Simplify some more

use-plancontext-more
Jelte Fennema 2020-01-03 16:05:31 +01:00
parent 093be287e2
commit 026db60075
4 changed files with 108 additions and 137 deletions

View File

@ -66,34 +66,10 @@ static uint64 NextPlanId = 1;
/* keep track of planner call stack levels */ /* keep track of planner call stack levels */
int PlannerLevel = 0; int PlannerLevel = 0;
typedef struct DistributedPlanningContext
{
/* The parsed query that is given to the planner. It is a slightly modified
* to work with the standard_planner */
Query *parse;
/* A copy of the original parsed query that is given to the planner. This
* doesn't contain most of the changes that are made to parse. There's one
* that change that is made for non fast path router queries though, which
* is the assigning of RTE identities using AssignRTEIdentities. This is
* NULL for non distributed plans, since those don't need it. */
Query *originalQuery;
/* the cursor options given to the planner */
int cursorOptions;
/* the ParamListInfo that is given to the planner */
ParamListInfo boundParams;
/* Our custom restriction context */
PlannerRestrictionContext *plannerRestrictionContext;
} DistributedPlanningContext;
static bool ListContainsDistributedTableRTE(List *rangeTableList); static bool ListContainsDistributedTableRTE(List *rangeTableList);
static bool IsUpdateOrDelete(Query *query); static bool IsUpdateOrDelete(Query *query);
static PlannedStmt * CreateDistributedPlannedStmt(PlannedStmt *localPlan, static PlannedStmt * CreateDistributedPlannedStmt(
DistributedPlanningContext *ctx); DistributedPlanningContext *planContext);
static DistributedPlan * CreateDistributedPlan(uint64 planId, Query *originalQuery, static DistributedPlan * CreateDistributedPlan(uint64 planId, Query *originalQuery,
Query *query, ParamListInfo boundParams, Query *query, ParamListInfo boundParams,
bool hasUnresolvedParams, bool hasUnresolvedParams,
@ -108,8 +84,6 @@ static int AssignRTEIdentities(List *rangeTableList, int rteIdCounter);
static void AssignRTEIdentity(RangeTblEntry *rangeTableEntry, int rteIdentifier); static void AssignRTEIdentity(RangeTblEntry *rangeTableEntry, int rteIdentifier);
static void AdjustPartitioningForDistributedPlanning(List *rangeTableList, static void AdjustPartitioningForDistributedPlanning(List *rangeTableList,
bool setPartitionedTablesInherited); bool setPartitionedTablesInherited);
static PlannedStmt * FinalizePlan(PlannedStmt *localPlan,
DistributedPlan *distributedPlan);
static PlannedStmt * FinalizeNonRouterPlan(PlannedStmt *localPlan, static PlannedStmt * FinalizeNonRouterPlan(PlannedStmt *localPlan,
DistributedPlan *distributedPlan, DistributedPlan *distributedPlan,
CustomScan *customScan); CustomScan *customScan);
@ -137,14 +111,10 @@ static bool HasUnresolvedExternParamsWalker(Node *expression, ParamListInfo boun
static bool IsLocalReferenceTableJoin(Query *parse, List *rangeTableList); static bool IsLocalReferenceTableJoin(Query *parse, List *rangeTableList);
static bool QueryIsNotSimpleSelect(Node *node); static bool QueryIsNotSimpleSelect(Node *node);
static bool UpdateReferenceTablesWithShard(Node *node, void *context); static bool UpdateReferenceTablesWithShard(Node *node, void *context);
static PlannedStmt * PlanFastPathDistributedStmt(DistributedPlanningContext *ctx, static PlannedStmt * PlanFastPathDistributedStmt(DistributedPlanningContext *planContext,
Const *distributionKeyValue); Const *distributionKeyValue);
static PlannedStmt * PlanDistributedStmt(DistributedPlanningContext *ctx, static PlannedStmt * PlanDistributedStmt(DistributedPlanningContext *planContext,
List *rangeTableList, int rteIdCounter); List *rangeTableList, int rteIdCounter);
static PlannedStmt * PlanDelegatedFunctionCall(DistributedPlanningContext *ctx,
DistributedPlan *delegatedPlan);
static PlannedStmt * PlanNonDistributedStmt(DistributedPlanningContext *ctx,
bool hasExternParam);
/* Distributed planner hook */ /* Distributed planner hook */
PlannedStmt * PlannedStmt *
@ -157,7 +127,7 @@ distributed_planner(Query *parse, int cursorOptions, ParamListInfo boundParams)
int rteIdCounter = 1; int rteIdCounter = 1;
bool fastPathRouterQuery = false; bool fastPathRouterQuery = false;
Const *distributionKeyValue = NULL; Const *distributionKeyValue = NULL;
DistributedPlanningContext ctx = { DistributedPlanningContext planContext = {
.parse = parse, .parse = parse,
.cursorOptions = cursorOptions, .cursorOptions = cursorOptions,
.boundParams = boundParams, .boundParams = boundParams,
@ -200,7 +170,7 @@ distributed_planner(Query *parse, int cursorOptions, ParamListInfo boundParams)
* too, but for those it needs to be done AFTER calling * too, but for those it needs to be done AFTER calling
* AssignRTEIdentities. * AssignRTEIdentities.
