexoidlist = RelationGetIndexList(relation);
/*
* For each index, we get the same type of lock that the executor will
* need, and do not release it. This saves a couple of trips to the
* shared lock manager while not creating any real loss of
* concurrency, because no schema changes could be happening on the
* index while we hold lock on the parent rel, and neither lock type
* blocks any other kind of index operation.
*/
if (rel->relid == root->parse->resultRelation)
lmode = RowExclusiveLock;
else
lmode = AccessShareLock;
foreach(l, indexoidlist)
{
Oid indexoid = lfirst_oid(l);
Relation indexRelation;
Form_pg_index index;
IndexOptInfo *info;
int ncolumns;
int i;
/*
* Extract info from the relation descriptor for the index.
*/
indexRelation = index_open(indexoid, lmode);
index = indexRelation->rd_index;
/*
* Ignore invalid indexes, since they can't safely be used for
* queries. Note that this is OK because the data structure we
* are constructing is only used by the planner --- the executor
* still needs to insert into "invalid" indexes!
*/
if (!index->indisvalid)
{
index_close(indexRelation, NoLock);
continue;
}
/*
* If the index is valid, but cannot yet be used, ignore it; but
* mark the plan we are generating as transient. See
* src/backend/access/heap/README.HOT for discussion.
*/
if (index->indcheckxmin &&
!TransactionIdPrecedes(HeapTupleHeaderGetXmin(indexRelation->rd_indextuple->t_data),
TransactionXmin))
{
root->glob->transientPlan = true;
index_close(indexRelation, NoLock);
continue;
}
info = makeNode(IndexOptInfo);
info->indexoid = index->indexrelid;
info->reltablespace =
RelationGetForm(indexRelation)->reltablespace;
info->rel = rel;
info->ncolumns = ncolumns = index->indnatts;
/*
* Allocate per-column info arrays. To save a few palloc cycles
* we allocate all the Oid-type arrays in one request. Note that
* the opfamily array needs an extra, terminating zero at the end.
* We pre-zero the ordering info in case the index is unordered.
*/
info->indexkeys = (int *) palloc(sizeof(int) * ncolumns);
info->opfamily = (Oid *) palloc0(sizeof(Oid) * (4 * ncolumns + 1));
info->opcintype = info->opfamily + (ncolumns + 1);
info->fwdsortop = info->opcintype + ncolumns;
info->revsortop = info->fwdsortop + ncolumns;
info->nulls_first = (bool *) palloc0(sizeof(bool) * ncolumns);
for (i = 0; i < ncolumns; i++)
{
info->indexkeys[i] = index->indkey.values[i];
info->opfamily[i] = indexRelation->rd_opfamily[i];
info->opcintype[i] = indexRelation->rd_opcintype[i];
}
info->relam = indexRelation->rd_rel->relam;
info->amcostestimate = indexRelation->rd_am->amcostestimate;
info->amoptionalkey = indexRelation->rd_am->amoptionalkey;
info->amsearchnulls = indexRelation->rd_am->amsearchnulls;
info->amhasgettuple = OidIsValid(indexRelation->rd_am->amgettuple);
info->amhasgetbitmap = OidIsValid(indexRelation->rd_am->amgetbitmap);
/*
* Fetch the ordering operators associated with the index, if any.
* We expect that all ordering-capable indexes use btree's
* strategy numbers for the ordering operators.
*/
if (indexRelation->rd_am->amcanorder)
{
int nstrat = indexRelation->rd_am->amstrategies;
for (i = 0; i < ncolumns; i++)
{
int16 opt = indexRelation->rd_indoption[i];
int fwdstrat;
int revstrat;
if (opt & INDOPTION_DESC)
{
fwdstrat = BTGreaterStrategyNumber;
revstrat = BTLessStrategyNumber;
}
else
{
fwdstrat = BTLessStrategyNumber;
revstrat = BTGreaterStrategyNumber;
}
/*
* Index AM must have a fixed set of strategies for it to
* make sense to specify amcanorder, so we need not allow
* the case amstrategies == 0.
*/
if (fwdstrat > 0)
{
Assert(fwdstrat <= nstrat);
info->fwdsortop[i] = indexRelation->rd_operator[i * nstrat + fwdstrat - 1];
}
if (revstrat > 0)
{
Assert(revstrat <= nstrat);
info->revsortop[i] = indexRelation->rd_operator[i * nstrat + revstrat - 1];
}
info->nulls_first[i] = (opt & INDOPTION_NULLS_FIRST) != 0;
}
}
/*
* Fetch the index expressions and predicate, if any. We must
* modify the copies we obtain from the relcache to have the
* correct varno for the parent relation, so that they match up
* correctly against qual clauses.
