Merge pull request #1232 from grafana/partitionedIdx

Add partitioned index
grafana · Mar 8, 2019 · 20f6fde · 20f6fde
2 parents 2a90de3 + b83a9a0
commit 20f6fde
Show file tree

Hide file tree

Showing 16 changed files with 1,018 additions and 131 deletions.
diff --git a/cmd/metrictank/metrictank.go b/cmd/metrictank/metrictank.go
@@ -182,7 +182,6 @@ func main() {
 	notifierKafka.ConfigProcess(*instance)
 	statsConfig.ConfigProcess(*instance)
 	mdata.ConfigProcess()
-	memory.ConfigProcess()
 	cassandra.ConfigProcess()
 	bigtable.ConfigProcess()
 	bigtableStore.ConfigProcess(mdata.MaxChunkSpan())

diff --git a/docker/docker-chaos/metrictank.ini b/docker/docker-chaos/metrictank.ini
@@ -394,6 +394,8 @@ match-cache-size = 1000
 rules-file = /etc/metrictank/index-rules.conf
 # maximum duration each second a prune job can lock the index.
 max-prune-lock-time = 100ms
+# use separate indexes per partition
+partitioned = false
 
 ### Bigtable index
 [bigtable-idx]

diff --git a/docker/docker-cluster/metrictank.ini b/docker/docker-cluster/metrictank.ini
@@ -394,6 +394,8 @@ match-cache-size = 1000
 rules-file = /etc/metrictank/index-rules.conf
 # maximum duration each second a prune job can lock the index.
 max-prune-lock-time = 100ms
+# use separate indexes per partition
+partitioned = false
 
 ### Bigtable index
 [bigtable-idx]

diff --git a/docker/docker-dev-custom-cfg-kafka/metrictank.ini b/docker/docker-dev-custom-cfg-kafka/metrictank.ini
@@ -394,6 +394,8 @@ match-cache-size = 1000
 rules-file = /etc/metrictank/index-rules.conf
 # maximum duration each second a prune job can lock the index.
 max-prune-lock-time = 100ms
+# use separate indexes per partition
+partitioned = false
 