*/ */
ctx.originalQuery = copyObject(parse); planContext.originalQuery = copyObject(parse);
} }
else if (needsDistributedPlanning) else if (needsDistributedPlanning)
{ {
@ -226,7 +196,7 @@ distributed_planner(Query *parse, int cursorOptions, ParamListInfo boundParams)
* to be present in the copied query too. * to be present in the copied query too.
*/ */
rteIdCounter = AssignRTEIdentities(rangeTableList, rteIdCounter); rteIdCounter = AssignRTEIdentities(rangeTableList, rteIdCounter);
ctx.originalQuery = copyObject(parse); planContext.originalQuery = copyObject(parse);
setPartitionedTablesInherited = false; setPartitionedTablesInherited = false;
AdjustPartitioningForDistributedPlanning(rangeTableList, AdjustPartitioningForDistributedPlanning(rangeTableList,
@ -240,7 +210,7 @@ distributed_planner(Query *parse, int cursorOptions, ParamListInfo boundParams)
ReplaceTableVisibleFunction((Node *) parse); ReplaceTableVisibleFunction((Node *) parse);
/* create a restriction context and put it at the end if context list */ /* create a restriction context and put it at the end if context list */
ctx.plannerRestrictionContext = CreateAndPushPlannerRestrictionContext(); planContext.plannerRestrictionContext = CreateAndPushPlannerRestrictionContext();
/* /*
* We keep track of how many times we've recursed into the planner, primarily * We keep track of how many times we've recursed into the planner, primarily
@ -253,25 +223,28 @@ distributed_planner(Query *parse, int cursorOptions, ParamListInfo boundParams)
PG_TRY(); PG_TRY();
{ {
DistributedPlan *delegatedPlan = NULL;
bool hasExternParam = false;
if (fastPathRouterQuery) if (fastPathRouterQuery)
{ {
result = PlanFastPathDistributedStmt(&ctx, distributionKeyValue); result = PlanFastPathDistributedStmt(&planContext, distributionKeyValue);
}
else if (needsDistributedPlanning)
{
result = PlanDistributedStmt(&ctx, rangeTableList, rteIdCounter);
}
else if ((delegatedPlan = TryToDelegateFunctionCall(ctx.parse,
&hasExternParam)))
{
result = PlanDelegatedFunctionCall(&ctx, delegatedPlan);
} }
else else
{ {
result = PlanNonDistributedStmt(&ctx, hasExternParam); /*
* Call into standard_planner because the Citus planner relies on both the
* restriction information per table and parse tree transformations made by
* postgres' planner.
*/
planContext.plan = standard_planner(planContext.parse,
planContext.cursorOptions,
planContext.boundParams);
if (needsDistributedPlanning)
{
result = PlanDistributedStmt(&planContext, rangeTableList, rteIdCounter);
}
else if ((result = TryToDelegateFunctionCall(&planContext)) == NULL)
{
result = planContext.plan;
}
} }
} }
PG_CATCH(); PG_CATCH();
@ -567,17 +540,18 @@ IsModifyDistributedPlan(DistributedPlan *distributedPlan)
* the FastPathPlanner. * the FastPathPlanner.
