[dbnode] Add configurability for regexp DFA and FSA limits (#2926)

go.mod

@@ -63,7 +63,7 @@ require (
 	github.com/m3db/stackmurmur3/v2 v2.0.2
 	github.com/m3db/tools v0.0.0-20181008195521-c6ded3f34878
 	github.com/m3dbx/pilosa v1.4.1
-	github.com/m3dbx/vellum v0.0.0-20200826162549-f94c029903de
+	github.com/m3dbx/vellum v0.0.0-20201119082309-5b47f7a70f69
 	github.com/mauricelam/genny v0.0.0-20180903214747-eb2c5232c885
 	github.com/mjibson/esc v0.1.0
 	github.com/niemeyer/pretty v0.0.0-20200227124842-a10e7caefd8e // indirect

go.sum

@@ -556,8 +556,8 @@ github.com/m3db/tools v0.0.0-20181008195521-c6ded3f34878 h1:kww0LtVVfGrXR7Ofpbi/
 github.com/m3db/tools v0.0.0-20181008195521-c6ded3f34878/go.mod h1:TxroQUZzb1wzOsq+4+TfVtT7z89YTz3v2UJAYfLNfLE=
 github.com/m3dbx/pilosa v1.4.1 h1:/Cpp1XAHSd6orpjceXGiKpCoDdYBP5BD/6NoqGG9eVg=
 github.com/m3dbx/pilosa v1.4.1/go.mod h1:Jt0+w9O08sa7qWDeRC58VBjb4OeOTDMOhfvVmyeVCO8=
-github.com/m3dbx/vellum v0.0.0-20200826162549-f94c029903de h1:C4DpCfTNzJf5RhJqxOtfWAnD2d6ls7KDnK1boBGUnVg=
-github.com/m3dbx/vellum v0.0.0-20200826162549-f94c029903de/go.mod h1:DOTAUfV4bzK6Nrb0dboT/oCG0DnQuX+/n0jfZPh6xxI=
+github.com/m3dbx/vellum v0.0.0-20201119082309-5b47f7a70f69 h1:dANuca0xuYlZR7qWdPIIAZKG0YHvsbLTzdenj53yQvc=
+github.com/m3dbx/vellum v0.0.0-20201119082309-5b47f7a70f69/go.mod h1:DOTAUfV4bzK6Nrb0dboT/oCG0DnQuX+/n0jfZPh6xxI=
 github.com/magiconair/properties v1.8.0/go.mod h1:PppfXfuXeibc/6YijjN8zIbojt8czPbwD3XqdrwzmxQ=
 github.com/magiconair/properties v1.8.1 h1:ZC2Vc7/ZFkGmsVC9KvOjumD+G5lXy2RtTKyzRKO2BQ4=
 github.com/magiconair/properties v1.8.1/go.mod h1:PppfXfuXeibc/6YijjN8zIbojt8czPbwD3XqdrwzmxQ=

src/cmd/services/m3dbnode/config/config.go

@@ -40,6 +40,7 @@ import (
 	xlog "github.com/m3db/m3/src/x/log"
 	"github.com/m3db/m3/src/x/opentracing"
 
+	"github.com/m3dbx/vellum/regexp"
 	"go.etcd.io/etcd/embed"
 	"go.etcd.io/etcd/pkg/transport"
 	"go.etcd.io/etcd/pkg/types"
@@ -379,6 +380,14 @@ type IndexConfiguration struct {
 	// as they are very CPU-intensive (regex and FST matching).
 	MaxQueryIDsConcurrency int `yaml:"maxQueryIDsConcurrency" validate:"min=0"`
 
+	// RegexpDFALimit is the limit on the max number of states used by a
+	// regexp deterministic finite automaton. Default is 10,000 states.
+	RegexpDFALimit *int `yaml:"regexpDFALimit"`
+
+	// RegexpFSALimit is the limit on the max number of bytes used by the
+	// finite state automaton. Default is 10mb (10 million as int).
+	RegexpFSALimit *uint `yaml:"regexpFSALimit"`
+
 	// ForwardIndexProbability determines the likelihood that an incoming write is
 	// written to the next block, when arriving close to the block boundary.
 	//
@@ -396,6 +405,24 @@ type IndexConfiguration struct {
 	ForwardIndexThreshold float64 `yaml:"forwardIndexThreshold" validate:"min=0.0,max=1.0"`
 }
 
+// RegexpDFALimitOrDefault returns the deterministic finite automaton states
+// limit or default.
+func (c IndexConfiguration) RegexpDFALimitOrDefault() int {
+	if c.RegexpDFALimit == nil {
+		return regexp.StateLimit()
+	}
+	return *c.RegexpDFALimit
+}
+
+// RegexpFSALimitOrDefault returns the finite state automaton size
+// limit or default.
+func (c IndexConfiguration) RegexpFSALimitOrDefault() uint {
+	if c.RegexpFSALimit == nil {
+		return regexp.DefaultLimit()
+	}
+	return *c.RegexpFSALimit
+}
+
 // TransformConfiguration contains configuration options that can transform
 // incoming writes.
 type TransformConfiguration struct {

src/cmd/services/m3dbnode/config/config_test.go

@@ -337,6 +337,8 @@ func TestConfiguration(t *testing.T) {
 	expected := `db:
   index:
     maxQueryIDsConcurrency: 0
+    regexpDFALimit: null
+    regexpFSALimit: null
     forwardIndexProbability: 0
     forwardIndexThreshold: 0
   transforms:

src/dbnode/server/server.go

@@ -91,6 +91,9 @@ import (
 	xsync "github.com/m3db/m3/src/x/sync"
 
 	apachethrift "github.com/apache/thrift/lib/go/thrift"
+	"github.com/m3dbx/vellum/levenshtein"
+	"github.com/m3dbx/vellum/levenshtein2"
+	"github.com/m3dbx/vellum/regexp"
 	opentracing "github.com/opentracing/opentracing-go"
 	"github.com/uber-go/tally"
 	"github.com/uber/tchannel-go"
@@ -371,6 +374,16 @@ func Run(runOpts RunOptions) {
 		logger.Warn("max index query IDs concurrency was not set, falling back to default value")
 	}
 
+	// Set global index options.
+	if n := cfg.Index.RegexpDFALimitOrDefault(); n > 0 {
+		regexp.SetStateLimit(n)
+		levenshtein.SetStateLimit(n)
+		levenshtein2.SetStateLimit(n)
+	}
+	if n := cfg.Index.RegexpFSALimitOrDefault(); n > 0 {
+		regexp.SetDefaultLimit(n)
+	}
+
 	buildReporter := instrument.NewBuildReporter(iOpts)
 	if err := buildReporter.Start(); err != nil {
 		logger.Fatal("unable to start build reporter", zap.Error(err))

src/m3ninx/index/segment/fst/regexp/regexp.go

@@ -41,7 +41,7 @@ func ParseRegexp(pattern string) (a *vregexp.Regexp, prefixBeg, prefixEnd []byte
 // ParsedRegexp uses the pre-parsed regexp pattern and creates an equivalent matching automaton, and
 // corresponding keys to bound prefix beginning and end during the FST search.
 func ParsedRegexp(pattern string, parsed *syntax.Regexp) (a *vregexp.Regexp, prefixBeg, prefixEnd []byte, err error) {
-	re, err := vregexp.NewParsedWithLimit(pattern, parsed, vregexp.DefaultLimit)
+	re, err := vregexp.NewParsedWithLimit(pattern, parsed, vregexp.DefaultLimit())
 	if err != nil {
 		return nil, nil, nil, err
 	}