Use sarama built in support for consumer groups (#6172)

This commit is contained in:
Daniel Nelson 2019-07-29 20:41:12 -07:00 committed by GitHub
parent 4d73290e13
commit aea09b3a20
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
8 changed files with 691 additions and 481 deletions

108
Gopkg.lock generated
View File

@ -61,6 +61,14 @@
revision = "1f7cd6cfe0adea687ad44a512dfe76140f804318" revision = "1f7cd6cfe0adea687ad44a512dfe76140f804318"
version = "v10.12.0" version = "v10.12.0"
[[projects]]
digest = "1:82041ab48e5c76da656b723fdc13a2b9ec716cdc736f82adaac77f5c39d4fca8"
name = "github.com/DataDog/zstd"
packages = ["."]
pruneopts = ""
revision = "2347a397da4ee9c6b8226d4aff82c302d0e52773"
version = "v1.4.1"
[[projects]] [[projects]]
branch = "master" branch = "master"
digest = "1:298712a3ee36b59c3ca91f4183bd75d174d5eaa8b4aed5072831f126e2e752f6" digest = "1:298712a3ee36b59c3ca91f4183bd75d174d5eaa8b4aed5072831f126e2e752f6"
@ -81,12 +89,12 @@
version = "v0.4.9" version = "v0.4.9"
[[projects]] [[projects]]
digest = "1:213b41361ad1cb4768add9d26c2e27794c65264eefdb24ed6ea34cdfeeff3f3c" digest = "1:5dd52495eaf9fad11f4742f341166aa9eb68f70061fc1a9b546f9481b284b6d8"
name = "github.com/Shopify/sarama" name = "github.com/Shopify/sarama"
packages = ["."] packages = ["."]
pruneopts = "" pruneopts = ""
revision = "a6144ae922fd99dd0ea5046c8137acfb7fab0914" revision = "46c83074a05474240f9620fb7c70fb0d80ca401a"
version = "v1.18.0" version = "v1.23.1"
[[projects]] [[projects]]
digest = "1:f82b8ac36058904227087141017bb82f4b0fc58272990a4cdae3e2d6d222644e" digest = "1:f82b8ac36058904227087141017bb82f4b0fc58272990a4cdae3e2d6d222644e"
@ -195,14 +203,6 @@
pruneopts = "" pruneopts = ""
revision = "3a771d992973f24aa725d07868b467d1ddfceafb" revision = "3a771d992973f24aa725d07868b467d1ddfceafb"
[[projects]]
digest = "1:c5978131c797af795972c27c25396c81d1bf53b7b6e8e3e0259e58375765c071"
name = "github.com/bsm/sarama-cluster"
packages = ["."]
pruneopts = ""
revision = "cf455bc755fe41ac9bb2861e7a961833d9c2ecc3"
version = "v2.1.13"
[[projects]] [[projects]]
digest = "1:e5691038f8e87e7da05280095d968e50c17d624e25cca095d4e4cd947a805563" digest = "1:e5691038f8e87e7da05280095d968e50c17d624e25cca095d4e4cd947a805563"
name = "github.com/caio/go-tdigest" name = "github.com/caio/go-tdigest"
@ -640,6 +640,14 @@
pruneopts = "" pruneopts = ""
revision = "6bb64b370b90e7ef1fa532be9e591a81c3493e00" revision = "6bb64b370b90e7ef1fa532be9e591a81c3493e00"
[[projects]]
digest = "1:0038a7f43b51c8b2a8cd03b5372e73f8eadfe156484c2ae8185ae836f8ebc2cd"
name = "github.com/hashicorp/go-uuid"
packages = ["."]
pruneopts = ""
revision = "4f571afc59f3043a65f8fe6bf46d887b10a01d43"
version = "v1.0.1"
[[projects]] [[projects]]
digest = "1:f72168ea995f398bab88e84bd1ff58a983466ba162fb8d50d47420666cd57fad" digest = "1:f72168ea995f398bab88e84bd1ff58a983466ba162fb8d50d47420666cd57fad"
name = "github.com/hashicorp/serf" name = "github.com/hashicorp/serf"
@ -710,6 +718,17 @@
revision = "8faa4453fc7051d1076053f8854077753ab912f2" revision = "8faa4453fc7051d1076053f8854077753ab912f2"
version = "v3.4.0" version = "v3.4.0"
[[projects]]
digest = "1:d45477e90c25c8c6d7d4237281167aa56079382fc042db4b44a8328071649bfa"
name = "github.com/jcmturner/gofork"
packages = [
"encoding/asn1",
"x/crypto/pbkdf2",
]
pruneopts = ""
revision = "dc7c13fece037a4a36e2b3c69db4991498d30692"
version = "v1.0.0"
[[projects]] [[projects]]
digest = "1:6f49eae0c1e5dab1dafafee34b207aeb7a42303105960944828c2079b92fc88e" digest = "1:6f49eae0c1e5dab1dafafee34b207aeb7a42303105960944828c2079b92fc88e"
name = "github.com/jmespath/go-jmespath" name = "github.com/jmespath/go-jmespath"
@ -1523,6 +1542,72 @@
revision = "d2d2541c53f18d2a059457998ce2876cc8e67cbf" revision = "d2d2541c53f18d2a059457998ce2876cc8e67cbf"
version = "v0.9.1" version = "v0.9.1"
[[projects]]
digest = "1:4777ba481cc12866b89aafb0a67529e7ac48b9aea06a25f3737b2cf5a3ffda12"
name = "gopkg.in/jcmturner/aescts.v1"
packages = ["."]