*/ */
static PlannedStmt * static PlannedStmt *
PlanFastPathDistributedStmt(DistributedPlanningContext *ctx, Const *distributionKeyValue) PlanFastPathDistributedStmt(DistributedPlanningContext *planContext,
Const *distributionKeyValue)
{ {
ctx->plannerRestrictionContext->fastPathRestrictionContext->fastPathRouterQuery = planContext->plannerRestrictionContext->fastPathRestrictionContext->
true; fastPathRouterQuery = true;
ctx->plannerRestrictionContext->fastPathRestrictionContext->distributionKeyValue = planContext->plannerRestrictionContext->fastPathRestrictionContext->
distributionKeyValue; distributionKeyValue = distributionKeyValue;
PlannedStmt *fastPathPlan = FastPathPlanner(ctx->originalQuery, ctx->parse, planContext->plan = FastPathPlanner(planContext->originalQuery, planContext->parse,
ctx->boundParams); planContext->boundParams);
return CreateDistributedPlannedStmt(fastPathPlan, ctx); return CreateDistributedPlannedStmt(planContext);
} }
@ -586,24 +560,16 @@ PlanFastPathDistributedStmt(DistributedPlanningContext *ctx, Const *distribution
* planner. * planner.
*/ */
static PlannedStmt * static PlannedStmt *
PlanDistributedStmt(DistributedPlanningContext *ctx, PlanDistributedStmt(DistributedPlanningContext *planContext,
List *rangeTableList, List *rangeTableList,
int rteIdCounter) int rteIdCounter)
{ {
/*
* Call into standard_planner because the Citus planner relies on both the
* restriction information per table and parse tree transformations made by
* postgres' planner.
*/
PlannedStmt *standardPlan = standard_planner(ctx->parse, ctx->cursorOptions,
ctx->boundParams);
/* may've inlined new relation rtes */ /* may've inlined new relation rtes */
rangeTableList = ExtractRangeTableEntryList(ctx->parse); rangeTableList = ExtractRangeTableEntryList(planContext->parse);
rteIdCounter = AssignRTEIdentities(rangeTableList, rteIdCounter); rteIdCounter = AssignRTEIdentities(rangeTableList, rteIdCounter);
PlannedStmt *result = CreateDistributedPlannedStmt(standardPlan, ctx); PlannedStmt *result = CreateDistributedPlannedStmt(planContext);
bool setPartitionedTablesInherited = true; bool setPartitionedTablesInherited = true;
AdjustPartitioningForDistributedPlanning(rangeTableList, AdjustPartitioningForDistributedPlanning(rangeTableList,
@ -614,41 +580,17 @@ PlanDistributedStmt(DistributedPlanningContext *ctx,
/* /*
* PlanDelegatedFunctionCall creates a plan by delagating the function call to * DissuadePlannerFromUsingPlan try dissuade planner when planning a plan that
* the worker. * potentially failed due to unresolved prepared statement parameters.
*/ */
static PlannedStmt * void
PlanDelegatedFunctionCall(DistributedPlanningContext *ctx, DistributedPlan *delegatedPlan) DissuadePlannerFromUsingPlan(PlannedStmt *plan)
{ {
PlannedStmt *result = standard_planner(ctx->parse, ctx->cursorOptions, /*
ctx->boundParams); * Arbitrarily high cost, but low enough that it can be added up
* without overflowing by choose_custom_plan().
result = FinalizePlan(result, delegatedPlan); */
plan->planTree->total_cost = FLT_MAX / 100000000;
return result;
}
/*
* PlanNonDistributedStmt creates a normal (non-distributed) planned statement
* using the PG planner.
*/
static PlannedStmt *
PlanNonDistributedStmt(DistributedPlanningContext *ctx, bool hasExternParam)
{
PlannedStmt *result = standard_planner(ctx->parse, ctx->cursorOptions,
ctx->boundParams);
if (hasExternParam)
{
/*
* As in CreateDistributedPlannedStmt, try dissuade planner when planning
* potentially failed due to unresolved prepared statement parameters.