 ### Bigtable index
 [bigtable-idx]

diff --git a/docs/config.md b/docs/config.md
@@ -461,6 +461,8 @@ match-cache-size = 1000
 rules-file = /etc/metrictank/index-rules.conf
 # maximum duration each second a prune job can lock the index.
 max-prune-lock-time = 100ms
+# use separate indexes per partition
+partitioned = false
 ```
 
 ### Bigtable index

diff --git a/idx/bigtable/bigtable.go b/idx/bigtable/bigtable.go
@@ -54,7 +54,7 @@ type writeReq struct {
 }
 
 type BigtableIdx struct {
-	memory.MemoryIdx
+	memory.MemoryIndex
 	cfg        *IdxConfig
 	tbl        *bigtable.Table
 	client     *bigtable.Client
@@ -70,9 +70,9 @@ func New(cfg *IdxConfig) *BigtableIdx {
 		log.Fatalf("bigtable-idx: %s", err)
 	}
 	idx := &BigtableIdx{
-		MemoryIdx: *memory.New(),
-		cfg:       cfg,
-		shutdown:  make(chan struct{}),
+		MemoryIndex: memory.New(),
+		cfg:         cfg,
+		shutdown:    make(chan struct{}),
 	}
 	if cfg.UpdateBigtableIdx {
 		idx.writeQueue = make(chan writeReq, cfg.WriteQueueSize-cfg.WriteMaxFlushSize)
@@ -165,7 +165,7 @@ func (b *BigtableIdx) InitBare() error {
 // rebuilds the in-memory index, sets up write queues, metrics and pruning routines
 func (b *BigtableIdx) Init() error {
 	log.Infof("bigtable-idx: Initializing. Project=%s, Instance=%s", b.cfg.GcpProject, b.cfg.BigtableInstance)
-	if err := b.MemoryIdx.Init(); err != nil {
+	if err := b.MemoryIndex.Init(); err != nil {
 		return err
 	}
 
@@ -190,7 +190,7 @@ func (b *BigtableIdx) Init() error {
 }
 
 func (b *BigtableIdx) Stop() {
-	b.MemoryIdx.Stop()
+	b.MemoryIndex.Stop()
 	close(b.shutdown)
 	if b.cfg.UpdateBigtableIdx {
 		close(b.writeQueue)
@@ -208,7 +208,7 @@ func (b *BigtableIdx) Stop() {
 func (b *BigtableIdx) Update(point schema.MetricPoint, partition int32) (idx.Archive, int32, bool) {
 	pre := time.Now()
 
-	archive, oldPartition, inMemory := b.MemoryIdx.Update(point, partition)
+	archive, oldPartition, inMemory := b.MemoryIndex.Update(point, partition)
 
 	if !b.cfg.UpdateBigtableIdx {
 		statUpdateDuration.Value(time.Since(pre))
@@ -241,7 +241,7 @@ func (b *BigtableIdx) Update(point schema.MetricPoint, partition int32) (idx.Arc
 func (b *BigtableIdx) AddOrUpdate(mkey schema.MKey, data *schema.MetricData, partition int32) (idx.Archive, int32, bool) {
 	pre := time.Now()
 
-	archive, oldPartition, inMemory := b.MemoryIdx.AddOrUpdate(mkey, data, partition)
+	archive, oldPartition, inMemory := b.MemoryIndex.AddOrUpdate(mkey, data, partition)
 
 	stat := statUpdateDuration
 	if !inMemory {
@@ -286,7 +286,7 @@ func (b *BigtableIdx) updateBigtable(now uint32, inMemory bool, archive idx.Arch
 		log.Debugf("bigtable-idx: updating def %s in index.", archive.MetricDefinition.Id)
 		b.writeQueue <- writeReq{recvTime: time.Now(), def: &archive.MetricDefinition}
 		archive.LastSave = now
-		b.MemoryIdx.UpdateArchive(archive)
+		b.MemoryIndex.UpdateArchive(archive)
 	} else {
 		// perform a non-blocking write to the writeQueue. If the queue is full, then
 		// this will fail and we won't update the LastSave timestamp. The next time
@@ -297,7 +297,7 @@ func (b *BigtableIdx) updateBigtable(now uint32, inMemory bool, archive idx.Arch
 		select {
 		case b.writeQueue <- writeReq{recvTime: time.Now(), def: &archive.MetricDefinition}:
 			archive.LastSave = now
-			b.MemoryIdx.UpdateArchive(archive)
+			b.MemoryIndex.UpdateArchive(archive)
 		default:
 			statSaveSkipped.Inc()
 			log.Debugf("bigtable-idx: writeQueue is full, update of %s not saved this time", archive.MetricDefinition.Id)
@@ -315,7 +315,7 @@ func (b *BigtableIdx) rebuildIndex() {
 	var defs []schema.MetricDefinition
 	for _, partition := range cluster.Manager.GetPartitions() {
 		defs = b.LoadPartition(partition, defs[:0], pre)
-		num += b.MemoryIdx.Load(defs)
+		num += b.MemoryIndex.LoadPartition(partition, defs)
 	}
 
 	log.Infof("bigtable-idx: Rebuilding Memory Index Complete. Imported %d. Took %s", num, time.Since(pre))
@@ -459,7 +459,7 @@ LOOP:
 
 func (b *BigtableIdx) Delete(orgId uint32, pattern string) ([]idx.Archive, error) {
 	pre := time.Now()
-	defs, err := b.MemoryIdx.Delete(orgId, pattern)
+	defs, err := b.MemoryIndex.Delete(orgId, pattern)
 	if err != nil {
 		return defs, err
 	}
@@ -498,7 +498,7 @@ func (b *BigtableIdx) deleteRow(key string) error {
 
 func (b *BigtableIdx) Prune(now time.Time) ([]idx.Archive, error) {
 	log.Info("bigtable-idx: start pruning of series")
-	pruned, err := b.MemoryIdx.Prune(now)
+	pruned, err := b.MemoryIndex.Prune(now)
 	duration := time.Since(now)
 	if err != nil {
 		log.Errorf("bigtable-idx: prune error. %s", err)

diff --git a/idx/cassandra/cassandra.go b/idx/cassandra/cassandra.go
@@ -57,7 +57,7 @@ type writeReq struct {
 
 // CasIdx implements the the "MetricIndex" interface
 type CasIdx struct {
-	memory.MemoryIdx