pruneopts = ""
revision = "f6abebb3171c4c1b1fea279cb7c7325020a26290"
version = "v1.0.1"
[[projects]]
digest = "1:84c5b1392ef65ad1bb64da4b4d0beb2f204eefc769d6d96082347bb7057cb7b1"
name = "gopkg.in/jcmturner/dnsutils.v1"
packages = ["."]
pruneopts = ""
revision = "13eeb8d49ffb74d7a75784c35e4d900607a3943c"
version = "v1.0.1"
[[projects]]
digest = "1:502ab576ba8c47c4de77fe3f2b2386adc1a1447bb5afae2ac7bf0edd2b6f7c52"
name = "gopkg.in/jcmturner/gokrb5.v7"
packages = [
"asn1tools",
"client",
"config",
"credentials",
"crypto",
"crypto/common",
"crypto/etype",
"crypto/rfc3961",
"crypto/rfc3962",
"crypto/rfc4757",
"crypto/rfc8009",
"gssapi",
"iana",
"iana/addrtype",
"iana/adtype",
"iana/asnAppTag",
"iana/chksumtype",
"iana/errorcode",
"iana/etypeID",
"iana/flags",
"iana/keyusage",
"iana/msgtype",
"iana/nametype",
"iana/patype",
"kadmin",
"keytab",
"krberror",
"messages",
"pac",
"types",
]
pruneopts = ""
revision = "363118e62befa8a14ff01031c025026077fe5d6d"
version = "v7.3.0"
[[projects]]
digest = "1:f9956ccc103c6208cd50c71ee5191b6fdcc635972c12624ef949c9b20b2bb9d1"
name = "gopkg.in/jcmturner/rpc.v1"
packages = [
"mstypes",
"ndr",
]
pruneopts = ""
revision = "99a8ce2fbf8b8087b6ed12a37c61b10f04070043"
version = "v1.1.0"
[[projects]] [[projects]]
digest = "1:367baf06b7dbd0ef0bbdd785f6a79f929c96b0c18e9d3b29c0eed1ac3f5db133" digest = "1:367baf06b7dbd0ef0bbdd785f6a79f929c96b0c18e9d3b29c0eed1ac3f5db133"
name = "gopkg.in/ldap.v2" name = "gopkg.in/ldap.v2"
@ -1598,7 +1683,6 @@
"github.com/aws/aws-sdk-go/service/cloudwatch", "github.com/aws/aws-sdk-go/service/cloudwatch",
"github.com/aws/aws-sdk-go/service/dynamodb", "github.com/aws/aws-sdk-go/service/dynamodb",
"github.com/aws/aws-sdk-go/service/kinesis", "github.com/aws/aws-sdk-go/service/kinesis",
"github.com/bsm/sarama-cluster",
"github.com/cisco-ie/nx-telemetry-proto/mdt_dialout", "github.com/cisco-ie/nx-telemetry-proto/mdt_dialout",
"github.com/cisco-ie/nx-telemetry-proto/telemetry_bis", "github.com/cisco-ie/nx-telemetry-proto/telemetry_bis",
"github.com/couchbase/go-couchbase", "github.com/couchbase/go-couchbase",

View File

@ -18,10 +18,6 @@
name = "github.com/aws/aws-sdk-go" name = "github.com/aws/aws-sdk-go"
version = "1.15.54" version = "1.15.54"
[[constraint]]
name = "github.com/bsm/sarama-cluster"
version = "2.1.13"
[[constraint]] [[constraint]]
name = "github.com/couchbase/go-couchbase" name = "github.com/couchbase/go-couchbase"
branch = "master" branch = "master"

View File

@ -1,9 +1,8 @@
ifeq ($(SHELL), cmd) ifeq ($(OS), Windows_NT)
VERSION := $(shell git describe --exact-match --tags 2>nil)
HOME := $(HOMEPATH)
else ifeq ($(SHELL), sh.exe)
VERSION := $(shell git describe --exact-match --tags 2>nil) VERSION := $(shell git describe --exact-match --tags 2>nil)
HOME := $(HOMEPATH) HOME := $(HOMEPATH)
CGO_ENABLED ?= 0
export CGO_ENABLED
else else
VERSION := $(shell git describe --exact-match --tags 2>/dev/null) VERSION := $(shell git describe --exact-match --tags 2>/dev/null)
endif endif
@ -48,7 +47,6 @@ install: telegraf
mkdir -p $(DESTDIR)$(PREFIX)/bin/ mkdir -p $(DESTDIR)$(PREFIX)/bin/
cp telegraf $(DESTDIR)$(PREFIX)/bin/ cp telegraf $(DESTDIR)$(PREFIX)/bin/
.PHONY: test .PHONY: test
test: test:
go test -short ./... go test -short ./...