*/
result->planTree->total_cost = FLT_MAX / 100000000;
}
return result;
} }
@ -657,25 +599,27 @@ PlanNonDistributedStmt(DistributedPlanningContext *ctx, bool hasExternParam)
* query into a distributed plan that is encapsulated by a PlannedStmt. * query into a distributed plan that is encapsulated by a PlannedStmt.
*/ */
static PlannedStmt * static PlannedStmt *
CreateDistributedPlannedStmt(PlannedStmt *localPlan, DistributedPlanningContext *ctx) CreateDistributedPlannedStmt(DistributedPlanningContext *planContext)
{ {
uint64 planId = NextPlanId++; uint64 planId = NextPlanId++;
bool hasUnresolvedParams = false; bool hasUnresolvedParams = false;
JoinRestrictionContext *joinRestrictionContext = JoinRestrictionContext *joinRestrictionContext =
ctx->plannerRestrictionContext->joinRestrictionContext; planContext->plannerRestrictionContext->joinRestrictionContext;
if (HasUnresolvedExternParamsWalker((Node *) ctx->originalQuery, ctx->boundParams)) if (HasUnresolvedExternParamsWalker((Node *) planContext->originalQuery,
planContext->boundParams))
{ {
hasUnresolvedParams = true; hasUnresolvedParams = true;
} }
ctx->plannerRestrictionContext->joinRestrictionContext = planContext->plannerRestrictionContext->joinRestrictionContext =
RemoveDuplicateJoinRestrictions(joinRestrictionContext); RemoveDuplicateJoinRestrictions(joinRestrictionContext);
DistributedPlan *distributedPlan = DistributedPlan *distributedPlan =
CreateDistributedPlan(planId, ctx->originalQuery, ctx->parse, CreateDistributedPlan(planId, planContext->originalQuery, planContext->parse,
ctx->boundParams, planContext->boundParams,
hasUnresolvedParams, ctx->plannerRestrictionContext); hasUnresolvedParams,
planContext->plannerRestrictionContext);
/* /*
* If no plan was generated, prepare a generic error to be emitted. * If no plan was generated, prepare a generic error to be emitted.
@ -720,7 +664,7 @@ CreateDistributedPlannedStmt(PlannedStmt *localPlan, DistributedPlanningContext
distributedPlan->planId = planId; distributedPlan->planId = planId;
/* create final plan by combining local plan with distributed plan */ /* create final plan by combining local plan with distributed plan */
PlannedStmt *resultPlan = FinalizePlan(localPlan, distributedPlan); PlannedStmt *resultPlan = FinalizePlan(planContext->plan, distributedPlan);
/* /*
* As explained above, force planning costs to be unrealistically high if * As explained above, force planning costs to be unrealistically high if
@ -728,14 +672,11 @@ CreateDistributedPlannedStmt(PlannedStmt *localPlan, DistributedPlanningContext
* if it is planned as a multi shard modify query. * if it is planned as a multi shard modify query.
*/ */
if ((distributedPlan->planningError || if ((distributedPlan->planningError ||
(IsUpdateOrDelete(ctx->originalQuery) && IsMultiTaskPlan(distributedPlan))) && (IsUpdateOrDelete(planContext->originalQuery) && IsMultiTaskPlan(
distributedPlan))) &&
hasUnresolvedParams) hasUnresolvedParams)
{ {
/* DissuadePlannerFromUsingPlan(resultPlan);
* Arbitrarily high cost, but low enough that it can be added up
* without overflowing by choose_custom_plan().
*/
resultPlan->planTree->total_cost = FLT_MAX / 100000000;
} }
return resultPlan; return resultPlan;
@ -1173,7 +1114,7 @@ GetDistributedPlan(CustomScan *customScan)
* FinalizePlan combines local plan with distributed plan and creates a plan * FinalizePlan combines local plan with distributed plan and creates a plan
* which can be run by the PostgreSQL executor. * which can be run by the PostgreSQL executor.