*/
info->indexprs = RelationGetIndexExpressions(indexRelation);
info->indpred = RelationGetIndexPredicate(indexRelation);
if (info->indexprs && varno != 1)
ChangeVarNodes((Node *) info->indexprs, 1, varno, 0);
if (info->indpred && varno != 1)
ChangeVarNodes((Node *) info->indpred, 1, varno, 0);
info->predOK = false; /* set later in indxpath.c */
info->unique = index->indisunique;
/*
* Estimate the index size. If it's not a partial index, we lock
* the number-of-tuples estimate to equal the parent table; if it
* is partial then we have to use the same methods as we would for
* a table, except we can be sure that the index is not larger
* than the table.
*/
if (info->indpred == NIL)
{
info->pages = RelationGetNumberOfBlocks(indexRelation);
info->tuples = rel->tuples;
}
else
{
estimate_rel_size(indexRelation, NULL,
&info->pages, &info->tuples);
if (info->tuples > rel->tuples)
info->tuples = rel->tuples;
}
index_close(indexRelation, NoLock);
indexinfos = lcons(info, indexinfos);
}
list_free(indexoidlist);
}
/*
* get_relation_info -
* Retrieves catalog information for a given relation.
*
* Given the Oid of the relation, return the following info into fields
* of the RelOptInfo struct:
*
* min_attr lowest valid AttrNumber
* max_attr highest valid AttrNumber
* indexlist list of IndexOptInfos for relation's indexes
* pages number of pages
* tuples number of tuples
*
* Also, initialize the attr_needed[] and attr_widths[] arrays. In most
* cases these are left as zeroes, but sometimes we need to compute attr
* widths here, and we may as well cache the results for costsize.c.
*/
void
get_relation_info(Oid relationObjectId, RelOptInfo *rel)
{
Index varno = rel->relid;
Relation relation;
bool hasindex;
List *indexinfos = NIL;
/*
* Normally, we can assume the rewriter already acquired at least
* AccessShareLock on each relation used in the query. However this will
* not be the case for relations added to the query because they are
* inheritance children of some relation mentioned explicitly. For them,
* this is the first access during the parse/rewrite/plan pipeline, and so
* we need to obtain and keep a suitable lock.
*
* XXX really, a suitable lock is RowShareLock if the relation is an
* UPDATE/DELETE target, and AccessShareLock otherwise. However we cannot
* easily tell here which to get, so for the moment just get
* AccessShareLock always. The executor will get the right lock when it
* runs, which means there is a very small chance of deadlock trying to
* upgrade our lock.
*/
if (rel->reloptkind == RELOPT_BASEREL)
relation = heap_open(relationObjectId, NoLock);
else
relation = heap_open(relationObjectId, AccessShareLock);
rel->min_attr = FirstLowInvalidHeapAttributeNumber + 1;
rel->max_attr = RelationGetNumberOfAttributes(relation);
Assert(rel->max_attr >= rel->min_attr);
rel->attr_needed = (Relids *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(Relids));
rel->attr_widths = (int32 *)
palloc0((rel->max_attr - rel->min_attr + 1) * sizeof(int32));
/*
* Estimate relation size.
*/
estimate_rel_size(relation, rel->attr_widths - rel->min_attr,
&rel->pages, &rel->tuples);
/*
* Make list of indexes. Ignore indexes on system catalogs if told to.
*/
if (IsIgnoringSystemIndexes() && IsSystemClass(relation->rd_rel))
hasindex = false;
else
hasindex = relation->rd_rel->relhasindex;
if (hasindex)
{
List *indexoidlist;
ListCell *l;
indexoidlist = RelationGetIndexList(relation);
foreach(l, indexoidlist)
{
Oid indexoid = lfirst_oid(l);
Relation indexRelation;
Form_pg_index index;
IndexOptInfo *info;
int ncolumns;
int i;
int16 amorderstrategy;
/*
* Extract info from the relation descriptor for the index.
*
* Note that we take no lock on the index; we assume our lock on
* the parent table will protect the index's schema information.
* When and if the executor actually uses the index, it will take
* a lock as needed to protect the access to the index contents.
*/
indexRelation = index_open(indexoid);
index = indexRelation->rd_index;
info = makeNode(IndexOptInfo);
info->indexoid = index->indexrelid;
info->rel = rel;
info->ncolumns = ncolumns = index->indnatts;
/*
* Need to make classlist and ordering arrays large enough to put
* a terminating 0 at the end of each one.