+	memory.MemoryIndex
 	cfg              *IdxConfig
 	cluster          *gocql.ClusterConfig
 	session          *gocql.Session
@@ -96,7 +96,7 @@ func New(cfg *IdxConfig) *CasIdx {
 	}
 
 	idx := &CasIdx{
-		MemoryIdx:        *memory.New(),
+		MemoryIndex:      memory.New(),
 		cfg:              cfg,
 		cluster:          cluster,
 		updateInterval32: uint32(cfg.updateInterval.Nanoseconds() / int64(time.Second)),
@@ -180,7 +180,7 @@ func (c *CasIdx) InitBare() error {
 // rebuilds the in-memory index, sets up write queues, metrics and pruning routines
 func (c *CasIdx) Init() error {
 	log.Infof("initializing cassandra-idx. Hosts=%s", c.cfg.hosts)
-	if err := c.MemoryIdx.Init(); err != nil {
+	if err := c.MemoryIndex.Init(); err != nil {
 		return err
 	}
 
@@ -207,7 +207,7 @@ func (c *CasIdx) Init() error {
 
 func (c *CasIdx) Stop() {
 	log.Info("cassandra-idx: stopping")
-	c.MemoryIdx.Stop()
+	c.MemoryIndex.Stop()
 
 	// if updateCassIdx is disabled then writeQueue should never have been initialized
 	if c.cfg.updateCassIdx {
@@ -222,7 +222,7 @@ func (c *CasIdx) Stop() {
 func (c *CasIdx) Update(point schema.MetricPoint, partition int32) (idx.Archive, int32, bool) {
 	pre := time.Now()
 
-	archive, oldPartition, inMemory := c.MemoryIdx.Update(point, partition)
+	archive, oldPartition, inMemory := c.MemoryIndex.Update(point, partition)
 
 	if !c.cfg.updateCassIdx {
 		statUpdateDuration.Value(time.Since(pre))
@@ -251,7 +251,7 @@ func (c *CasIdx) Update(point schema.MetricPoint, partition int32) (idx.Archive,
 func (c *CasIdx) AddOrUpdate(mkey schema.MKey, data *schema.MetricData, partition int32) (idx.Archive, int32, bool) {
 	pre := time.Now()
 
-	archive, oldPartition, inMemory := c.MemoryIdx.AddOrUpdate(mkey, data, partition)
+	archive, oldPartition, inMemory := c.MemoryIndex.AddOrUpdate(mkey, data, partition)
 
 	stat := statUpdateDuration
 	if !inMemory {
@@ -291,7 +291,7 @@ func (c *CasIdx) updateCassandra(now uint32, inMemory bool, archive idx.Archive,
 		log.Debugf("cassandra-idx: updating def %s in index.", archive.MetricDefinition.Id)
 		c.writeQueue <- writeReq{recvTime: time.Now(), def: &archive.MetricDefinition}
 		archive.LastSave = now
-		c.MemoryIdx.UpdateArchive(archive)
+		c.MemoryIndex.UpdateArchive(archive)
 	} else {
 		// perform a non-blocking write to the writeQueue. If the queue is full, then
 		// this will fail and we won't update the LastSave timestamp. The next time
@@ -302,7 +302,7 @@ func (c *CasIdx) updateCassandra(now uint32, inMemory bool, archive idx.Archive,
 		select {
 		case c.writeQueue <- writeReq{recvTime: time.Now(), def: &archive.MetricDefinition}:
 			archive.LastSave = now
-			c.MemoryIdx.UpdateArchive(archive)
+			c.MemoryIndex.UpdateArchive(archive)
 		default:
 			statSaveSkipped.Inc()
 			log.Debugf("cassandra-idx: writeQueue is full, update of %s not saved this time.", archive.MetricDefinition.Id)
@@ -315,8 +315,12 @@ func (c *CasIdx) updateCassandra(now uint32, inMemory bool, archive idx.Archive,
 func (c *CasIdx) rebuildIndex() {
 	log.Info("cassandra-idx: Rebuilding Memory Index from metricDefinitions in Cassandra")
 	pre := time.Now()
-	defs := c.LoadPartitions(cluster.Manager.GetPartitions(), nil, pre)
-	num := c.MemoryIdx.Load(defs)
+	var defs []schema.MetricDefinition
+	var num int
+	for _, partition := range cluster.Manager.GetPartitions() {
+		defs = c.LoadPartitions([]int32{partition}, defs[:0], pre)
+		num += c.MemoryIndex.LoadPartition(partition, defs)
+	}
 	log.Infof("cassandra-idx: Rebuilding Memory Index Complete. Imported %d. Took %s", num, time.Since(pre))
 }
 
@@ -549,7 +553,7 @@ func (c *CasIdx) addDefToArchive(def schema.MetricDefinition) error {
 
 func (c *CasIdx) Delete(orgId uint32, pattern string) ([]idx.Archive, error) {
 	pre := time.Now()
-	defs, err := c.MemoryIdx.Delete(orgId, pattern)
+	defs, err := c.MemoryIndex.Delete(orgId, pattern)
 	if err != nil {
 		return defs, err
 	}
@@ -596,7 +600,7 @@ func (c *CasIdx) deleteDefAsync(key schema.MKey, part int32) {
 
 func (c *CasIdx) Prune(now time.Time) ([]idx.Archive, error) {
 	log.Info("cassandra-idx: start pruning of series")
-	pruned, err := c.MemoryIdx.Prune(now)
+	pruned, err := c.MemoryIndex.Prune(now)
 	duration := time.Since(now)
 	if err != nil {
 		log.Errorf("cassandra-idx: pruning error: %s", err)

diff --git a/idx/cassandra/cassandra_test.go b/idx/cassandra/cassandra_test.go
@@ -79,7 +79,7 @@ func init() {
 }
 
 func initForTests(c *CasIdx) error {
-	return c.MemoryIdx.Init()
+	return c.MemoryIndex.Init()
 }
 
 func getSeriesNames(depth, count int, prefix string) []string {
@@ -293,7 +293,7 @@ func TestAddToWriteQueue(t *testing.T) {
 			So(time.Now(), ShouldHappenAfter, pre.Add(time.Second))
 		})
 	})
-	ix.MemoryIdx.Stop()
+	ix.MemoryIndex.Stop()
 	close(ix.writeQueue)
 }