coprocessor.go

// Copyright 2016 PingCAP, Inc.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// See the License for the specific language governing permissions and
// limitations under the License.

package tikv

import (
	"bytes"
	"fmt"
	"sync"
	"time"

	log "github.com/Sirupsen/logrus"
	"github.com/juju/errors"
	"github.com/pingcap/kvproto/pkg/coprocessor"
	"github.com/pingcap/kvproto/pkg/kvrpcpb"
	"github.com/pingcap/tidb/kv"
	"github.com/pingcap/tidb/store/tikv/tikvrpc"
	"github.com/pingcap/tidb/util/goroutine_pool"
	"github.com/pingcap/tipb/go-tipb"
	goctx "golang.org/x/net/context"
)

var copIteratorGP = gp.New(time.Minute)

// CopClient is coprocessor client.
type CopClient struct {
	store *tikvStore
}

// IsRequestTypeSupported checks whether reqType is supported.
func (c *CopClient) IsRequestTypeSupported(reqType, subType int64) bool {
	switch reqType {
	case kv.ReqTypeSelect, kv.ReqTypeIndex:
		switch subType {
		case kv.ReqSubTypeGroupBy, kv.ReqSubTypeBasic, kv.ReqSubTypeTopN:
			return true
		default:
			return c.supportExpr(tipb.ExprType(subType))
		}
	case kv.ReqTypeDAG:
		return c.supportExpr(tipb.ExprType(subType))
	case kv.ReqTypeAnalyze:
		return true
	}
	return false
}

func (c *CopClient) supportExpr(exprType tipb.ExprType) bool {
	switch exprType {
	case tipb.ExprType_Null, tipb.ExprType_Int64, tipb.ExprType_Uint64, tipb.ExprType_String, tipb.ExprType_Bytes,
		tipb.ExprType_MysqlDuration, tipb.ExprType_MysqlTime, tipb.ExprType_MysqlDecimal,
		tipb.ExprType_ColumnRef:
		return true
	// logic operators.
	case tipb.ExprType_And, tipb.ExprType_Or, tipb.ExprType_Not:
		return true
	// compare operators.
	case tipb.ExprType_LT, tipb.ExprType_LE, tipb.ExprType_EQ, tipb.ExprType_NE,
		tipb.ExprType_GE, tipb.ExprType_GT, tipb.ExprType_NullEQ,
		tipb.ExprType_In, tipb.ExprType_ValueList, tipb.ExprType_IsNull,
		tipb.ExprType_Like:
		return true
	// arithmetic operators.
	case tipb.ExprType_Plus, tipb.ExprType_Div, tipb.ExprType_Minus, tipb.ExprType_Mul:
		return true
	// control functions
	case tipb.ExprType_Case, tipb.ExprType_If, tipb.ExprType_IfNull, tipb.ExprType_Coalesce:
		return true
	// aggregate functions.
	case tipb.ExprType_Count, tipb.ExprType_First, tipb.ExprType_Max, tipb.ExprType_Min, tipb.ExprType_Sum, tipb.ExprType_Avg:
		return true
	// json functions.
	case tipb.ExprType_JsonType, tipb.ExprType_JsonExtract, tipb.ExprType_JsonUnquote,
		tipb.ExprType_JsonObject, tipb.ExprType_JsonArray, tipb.ExprType_JsonMerge,
		tipb.ExprType_JsonSet, tipb.ExprType_JsonInsert, tipb.ExprType_JsonReplace, tipb.ExprType_JsonRemove:
		return true
	case kv.ReqSubTypeDesc:
		return true
	case kv.ReqSubTypeSignature:
		return true
	default:
		return false
	}
}

// Send builds the request and gets the coprocessor iterator response.
func (c *CopClient) Send(ctx goctx.Context, req *kv.Request) kv.Response {
	coprocessorCounter.WithLabelValues("send").Inc()

	bo := NewBackoffer(copBuildTaskMaxBackoff, ctx)
	tasks, err := buildCopTasks(bo, c.store.regionCache, &copRanges{mid: req.KeyRanges}, req.Desc)
	if err != nil {
		return copErrorResponse{err}
	}
	it := &copIterator{
		store:       c.store,
		req:         req,
		concurrency: req.Concurrency,
		finished:    make(chan struct{}),
	}
	it.tasks = tasks
	if it.concurrency > len(tasks) {
		it.concurrency = len(tasks)
	}
	if it.concurrency < 1 {
		// Make sure that there is at least one worker.
		it.concurrency = 1
	}
	if !it.req.KeepOrder {
		it.respChan = make(chan copResponse, it.concurrency)
	}
	it.run(ctx)
	return it
}