View File

@ -14,7 +14,6 @@ following works:
- github.com/aws/aws-sdk-go [Apache License 2.0](https://github.com/aws/aws-sdk-go/blob/master/LICENSE.txt) - github.com/aws/aws-sdk-go [Apache License 2.0](https://github.com/aws/aws-sdk-go/blob/master/LICENSE.txt)
- github.com/Azure/go-autorest [Apache License 2.0](https://github.com/Azure/go-autorest/blob/master/LICENSE) - github.com/Azure/go-autorest [Apache License 2.0](https://github.com/Azure/go-autorest/blob/master/LICENSE)
- github.com/beorn7/perks [MIT License](https://github.com/beorn7/perks/blob/master/LICENSE) - github.com/beorn7/perks [MIT License](https://github.com/beorn7/perks/blob/master/LICENSE)
- github.com/bsm/sarama-cluster [MIT License](https://github.com/bsm/sarama-cluster/blob/master/LICENSE)
- github.com/cenkalti/backoff [MIT License](https://github.com/cenkalti/backoff/blob/master/LICENSE) - github.com/cenkalti/backoff [MIT License](https://github.com/cenkalti/backoff/blob/master/LICENSE)
- github.com/cisco-ie/nx-telemetry-proto [Apache License 2.0](https://github.com/cisco-ie/nx-telemetry-proto/blob/master/LICENSE) - github.com/cisco-ie/nx-telemetry-proto [Apache License 2.0](https://github.com/cisco-ie/nx-telemetry-proto/blob/master/LICENSE)
- github.com/couchbase/go-couchbase [MIT License](https://github.com/couchbase/go-couchbase/blob/master/LICENSE) - github.com/couchbase/go-couchbase [MIT License](https://github.com/couchbase/go-couchbase/blob/master/LICENSE)

View File

@ -10,11 +10,13 @@ and use the old zookeeper connection method.
```toml ```toml
[[inputs.kafka_consumer]] [[inputs.kafka_consumer]]
## kafka servers ## Kafka brokers.
brokers = ["localhost:9092"] brokers = ["localhost:9092"]
## topic(s) to consume
## Topics to consume.
topics = ["telegraf"] topics = ["telegraf"]
## Add topic as tag if topic_tag is not empty
## When set this tag will be added to all metrics with the topic as the value.
# topic_tag = "" # topic_tag = ""
## Optional Client id ## Optional Client id
@ -37,10 +39,11 @@ and use the old zookeeper connection method.
# sasl_username = "kafka" # sasl_username = "kafka"
# sasl_password = "secret" # sasl_password = "secret"
## the name of the consumer group ## Name of the consumer group.
consumer_group = "telegraf_metrics_consumers" # consumer_group = "telegraf_metrics_consumers"
## Offset (must be either "oldest" or "newest")
offset = "oldest" ## Initial offset position; one of "oldest" or "newest".
# offset = "oldest"
## Maximum length of a message to consume, in bytes (default 0/unlimited); ## Maximum length of a message to consume, in bytes (default 0/unlimited);
## larger messages are dropped ## larger messages are dropped

View File

@ -8,61 +8,20 @@ import (
"sync" "sync"
"github.com/Shopify/sarama" "github.com/Shopify/sarama"
cluster "github.com/bsm/sarama-cluster"
"github.com/influxdata/telegraf" "github.com/influxdata/telegraf"
"github.com/influxdata/telegraf/internal/tls" "github.com/influxdata/telegraf/internal/tls"
"github.com/influxdata/telegraf/plugins/inputs" "github.com/influxdata/telegraf/plugins/inputs"
"github.com/influxdata/telegraf/plugins/parsers" "github.com/influxdata/telegraf/plugins/parsers"
) )
const ( const sampleConfig = `
defaultMaxUndeliveredMessages = 1000 ## Kafka brokers.
)
type empty struct{}
type semaphore chan empty
type Consumer interface {
Errors() <-chan error
Messages() <-chan *sarama.ConsumerMessage
MarkOffset(msg *sarama.ConsumerMessage, metadata string)
Close() error
}
type Kafka struct {
ConsumerGroup string `toml:"consumer_group"`
ClientID string `toml:"client_id"`
Topics []string `toml:"topics"`
Brokers []string `toml:"brokers"`
MaxMessageLen int `toml:"max_message_len"`
Version string `toml:"version"`
MaxUndeliveredMessages int `toml:"max_undelivered_messages"`
Offset string `toml:"offset"`
SASLUsername string `toml:"sasl_username"`
SASLPassword string `toml:"sasl_password"`
TopicTag string `toml:"topic_tag"`
tls.ClientConfig
cluster Consumer
parser parsers.Parser
wg *sync.WaitGroup
cancel context.CancelFunc
// Unconfirmed messages
messages map[telegraf.TrackingID]*sarama.ConsumerMessage
// doNotCommitMsgs tells the parser not to call CommitUpTo on the consumer
// this is mostly for test purposes, but there may be a use-case for it later.