*/ */
static PlannedStmt * PlannedStmt *
FinalizePlan(PlannedStmt *localPlan, DistributedPlan *distributedPlan) FinalizePlan(PlannedStmt *localPlan, DistributedPlan *distributedPlan)
{ {
PlannedStmt *finalPlan = NULL; PlannedStmt *finalPlan = NULL;

View File

@ -95,8 +95,8 @@ contain_param_walker(Node *node, void *context)
* forms involving multiple function calls, FROM clauses, WHERE clauses, * forms involving multiple function calls, FROM clauses, WHERE clauses,
* ... Those complex forms are handled in the coordinator. * ... Those complex forms are handled in the coordinator.
*/ */
DistributedPlan * PlannedStmt *
TryToDelegateFunctionCall(Query *query, bool *hasExternParam) TryToDelegateFunctionCall(DistributedPlanningContext *planContext)
{ {
List *targetList = NIL; List *targetList = NIL;
TargetEntry *targetEntry = NULL; TargetEntry *targetEntry = NULL;
@ -114,12 +114,9 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
StringInfo queryString = NULL; StringInfo queryString = NULL;
Task *task = NULL; Task *task = NULL;
Job *job = NULL; Job *job = NULL;
DistributedPlan *distributedPlan = NULL; DistributedPlan *distributedPlan = CitusMakeNode(DistributedPlan);
struct ParamWalkerContext walkerParamContext = { 0 }; struct ParamWalkerContext walkerParamContext = { 0 };
/* set hasExternParam now in case of early exit */
*hasExternParam = false;
if (!CitusHasBeenLoaded() || !CheckCitusVersion(DEBUG4)) if (!CitusHasBeenLoaded() || !CheckCitusVersion(DEBUG4))
{ {
/* Citus is not ready to determine whether function is distributed */ /* Citus is not ready to determine whether function is distributed */
@ -133,19 +130,19 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
return NULL; return NULL;
} }
if (query == NULL) if (planContext->parse == NULL)
{ {
/* no query (mostly here to be defensive) */ /* no query (mostly here to be defensive) */
return NULL; return NULL;
} }
if (query->commandType != CMD_SELECT) if (planContext->parse->commandType != CMD_SELECT)
{ {
/* not a SELECT */ /* not a SELECT */
return NULL; return NULL;
} }
FromExpr *joinTree = query->jointree; FromExpr *joinTree = planContext->parse->jointree;
if (joinTree == NULL) if (joinTree == NULL)
{ {
/* no join tree (mostly here to be defensive) */ /* no join tree (mostly here to be defensive) */
@ -174,7 +171,8 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
if (IsA(reference, RangeTblRef)) if (IsA(reference, RangeTblRef))
{ {
RangeTblEntry *rtentry = rt_fetch(reference->rtindex, query->rtable); RangeTblEntry *rtentry = rt_fetch(reference->rtindex,
planContext->parse->rtable);
if (rtentry->rtekind != RTE_RESULT) if (rtentry->rtekind != RTE_RESULT)
{ {
/* e.g. SELECT f() FROM rel */ /* e.g. SELECT f() FROM rel */
@ -203,8 +201,8 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
#endif #endif
} }
targetList = query->targetList; targetList = planContext->parse->targetList;
if (list_length(query->targetList) != 1) if (list_length(planContext->parse->targetList) != 1)
{ {
/* multiple target list items */ /* multiple target list items */
return NULL; return NULL;
@ -288,7 +286,7 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
if (partitionParam->paramkind == PARAM_EXTERN) if (partitionParam->paramkind == PARAM_EXTERN)
{ {
/* Don't log a message, we should end up here again without a parameter */ /* Don't log a message, we should end up here again without a parameter */
*hasExternParam = true; DissuadePlannerFromUsingPlan(planContext->plan);
return NULL; return NULL;
} }
} }
@ -354,7 +352,7 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
if (walkerParamContext.paramKind == PARAM_EXTERN) if (walkerParamContext.paramKind == PARAM_EXTERN)
{ {
/* Don't log a message, we should end up here again without a parameter */ /* Don't log a message, we should end up here again without a parameter */