// copTask contains a related Region and KeyRange for a kv.Request.
type copTask struct {
	region RegionVerID
	ranges *copRanges

	respChan  chan copResponse
	storeAddr string
}

func (r *copTask) String() string {
	return fmt.Sprintf("region(%d %d %d) ranges(%d) store(%s)",
		r.region.id, r.region.confVer, r.region.ver, r.ranges.len(), r.storeAddr)
}

// copRanges is like []kv.KeyRange, but may has extra elements at head/tail.
// It's for avoiding alloc big slice during build copTask.
type copRanges struct {
	first *kv.KeyRange
	mid   []kv.KeyRange
	last  *kv.KeyRange
}

func (r *copRanges) String() string {
	var s string
	r.do(func(ran *kv.KeyRange) {
		s += fmt.Sprintf("[%q, %q]", ran.StartKey, ran.EndKey)
	})
	return s
}

func (r *copRanges) len() int {
	var l int
	if r.first != nil {
		l++
	}
	l += len(r.mid)
	if r.last != nil {
		l++
	}
	return l
}

func (r *copRanges) at(i int) kv.KeyRange {
	if r.first != nil {
		if i == 0 {
			return *r.first
		}
		i--
	}
	if i < len(r.mid) {
		return r.mid[i]
	}
	return *r.last
}

func (r *copRanges) slice(from, to int) *copRanges {
	var ran copRanges
	if r.first != nil {
		if from == 0 && to > 0 {
			ran.first = r.first
		}
		if from > 0 {
			from--
		}
		if to > 0 {
			to--
		}
	}
	if to <= len(r.mid) {
		ran.mid = r.mid[from:to]
	} else {
		if from <= len(r.mid) {
			ran.mid = r.mid[from:]
		}
		if from < to {
			ran.last = r.last
		}
	}
	return &ran
}

func (r *copRanges) do(f func(ran *kv.KeyRange)) {
	if r.first != nil {
		f(r.first)
	}
	for _, ran := range r.mid {
		f(&ran)
	}
	if r.last != nil {
		f(r.last)
	}
}

func (r *copRanges) toPBRanges() []*coprocessor.KeyRange {
	ranges := make([]*coprocessor.KeyRange, 0, r.len())
	r.do(func(ran *kv.KeyRange) {
		ranges = append(ranges, &coprocessor.KeyRange{
			Start: ran.StartKey,
			End:   ran.EndKey,
		})
	})
	return ranges
}

func buildCopTasks(bo *Backoffer, cache *RegionCache, ranges *copRanges, desc bool) ([]*copTask, error) {
	coprocessorCounter.WithLabelValues("build_task").Inc()

	start := time.Now()
	rangesLen := ranges.len()

	var tasks []*copTask
	appendTask := func(region RegionVerID, ranges *copRanges) {
		tasks = append(tasks, &copTask{
			region:   region,
			ranges:   ranges,
			respChan: make(chan copResponse, 1),
		})
	}

	for ranges.len() > 0 {
		loc, err := cache.LocateKey(bo, ranges.at(0).StartKey)
		if err != nil {
			return nil, errors.Trace(err)
		}