*/
info->indexkeys = (int *) palloc(sizeof(int) * ncolumns);
info->classlist = (Oid *) palloc0(sizeof(Oid) * (ncolumns + 1));
info->ordering = (Oid *) palloc0(sizeof(Oid) * (ncolumns + 1));
for (i = 0; i < ncolumns; i++)
{
info->classlist[i] = indexRelation->rd_indclass->values[i];
info->indexkeys[i] = index->indkey.values[i];
}
info->relam = indexRelation->rd_rel->relam;
info->amcostestimate = indexRelation->rd_am->amcostestimate;
info->amoptionalkey = indexRelation->rd_am->amoptionalkey;
/*
* Fetch the ordering operators associated with the index, if any.
*/
amorderstrategy = indexRelation->rd_am->amorderstrategy;
if (amorderstrategy != 0)
{
int oprindex = amorderstrategy - 1;
for (i = 0; i < ncolumns; i++)
{
info->ordering[i] = indexRelation->rd_operator[oprindex];
oprindex += indexRelation->rd_am->amstrategies;
}
}
/*
* Fetch the index expressions and predicate, if any. We must
* modify the copies we obtain from the relcache to have the
* correct varno for the parent relation, so that they match up
* correctly against qual clauses.
*/
info->indexprs = RelationGetIndexExpressions(indexRelation);
info->indpred = RelationGetIndexPredicate(indexRelation);
if (info->indexprs && varno != 1)
ChangeVarNodes((Node *) info->indexprs, 1, varno, 0);
if (info->indpred && varno != 1)
ChangeVarNodes((Node *) info->indpred, 1, varno, 0);
info->predOK = false; /* set later in indxpath.c */
info->unique = index->indisunique;
/*
* Estimate the index size. If it's not a partial index, we lock
* the number-of-tuples estimate to equal the parent table; if it
* is partial then we have to use the same methods as we would for
* a table, except we can be sure that the index is not larger
* than the table.
*/
if (info->indpred == NIL)
{
info->pages = RelationGetNumberOfBlocks(indexRelation);
info->tuples = rel->tuples;
}
else
{
estimate_rel_size(indexRelation, NULL,
&info->pages, &info->tuples);
if (info->tuples > rel->tuples)
info->tuples = rel->tuples;
}
index_close(indexRelation);
indexinfos = lcons(info, indexinfos);
}
list_free(indexoidlist);
}
/*
* ErrorIfNotSupportedConstraint run checks related to unique index / exclude
* constraints.
*
* Forbid UNIQUE, PRIMARY KEY, or EXCLUDE constraints on append partitioned
* tables, since currently there is no way of enforcing uniqueness for
* overlapping shards.
*
* Similarly, do not allow such constraints if they do not include partition
* column. This check is important for two reasons:
* i. First, currently Citus does not enforce uniqueness constraint on multiple
* shards.
* ii. Second, INSERT INTO .. ON CONFLICT (i.e., UPSERT) queries can be executed
* with no further check for constraints.
*/
static void
ErrorIfNotSupportedConstraint(Relation relation, char distributionMethod,
Var *distributionColumn, uint32 colocationId)
{
char *relationName = RelationGetRelationName(relation);
List *indexOidList = RelationGetIndexList(relation);
ListCell *indexOidCell = NULL;
foreach(indexOidCell, indexOidList)
{
Oid indexOid = lfirst_oid(indexOidCell);
Relation indexDesc = index_open(indexOid, RowExclusiveLock);
IndexInfo *indexInfo = NULL;
AttrNumber *attributeNumberArray = NULL;
bool hasDistributionColumn = false;
int attributeCount = 0;
int attributeIndex = 0;
/* extract index key information from the index's pg_index info */
indexInfo = BuildIndexInfo(indexDesc);
/* only check unique indexes and exclusion constraints. */
if (indexInfo->ii_Unique == false && indexInfo->ii_ExclusionOps == NULL)
{
index_close(indexDesc, NoLock);
continue;
}
/*
* Citus cannot enforce uniqueness/exclusion constraints with overlapping shards.
* Thus, emit a warning for unique indexes and exclusion constraints on
* append partitioned tables.