*hasExternParam = true; DissuadePlannerFromUsingPlan(planContext->plan);
} }
else else
{ {
@ -367,7 +365,7 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
ereport(DEBUG1, (errmsg("pushing down the function call"))); ereport(DEBUG1, (errmsg("pushing down the function call")));
queryString = makeStringInfo(); queryString = makeStringInfo();
pg_get_query_def(query, queryString); pg_get_query_def(planContext->parse, queryString);
task = CitusMakeNode(Task); task = CitusMakeNode(Task);
task->taskType = SELECT_TASK; task->taskType = SELECT_TASK;
@ -378,7 +376,7 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
job = CitusMakeNode(Job); job = CitusMakeNode(Job);
job->jobId = UniqueJobId(); job->jobId = UniqueJobId();
job->jobQuery = query; job->jobQuery = planContext->parse;
job->taskList = list_make1(task); job->taskList = list_make1(task);
distributedPlan = CitusMakeNode(DistributedPlan); distributedPlan = CitusMakeNode(DistributedPlan);
@ -390,5 +388,5 @@ TryToDelegateFunctionCall(Query *query, bool *hasExternParam)
/* worker will take care of any necessary locking, treat query as read-only */ /* worker will take care of any necessary locking, treat query as read-only */
distributedPlan->modLevel = ROW_MODIFY_READONLY; distributedPlan->modLevel = ROW_MODIFY_READONLY;
return distributedPlan; return FinalizePlan(planContext->plan, distributedPlan);
} }

View File

@ -129,6 +129,33 @@ typedef struct RelationRowLock
} RelationRowLock; } RelationRowLock;
typedef struct DistributedPlanningContext
{
/* The parsed query that is given to the planner. It is a slightly modified
* to work with the standard_planner */
Query *parse;
/* A copy of the original parsed query that is given to the planner. This
* doesn't contain most of the changes that are made to parse. There's one
* that change that is made for non fast path router queries though, which
* is the assigning of RTE identities using AssignRTEIdentities. This is
* NULL for non distributed plans, since those don't need it. */
Query *originalQuery;
/* the cursor options given to the planner */
int cursorOptions;
/* the ParamListInfo that is given to the planner */
ParamListInfo boundParams;
/* Plan created either by standard_planner or by FastPathPlanner */
PlannedStmt *plan;
/* Our custom restriction context */
PlannerRestrictionContext *plannerRestrictionContext;
} DistributedPlanningContext;
extern PlannedStmt * distributed_planner(Query *parse, int cursorOptions, extern PlannedStmt * distributed_planner(Query *parse, int cursorOptions,
ParamListInfo boundParams); ParamListInfo boundParams);
extern List * ExtractRangeTableEntryList(Query *query); extern List * ExtractRangeTableEntryList(Query *query);
@ -150,5 +177,8 @@ extern bool IsMultiTaskPlan(struct DistributedPlan *distributedPlan);
extern RangeTblEntry * RemoteScanRangeTableEntry(List *columnNameList); extern RangeTblEntry * RemoteScanRangeTableEntry(List *columnNameList);
extern int GetRTEIdentity(RangeTblEntry *rte); extern int GetRTEIdentity(RangeTblEntry *rte);
extern int32 BlessRecordExpression(Expr *expr); extern int32 BlessRecordExpression(Expr *expr);
extern void DissuadePlannerFromUsingPlan(PlannedStmt *plan);
extern PlannedStmt * FinalizePlan(PlannedStmt *localPlan,
struct DistributedPlan *distributedPlan);
#endif /* DISTRIBUTED_PLANNER_H */ #endif /* DISTRIBUTED_PLANNER_H */

View File

@ -10,10 +10,12 @@
#define FUNCTION_CALL_DELEGATION_H #define FUNCTION_CALL_DELEGATION_H
#include "postgres.h" #include "postgres.h"
#include "distributed/distributed_planner.h"
#include "distributed/multi_physical_planner.h" #include "distributed/multi_physical_planner.h"
DistributedPlan * TryToDelegateFunctionCall(Query *query, bool *hasParam); PlannedStmt * TryToDelegateFunctionCall(DistributedPlanningContext *planContext);
#endif /* FUNCTION_CALL_DELEGATION_H */ #endif /* FUNCTION_CALL_DELEGATION_H */