		// Iterate to the first range that is not complete in the region.
		var i int
		for ; i < ranges.len(); i++ {
			r := ranges.at(i)
			if !(loc.Contains(r.EndKey) || bytes.Equal(loc.EndKey, r.EndKey)) {
				break
			}
		}
		// All rest ranges belong to the same region.
		if i == ranges.len() {
			appendTask(loc.Region, ranges)
			break
		}

		r := ranges.at(i)
		if loc.Contains(r.StartKey) {
			// Part of r is not in the region. We need to split it.
			taskRanges := ranges.slice(0, i)
			taskRanges.last = &kv.KeyRange{
				StartKey: r.StartKey,
				EndKey:   loc.EndKey,
			}
			appendTask(loc.Region, taskRanges)

			ranges = ranges.slice(i+1, ranges.len())
			ranges.first = &kv.KeyRange{
				StartKey: loc.EndKey,
				EndKey:   r.EndKey,
			}
		} else {
			// rs[i] is not in the region.
			appendTask(loc.Region, ranges.slice(0, i))
			ranges = ranges.slice(i, ranges.len())
		}
	}

	if desc {
		reverseTasks(tasks)
	}
	if elapsed := time.Since(start); elapsed > time.Millisecond*500 {
		log.Warnf("buildCopTasks takes too much time (%v), range len %v, task len %v", elapsed, rangesLen, len(tasks))
	}
	txnRegionsNumHistogram.WithLabelValues("coprocessor").Observe(float64(len(tasks)))
	return tasks, nil
}

func reverseTasks(tasks []*copTask) {
	for i := 0; i < len(tasks)/2; i++ {
		j := len(tasks) - i - 1
		tasks[i], tasks[j] = tasks[j], tasks[i]
	}
}

type copIterator struct {
	store       *tikvStore
	req         *kv.Request
	concurrency int
	finished    chan struct{}

	// If keepOrder, results are stored in copTask.respChan, read them out one by one.
	tasks []*copTask
	curr  int

	// Otherwise, results are stored in respChan.
	respChan chan copResponse
	wg       sync.WaitGroup
}

type copResponse struct {
	*coprocessor.Response
	err error
}

const minLogCopTaskTime = 300 * time.Millisecond

// work is a worker function that get a copTask from channel, handle it and
// send the result back.
func (it *copIterator) work(ctx goctx.Context, taskCh <-chan *copTask) {
	defer it.wg.Done()
	for task := range taskCh {
		bo := NewBackoffer(copNextMaxBackoff, ctx)
		startTime := time.Now()
		resps := it.handleTask(bo, task)
		costTime := time.Since(startTime)
		if costTime > minLogCopTaskTime {
			log.Infof("[TIME_COP_TASK] %s%s %s", costTime, bo, task)
		}
		coprocessorHistogram.Observe(costTime.Seconds())
		if bo.totalSleep > 0 {
			backoffHistogram.Observe(float64(bo.totalSleep) / 1000)
		}
		var ch chan copResponse
		if !it.req.KeepOrder {
			ch = it.respChan
		} else {
			ch = task.respChan
		}

		for _, resp := range resps {
			select {
			case ch <- resp:
			case <-ctx.Done():
				return
			case <-it.finished:
				return
			}
		}
		if it.req.KeepOrder {
			close(ch)
		}
	}
}

func (it *copIterator) run(ctx goctx.Context) {
	taskCh := make(chan *copTask, 1)
	it.wg.Add(it.concurrency)
	// Start it.concurrency number of workers to handle cop requests.
	for i := 0; i < it.concurrency; i++ {
		copIteratorGP.Go(func() {
			childCtx, cancel := goctx.WithCancel(ctx)
			defer cancel()
			it.work(childCtx, taskCh)
		})
	}

	copIteratorGP.Go(func() {
		// Send tasks to feed the worker goroutines.
		childCtx, cancel := goctx.WithCancel(ctx)
		defer cancel()
		for _, t := range it.tasks {
			finished, canceled := it.sendToTaskCh(childCtx, t, taskCh)
			if finished || canceled {
				break
			}
		}
		close(taskCh)

		// Wait for worker goroutines to exit.
		it.wg.Wait()
		if !it.req.KeepOrder {
			close(it.respChan)
		}
	})
}

func (it *copIterator) sendToTaskCh(ctx goctx.Context, t *copTask, taskCh chan<- *copTask) (finished bool, canceled bool) {
	select {
	case taskCh <- t:
	case <-it.finished:
		finished = true
	case <-ctx.Done():
		canceled = true
	}
	return
}