doNotCommitMsgs bool
}
var sampleConfig = `
## kafka servers
brokers = ["localhost:9092"] brokers = ["localhost:9092"]
## topic(s) to consume
## Topics to consume.
topics = ["telegraf"] topics = ["telegraf"]
## Add topic as tag if topic_tag is not empty
## When set this tag will be added to all metrics with the topic as the value.
# topic_tag = "" # topic_tag = ""
## Optional Client id ## Optional Client id
@ -85,10 +44,12 @@ var sampleConfig = `
# sasl_username = "kafka" # sasl_username = "kafka"
# sasl_password = "secret" # sasl_password = "secret"
## the name of the consumer group ## Name of the consumer group.
consumer_group = "telegraf_metrics_consumers" # consumer_group = "telegraf_metrics_consumers"
## Offset (must be either "oldest" or "newest")
offset = "oldest" ## Initial offset position; one of "oldest" or "newest".
# offset = "oldest"
## Maximum length of a message to consume, in bytes (default 0/unlimited); ## Maximum length of a message to consume, in bytes (default 0/unlimited);
## larger messages are dropped ## larger messages are dropped
max_message_len = 1000000 max_message_len = 1000000
@ -110,22 +71,77 @@ var sampleConfig = `
data_format = "influx" data_format = "influx"
` `
func (k *Kafka) SampleConfig() string { const (
defaultMaxUndeliveredMessages = 1000
defaultMaxMessageLen = 1000000
defaultConsumerGroup = "telegraf_metrics_consumers"
)
type empty struct{}
type semaphore chan empty
type KafkaConsumer struct {
Brokers []string `toml:"brokers"`
ClientID string `toml:"client_id"`
ConsumerGroup string `toml:"consumer_group"`
MaxMessageLen int `toml:"max_message_len"`
MaxUndeliveredMessages int `toml:"max_undelivered_messages"`
Offset string `toml:"offset"`
Topics []string `toml:"topics"`
TopicTag string `toml:"topic_tag"`
Version string `toml:"version"`
SASLPassword string `toml:"sasl_password"`
SASLUsername string `toml:"sasl_username"`
tls.ClientConfig
ConsumerCreator ConsumerGroupCreator `toml:"-"`
consumer ConsumerGroup
config *sarama.Config
parser parsers.Parser
wg sync.WaitGroup
cancel context.CancelFunc
}
type ConsumerGroup interface {
Consume(ctx context.Context, topics []string, handler sarama.ConsumerGroupHandler) error
Errors() <-chan error
Close() error
}
type ConsumerGroupCreator interface {
Create(brokers []string, group string, config *sarama.Config) (ConsumerGroup, error)
}
type SaramaCreator struct{}
func (*SaramaCreator) Create(brokers []string, group string, config *sarama.Config) (ConsumerGroup, error) {
return sarama.NewConsumerGroup(brokers, group, config)
}
func (k *KafkaConsumer) SampleConfig() string {
return sampleConfig return sampleConfig
} }
func (k *Kafka) Description() string { func (k *KafkaConsumer) Description() string {
return "Read metrics from Kafka topic(s)" return "Read metrics from Kafka topics"
} }
func (k *Kafka) SetParser(parser parsers.Parser) { func (k *KafkaConsumer) SetParser(parser parsers.Parser) {
k.parser = parser k.parser = parser
} }
func (k *Kafka) Start(acc telegraf.Accumulator) error { func (k *KafkaConsumer) Init() error {
var clusterErr error if k.MaxUndeliveredMessages == 0 {
k.MaxUndeliveredMessages = defaultMaxUndeliveredMessages
}
if k.ConsumerGroup == "" {
k.ConsumerGroup = defaultConsumerGroup
}
config := cluster.NewConfig() config := sarama.NewConfig()
config.Consumer.Return.Errors = true
if k.Version != "" { if k.Version != "" {
version, err := sarama.ParseKafkaVersion(k.Version) version, err := sarama.ParseKafkaVersion(k.Version)
@ -135,172 +151,255 @@ func (k *Kafka) Start(acc telegraf.Accumulator) error {
config.Version = version config.Version = version
} }
config.Consumer.Return.Errors = true
tlsConfig, err := k.ClientConfig.TLSConfig() tlsConfig, err := k.ClientConfig.TLSConfig()
if err != nil { if err != nil {
return err return err
} }
if tlsConfig != nil {
config.Net.TLS.Config = tlsConfig
config.Net.TLS.Enable = true
}
if k.SASLUsername != "" && k.SASLPassword != "" {
config.Net.SASL.User = k.SASLUsername
config.Net.SASL.Password = k.SASLPassword