*/
if (distributionMethod == DISTRIBUTE_BY_APPEND)
{
ereport(WARNING, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("table \"%s\" has a UNIQUE or EXCLUDE constraint",
relationName),
errdetail("UNIQUE constraints, EXCLUDE constraints, "
"and PRIMARY KEYs on "
"append-partitioned tables cannot be enforced."),
errhint("Consider using hash partitioning.")));
}
attributeCount = indexInfo->ii_NumIndexAttrs;
attributeNumberArray = indexInfo->ii_KeyAttrNumbers;
for (attributeIndex = 0; attributeIndex < attributeCount; attributeIndex++)
{
AttrNumber attributeNumber = attributeNumberArray[attributeIndex];
bool uniqueConstraint = false;
bool exclusionConstraintWithEquality = false;
if (distributionColumn->varattno != attributeNumber)
{
continue;
}
uniqueConstraint = indexInfo->ii_Unique;
exclusionConstraintWithEquality = (indexInfo->ii_ExclusionOps != NULL &&
OperatorImplementsEquality(
indexInfo->ii_ExclusionOps[
attributeIndex]));
if (uniqueConstraint || exclusionConstraintWithEquality)
{
hasDistributionColumn = true;
break;
}
}
if (!hasDistributionColumn)
{
ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot distribute relation: \"%s\"",
relationName),
errdetail("Distributed relations cannot have UNIQUE, "
"EXCLUDE, or PRIMARY KEY constraints that do not "
"include the partition column (with an equality "
"operator if EXCLUDE).")));
}
index_close(indexDesc, NoLock);
}
Datum
triggered_change_notification(PG_FUNCTION_ARGS)
{
TriggerData *trigdata = (TriggerData *) fcinfo->context;
Trigger *trigger;
int nargs;
HeapTuple trigtuple;
Relation rel;
TupleDesc tupdesc;
char *channel;
char operation;
StringInfo payload = makeStringInfo();
bool foundPK;
List *indexoidlist;
ListCell *indexoidscan;
/* make sure it's called as a trigger */
if (!CALLED_AS_TRIGGER(fcinfo))
ereport(ERROR,
(errcode(ERRCODE_E_R_I_E_TRIGGER_PROTOCOL_VIOLATED),
errmsg("triggered_change_notification: must be called as trigger")));
/* and that it's called after the change */
if (!TRIGGER_FIRED_AFTER(trigdata->tg_event))
ereport(ERROR,
(errcode(ERRCODE_E_R_I_E_TRIGGER_PROTOCOL_VIOLATED),
errmsg("triggered_change_notification: must be called after the change")));
/* and that it's called for each row */
if (!TRIGGER_FIRED_FOR_ROW(trigdata->tg_event))
ereport(ERROR,
(errcode(ERRCODE_E_R_I_E_TRIGGER_PROTOCOL_VIOLATED),
errmsg("triggered_change_notification: must be called for each row")));
if (TRIGGER_FIRED_BY_INSERT(trigdata->tg_event))
operation = 'I';
else if (TRIGGER_FIRED_BY_UPDATE(trigdata->tg_event))
operation = 'U';
else if (TRIGGER_FIRED_BY_DELETE(trigdata->tg_event))
operation = 'D';
else
{
elog(ERROR, "triggered_change_notification: trigger fired by unrecognized operation");
operation = 'X'; /* silence compiler warning */
}
trigger = trigdata->tg_trigger;
nargs = trigger->tgnargs;
if (nargs > 1)
ereport(ERROR,
(errcode(ERRCODE_E_R_I_E_TRIGGER_PROTOCOL_VIOLATED),
errmsg("triggered_change_notification: must not be called with more than one parameter")));
if (nargs == 0)
channel = "tcn";
else
channel = trigger->tgargs[0];
/* get tuple data */
trigtuple = trigdata->tg_trigtuple;
rel = trigdata->tg_relation;
tupdesc = rel->rd_att;
foundPK = false;
/*
* Get the list of index OIDs for the table from the relcache, and look up
* each one in the pg_index syscache until we find one marked primary key
* (hopefully there isn't more than one such).
*/
indexoidlist = RelationGetIndexList(rel);
foreach(indexoidscan, indexoidlist)
{
Oid indexoid = lfirst_oid(indexoidscan);
HeapTuple indexTuple;
Form_pg_index index;
indexTuple = SearchSysCache1(INDEXRELID, ObjectIdGetDatum(indexoid));
if (!HeapTupleIsValid(indexTuple)) /* should not happen */
elog(ERROR, "cache lookup failed for index %u", indexoid);
index = (Form_pg_index) GETSTRUCT(indexTuple);
/* we're only interested if it is the primary key and valid */
if (index->indisprimary && IndexIsValid(index))
{
int numatts = index->indnatts;
if (numatts > 0)
{
int i;
foundPK = true;
strcpy_quoted(payload, RelationGetRelationName(rel), '"');
appendStringInfoCharMacro(payload, ',');
appendStringInfoCharMacro(payload, operation);
for (i = 0; i < numatts; i++)
{
int colno = index->indkey.values[i];
appendStringInfoCharMacro(payload, ',');
strcpy_quoted(payload, NameStr((tupdesc->attrs[colno - 1])->attname), '"');
appendStringInfoCharMacro(payload, '=');
strcpy_quoted(payload, SPI_getvalue(trigtuple, tupdesc, colno), '\'');
}
Async_Notify(channel, payload->data);
}
ReleaseSysCache(indexTuple);
break;
}
ReleaseSysCache(indexTuple);
}