// Next returns next coprocessor result.
func (it *copIterator) Next() ([]byte, error) {
	coprocessorCounter.WithLabelValues("next").Inc()

	var (
		resp copResponse
		ok   bool
	)
	// If data order matters, response should be returned in the same order as copTask slice.
	// Otherwise all responses are returned from a single channel.
	if !it.req.KeepOrder {
		// Get next fetched resp from chan
		resp, ok = <-it.respChan
		if !ok {
			return nil, nil
		}
	} else {
		for {
			if it.curr >= len(it.tasks) {
				// Resp will be nil if iterator is finished.
				return nil, nil
			}
			task := it.tasks[it.curr]
			resp, ok = <-task.respChan
			if ok {
				break
			}
			// Switch to next task.
			it.tasks[it.curr] = nil
			it.curr++
		}
	}

	if resp.err != nil {
		return nil, errors.Trace(resp.err)
	}
	if resp.Data == nil {
		return []byte{}, nil
	}

	err := it.store.CheckVisibility(it.req.StartTs)
	if err != nil {
		return nil, errors.Trace(err)
	}

	return resp.Data, nil
}

// handleTask handles single copTask.
func (it *copIterator) handleTask(bo *Backoffer, task *copTask) []copResponse {
	coprocessorCounter.WithLabelValues("handle_task").Inc()
	sender := NewRegionRequestSender(it.store.regionCache, it.store.client)
	for {
		select {
		case <-it.finished:
			return nil
		default:
		}

		req := &tikvrpc.Request{
			Type: tikvrpc.CmdCop,
			Cop: &coprocessor.Request{
				Tp:     it.req.Tp,
				Data:   it.req.Data,
				Ranges: task.ranges.toPBRanges(),
			},
			Context: kvrpcpb.Context{
				IsolationLevel: pbIsolationLevel(it.req.IsolationLevel),
				Priority:       kvPriorityToCommandPri(it.req.Priority),
				NotFillCache:   it.req.NotFillCache,
			},
		}
		resp, err := sender.SendReq(bo, req, task.region, readTimeoutMedium)
		if err != nil {
			return []copResponse{{err: errors.Trace(err)}}
		}
		if regionErr := resp.Cop.GetRegionError(); regionErr != nil {
			err = bo.Backoff(boRegionMiss, errors.New(regionErr.String()))
			if err != nil {
				return []copResponse{{err: errors.Trace(err)}}
			}
			return it.handleRegionErrorTask(bo, task)
		}
		if e := resp.Cop.GetLocked(); e != nil {
			log.Debugf("coprocessor encounters lock: %v", e)
			ok, err1 := it.store.lockResolver.ResolveLocks(bo, []*Lock{newLock(e)})
			if err1 != nil {
				return []copResponse{{err: errors.Trace(err1)}}
			}
			if !ok {
				err = bo.Backoff(boTxnLockFast, errors.New(e.String()))
				if err != nil {
					return []copResponse{{err: errors.Trace(err)}}
				}
			}
			continue
		}
		if e := resp.Cop.GetOtherError(); e != "" {
			err = errors.Errorf("other error: %s", e)
			log.Warnf("coprocessor err: %v", err)
			return []copResponse{{err: errors.Trace(err)}}
		}
		task.storeAddr = sender.storeAddr
		return []copResponse{{Response: resp.Cop}}
	}
}

// handleRegionErrorTask handles current task. It may be split into multiple tasks (in region split scenario).
func (it *copIterator) handleRegionErrorTask(bo *Backoffer, task *copTask) []copResponse {
	coprocessorCounter.WithLabelValues("rebuild_task").Inc()

	newTasks, err := buildCopTasks(bo, it.store.regionCache, task.ranges, it.req.Desc)
	if err != nil {
		return []copResponse{{err: errors.Trace(err)}}
	}
	if newTasks == nil {
		// TODO: check this, this should never happen.
		return nil
	}

	var ret []copResponse
	for _, t := range newTasks {
		resp := it.handleTask(bo, t)
		ret = append(ret, resp...)
	}
	return ret
}

func (it *copIterator) Close() error {
	close(it.finished)
	it.wg.Wait()
	return nil
}

// copErrorResponse returns error when calling Next()
type copErrorResponse struct{ error }

func (it copErrorResponse) Next() ([]byte, error) {
	return nil, it.error
}

func (it copErrorResponse) Close() error {
	return nil
}