config.Net.SASL.Enable = true
}
if k.ClientID != "" { if k.ClientID != "" {
config.ClientID = k.ClientID config.ClientID = k.ClientID
} else { } else {
config.ClientID = "Telegraf" config.ClientID = "Telegraf"
} }
if tlsConfig != nil {
log.Printf("D! TLS Enabled")
config.Net.TLS.Config = tlsConfig
config.Net.TLS.Enable = true
}
if k.SASLUsername != "" && k.SASLPassword != "" {
log.Printf("D! Using SASL auth with username '%s',",
k.SASLUsername)
config.Net.SASL.User = k.SASLUsername
config.Net.SASL.Password = k.SASLPassword
config.Net.SASL.Enable = true
}
switch strings.ToLower(k.Offset) { switch strings.ToLower(k.Offset) {
case "oldest", "": case "oldest", "":
config.Consumer.Offsets.Initial = sarama.OffsetOldest config.Consumer.Offsets.Initial = sarama.OffsetOldest
case "newest": case "newest":
config.Consumer.Offsets.Initial = sarama.OffsetNewest config.Consumer.Offsets.Initial = sarama.OffsetNewest
default: default:
log.Printf("I! WARNING: Kafka consumer invalid offset '%s', using 'oldest'", return fmt.Errorf("invalid offset %q", k.Offset)
k.Offset)
config.Consumer.Offsets.Initial = sarama.OffsetOldest
} }
if k.cluster == nil { if k.ConsumerCreator == nil {
k.cluster, clusterErr = cluster.NewConsumer( k.ConsumerCreator = &SaramaCreator{}
}
k.config = config
return nil
}
func (k *KafkaConsumer) Start(acc telegraf.Accumulator) error {
var err error
k.consumer, err = k.ConsumerCreator.Create(
k.Brokers, k.Brokers,
k.ConsumerGroup, k.ConsumerGroup,
k.Topics, k.config,
config,
) )
if err != nil {
if clusterErr != nil { return err
log.Printf("E! Error when creating Kafka Consumer, brokers: %v, topics: %v",
k.Brokers, k.Topics)
return clusterErr
}
} }
ctx, cancel := context.WithCancel(context.Background()) ctx, cancel := context.WithCancel(context.Background())
k.cancel = cancel k.cancel = cancel
// Start consumer goroutine // Start consumer goroutine
k.wg = &sync.WaitGroup{}
k.wg.Add(1) k.wg.Add(1)
go func() { go func() {
defer k.wg.Done() defer k.wg.Done()
k.receiver(ctx, acc) for ctx.Err() == nil {
handler := NewConsumerGroupHandler(acc, k.MaxUndeliveredMessages, k.parser)
handler.MaxMessageLen = k.MaxMessageLen
handler.TopicTag = k.TopicTag
err := k.consumer.Consume(ctx, k.Topics, handler)
if err != nil {
acc.AddError(err)
}
}
err = k.consumer.Close()
if err != nil {
acc.AddError(err)
}
}()
k.wg.Add(1)
go func() {
defer k.wg.Done()
for err := range k.consumer.Errors() {
acc.AddError(err)
}
}() }()
log.Printf("I! Started the kafka consumer service, brokers: %v, topics: %v",
k.Brokers, k.Topics)
return nil return nil
} }
// receiver() reads all incoming messages from the consumer, and parses them into func (k *KafkaConsumer) Gather(acc telegraf.Accumulator) error {
// influxdb metric points. return nil
func (k *Kafka) receiver(ctx context.Context, ac telegraf.Accumulator) { }
k.messages = make(map[telegraf.TrackingID]*sarama.ConsumerMessage)
acc := ac.WithTracking(k.MaxUndeliveredMessages) func (k *KafkaConsumer) Stop() {
sem := make(semaphore, k.MaxUndeliveredMessages) k.cancel()
k.wg.Wait()
}
// Message is an aggregate type binding the Kafka message and the session so
// that offsets can be updated.
type Message struct {
message *sarama.ConsumerMessage
session sarama.ConsumerGroupSession
}
func NewConsumerGroupHandler(acc telegraf.Accumulator, maxUndelivered int, parser parsers.Parser) *ConsumerGroupHandler {
handler := &ConsumerGroupHandler{
acc: acc.WithTracking(maxUndelivered),
sem: make(chan empty, maxUndelivered),
undelivered: make(map[telegraf.TrackingID]Message, maxUndelivered),
parser: parser,
}
return handler
}
// ConsumerGroupHandler is a sarama.ConsumerGroupHandler implementation.
type ConsumerGroupHandler struct {
MaxMessageLen int
TopicTag string
acc telegraf.TrackingAccumulator
sem semaphore
parser parsers.Parser
wg sync.WaitGroup
cancel context.CancelFunc
mu sync.Mutex
undelivered map[telegraf.TrackingID]Message
}
// Setup is called once when a new session is opened. It setups up the handler
// and begins processing delivered messages.
func (h *ConsumerGroupHandler) Setup(sarama.ConsumerGroupSession) error {
h.undelivered = make(map[telegraf.TrackingID]Message)
ctx, cancel := context.WithCancel(context.Background())
h.cancel = cancel
h.wg.Add(1)
go func() {
defer h.wg.Done()
h.run(ctx)
}()
return nil
}
// Run processes any delivered metrics during the lifetime of the session.
func (h *ConsumerGroupHandler) run(ctx context.Context) error {
for { for {
select { select {
case <-ctx.Done(): case <-ctx.Done():
return
case track := <-acc.Delivered():
<-sem
k.onDelivery(track)
case err := <-k.cluster.Errors():
acc.AddError(err)
case sem <- empty{}:
select {
case <-ctx.Done():
return
case track := <-acc.Delivered():
// Once for the delivered message, once to leave the case
<-sem
<-sem
k.onDelivery(track)
case err := <-k.cluster.Errors():
<-sem
acc.AddError(err)
case msg := <-k.cluster.Messages():
err := k.onMessage(acc, msg)
if err != nil {
acc.AddError(err)
<-sem
}
}
}
}
}
func (k *Kafka) markOffset(msg *sarama.ConsumerMessage) {
if !k.doNotCommitMsgs {
k.cluster.MarkOffset(msg, "")
}
}
func (k *Kafka) onMessage(acc telegraf.TrackingAccumulator, msg *sarama.ConsumerMessage) error {
if k.MaxMessageLen != 0 && len(msg.Value) > k.MaxMessageLen {
k.markOffset(msg)
return fmt.Errorf("Message longer than max_message_len (%d > %d)",
len(msg.Value), k.MaxMessageLen)
}
metrics, err := k.parser.Parse(msg.Value)
if err != nil {
return err
}
if len(k.TopicTag) > 0 {
for _, metric := range metrics {
metric.AddTag(k.TopicTag, msg.Topic)
}
}
id := acc.AddTrackingMetricGroup(metrics)
k.messages[id] = msg
return nil return nil
case track := <-h.acc.Delivered():
h.onDelivery(track)
}
}
} }
func (k *Kafka) onDelivery(track telegraf.DeliveryInfo) { func (h *ConsumerGroupHandler) onDelivery(track telegraf.DeliveryInfo) {
msg, ok := k.messages[track.ID()] h.mu.Lock()
defer h.mu.Unlock()
msg, ok := h.undelivered[track.ID()]
if !ok { if !ok {
log.Printf("E! [inputs.kafka_consumer] Could not mark message delivered: %d", track.ID()) log.Printf("E! [inputs.kafka_consumer] Could not mark message delivered: %d", track.ID())
return return
} }
if track.Delivered() { if track.Delivered() {
k.markOffset(msg) msg.session.MarkMessage(msg.message, "")
} }
delete(k.messages, track.ID())
delete(h.undelivered, track.ID())
<-h.sem
} }
func (k *Kafka) Stop() { // Reserve blocks until there is an available slot for a new message.
k.cancel() func (h *ConsumerGroupHandler) Reserve(ctx context.Context) error {
k.wg.Wait() select {
case <-ctx.Done():
if err := k.cluster.Close(); err != nil { return ctx.Err()
log.Printf("E! [inputs.kafka_consumer] Error closing consumer: %v", err) case h.sem <- empty{}:
return nil
} }
} }
func (k *Kafka) Gather(acc telegraf.Accumulator) error { func (h *ConsumerGroupHandler) release() {
<-h.sem
}
// Handle processes a message and if successful saves it to be acknowledged
// after delivery.
func (h *ConsumerGroupHandler) Handle(session sarama.ConsumerGroupSession, msg *sarama.ConsumerMessage) error {
if h.MaxMessageLen != 0 && len(msg.Value) > h.MaxMessageLen {
session.MarkMessage(msg, "")
h.release()
return fmt.Errorf("message exceeds max_message_len (actual %d, max %d)",
len(msg.Value), h.MaxMessageLen)
}
metrics, err := h.parser.Parse(msg.Value)
if err != nil {
h.release()
return err
}
if len(h.TopicTag) > 0 {
for _, metric := range metrics {
metric.AddTag(h.TopicTag, msg.Topic)
}
}
id := h.acc.AddTrackingMetricGroup(metrics)
h.mu.Lock()
h.undelivered[id] = Message{session: session, message: msg}
h.mu.Unlock()
return nil
}
// ConsumeClaim is called once each claim in a goroutine and must be
// thread-safe. Should run until the claim is closed.
func (h *ConsumerGroupHandler) ConsumeClaim(session sarama.ConsumerGroupSession, claim sarama.ConsumerGroupClaim) error {
ctx := session.Context()
for {
err := h.Reserve(ctx)
if err != nil {
return nil
}
select {
case <-ctx.Done():
return nil
case msg, ok := <-claim.Messages():
if !ok {
return nil
}
err := h.Handle(session, msg)
if err != nil {
h.acc.AddError(err)
}
}
}
}
// Cleanup stops the internal goroutine and is called after all ConsumeClaim
// functions have completed.
func (h *ConsumerGroupHandler) Cleanup(sarama.ConsumerGroupSession) error {
h.cancel()
h.wg.Wait()
return nil return nil
} }
func init() { func init() {
inputs.Add("kafka_consumer", func() telegraf.Input { inputs.Add("kafka_consumer", func() telegraf.Input {
return &Kafka{ return &KafkaConsumer{}
MaxUndeliveredMessages: defaultMaxUndeliveredMessages,
}
}) })
} }

View File

@ -1,94 +0,0 @@
package kafka_consumer
import (
"fmt"
"testing"
"time"
"github.com/Shopify/sarama"
"github.com/influxdata/telegraf/testutil"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/influxdata/telegraf/plugins/parsers"
)
func TestReadsMetricsFromKafka(t *testing.T) {
if testing.Short() {
t.Skip("Skipping integration test in short mode")
}
brokerPeers := []string{testutil.GetLocalHost() + ":9092"}
testTopic := fmt.Sprintf("telegraf_test_topic_%d", time.Now().Unix())
// Send a Kafka message to the kafka host
msg := "cpu_load_short,direction=in,host=server01,region=us-west value=23422.0 1422568543702900257\n"
producer, err := sarama.NewSyncProducer(brokerPeers, nil)
require.NoError(t, err)
_, _, err = producer.SendMessage(
&sarama.ProducerMessage{
Topic: testTopic,
Value: sarama.StringEncoder(msg),
})
require.NoError(t, err)
defer producer.Close()
// Start the Kafka Consumer
k := &Kafka{
ConsumerGroup: "telegraf_test_consumers",
Topics: []string{testTopic},
Brokers: brokerPeers,
Offset: "oldest",
}
p, _ := parsers.NewInfluxParser()
k.SetParser(p)
// Verify that we can now gather the sent message
var acc testutil.Accumulator
// Sanity check
assert.Equal(t, 0, len(acc.Metrics), "There should not be any points")
if err := k.Start(&acc); err != nil {
t.Fatal(err.Error())
} else {
defer k.Stop()
}
waitForPoint(&acc, t)
// Gather points
err = acc.GatherError(k.Gather)
require.NoError(t, err)
if len(acc.Metrics) == 1 {
point := acc.Metrics[0]
assert.Equal(t, "cpu_load_short", point.Measurement)
assert.Equal(t, map[string]interface{}{"value": 23422.0}, point.Fields)
assert.Equal(t, map[string]string{
"host": "server01",
"direction": "in",
"region": "us-west",
}, point.Tags)
assert.Equal(t, time.Unix(0, 1422568543702900257).Unix(), point.Time.Unix())
} else {
t.Errorf("No points found in accumulator, expected 1")
}
}
// Waits for the metric that was sent to the kafka broker to arrive at the kafka
// consumer
func waitForPoint(acc *testutil.Accumulator, t *testing.T) {
// Give the kafka container up to 2 seconds to get the point to the consumer
ticker := time.NewTicker(5 * time.Millisecond)
counter := 0
for {
select {
case <-ticker.C:
counter++
if counter > 1000 {
t.Fatal("Waited for 5s, point never arrived to consumer")
} else if acc.NFields() == 1 {
return
}
}
}
}

View File

@ -2,219 +2,344 @@ package kafka_consumer
import ( import (
"context" "context"
"strings"
"testing" "testing"
"time"
"github.com/Shopify/sarama" "github.com/Shopify/sarama"
"github.com/influxdata/telegraf" "github.com/influxdata/telegraf"
"github.com/influxdata/telegraf/plugins/parsers" "github.com/influxdata/telegraf/plugins/parsers/value"
"github.com/influxdata/telegraf/testutil" "github.com/influxdata/telegraf/testutil"
"github.com/stretchr/testify/assert" "github.com/stretchr/testify/require"
) )
const ( type FakeConsumerGroup struct {
testMsg = "cpu_load_short,host=server01 value=23422.0 1422568543702900257\n" brokers []string
testMsgGraphite = "cpu.load.short.graphite 23422 1454780029" group string
testMsgJSON = "{\"a\": 5, \"b\": {\"c\": 6}}\n" config *sarama.Config
invalidMsg = "cpu_load_short,host=server01 1422568543702900257\n"
)
type TestConsumer struct { handler sarama.ConsumerGroupHandler
errors chan error errors chan error
messages chan *sarama.ConsumerMessage
} }
func (c *TestConsumer) Errors() <-chan error { func (g *FakeConsumerGroup) Consume(ctx context.Context, topics []string, handler sarama.ConsumerGroupHandler) error {
return c.errors g.handler = handler
} g.handler.Setup(nil)
func (c *TestConsumer) Messages() <-chan *sarama.ConsumerMessage {
return c.messages
}
func (c *TestConsumer) MarkOffset(msg *sarama.ConsumerMessage, metadata string) {
}
func (c *TestConsumer) Close() error {
return nil return nil
} }
func (c *TestConsumer) Inject(msg *sarama.ConsumerMessage) { func (g *FakeConsumerGroup) Errors() <-chan error {
c.messages <- msg return g.errors
} }
func newTestKafka() (*Kafka, *TestConsumer) { func (g *FakeConsumerGroup) Close() error {
consumer := &TestConsumer{ close(g.errors)
errors: make(chan error), return nil
messages: make(chan *sarama.ConsumerMessage, 1000), }
type FakeCreator struct {
ConsumerGroup *FakeConsumerGroup
}
func (c *FakeCreator) Create(brokers []string, group string, config *sarama.Config) (ConsumerGroup, error) {
c.ConsumerGroup.brokers = brokers
c.ConsumerGroup.group = group
c.ConsumerGroup.config = config
return c.ConsumerGroup, nil
}
func TestInit(t *testing.T) {
tests := []struct {
name string
plugin *KafkaConsumer
initError bool
check func(t *testing.T, plugin *KafkaConsumer)
}{
{
name: "default config",
plugin: &KafkaConsumer{},
check: func(t *testing.T, plugin *KafkaConsumer) {
require.Equal(t, plugin.ConsumerGroup, defaultConsumerGroup)
require.Equal(t, plugin.MaxUndeliveredMessages, defaultMaxUndeliveredMessages)
require.Equal(t, plugin.config.ClientID, "Telegraf")
require.Equal(t, plugin.config.Consumer.Offsets.Initial, sarama.OffsetOldest)
},
},
{
name: "parses valid version string",
plugin: &KafkaConsumer{
Version: "1.0.0",
},
check: func(t *testing.T, plugin *KafkaConsumer) {
require.Equal(t, plugin.config.Version, sarama.V1_0_0_0)
},
},
{
name: "invalid version string",
plugin: &KafkaConsumer{
Version: "100",
},
initError: true,
},
{
name: "custom client_id",
plugin: &KafkaConsumer{
ClientID: "custom",
},
check: func(t *testing.T, plugin *KafkaConsumer) {
require.Equal(t, plugin.config.ClientID, "custom")
},
},
{
name: "custom offset",
plugin: &KafkaConsumer{
Offset: "newest",
},
check: func(t *testing.T, plugin *KafkaConsumer) {
require.Equal(t, plugin.config.Consumer.Offsets.Initial, sarama.OffsetNewest)
},
},
{
name: "invalid offset",
plugin: &KafkaConsumer{
Offset: "middle",
},
initError: true,
},
} }
k := Kafka{ for _, tt := range tests {
cluster: consumer, t.Run(tt.name, func(t *testing.T) {
ConsumerGroup: "test", cg := &FakeConsumerGroup{}
Topics: []string{"telegraf"}, tt.plugin.ConsumerCreator = &FakeCreator{ConsumerGroup: cg}
Brokers: []string{"localhost:9092"}, err := tt.plugin.Init()
Offset: "oldest", if tt.initError {
MaxUndeliveredMessages: defaultMaxUndeliveredMessages, require.Error(t, err)
doNotCommitMsgs: true, return
messages: make(map[telegraf.TrackingID]*sarama.ConsumerMessage),
} }
return &k, consumer
}
func newTestKafkaWithTopicTag() (*Kafka, *TestConsumer) { tt.check(t, tt.plugin)
consumer := &TestConsumer{
errors: make(chan error),
messages: make(chan *sarama.ConsumerMessage, 1000),
}
k := Kafka{
cluster: consumer,
ConsumerGroup: "test",
Topics: []string{"telegraf"},
Brokers: []string{"localhost:9092"},
Offset: "oldest",
MaxUndeliveredMessages: defaultMaxUndeliveredMessages,
doNotCommitMsgs: true,
messages: make(map[telegraf.TrackingID]*sarama.ConsumerMessage),
TopicTag: "topic",
}
return &k, consumer
}
// Test that the parser parses kafka messages into points
func TestRunParser(t *testing.T) {
k, consumer := newTestKafka()
acc := testutil.Accumulator{}
ctx := context.Background()
k.parser, _ = parsers.NewInfluxParser()
go k.receiver(ctx, &acc)
consumer.Inject(saramaMsg(testMsg))
acc.Wait(1)
assert.Equal(t, acc.NFields(), 1)
}
// Test that the parser parses kafka messages into points
// and adds the topic tag
func TestRunParserWithTopic(t *testing.T) {
k, consumer := newTestKafkaWithTopicTag()
acc := testutil.Accumulator{}
ctx := context.Background()
k.parser, _ = parsers.NewInfluxParser()
go k.receiver(ctx, &acc)
consumer.Inject(saramaMsgWithTopic(testMsg, "test_topic"))
acc.Wait(1)
assert.Equal(t, acc.NFields(), 1)
assert.True(t, acc.HasTag("cpu_load_short", "topic"))
}
// Test that the parser ignores invalid messages
func TestRunParserInvalidMsg(t *testing.T) {
k, consumer := newTestKafka()
acc := testutil.Accumulator{}
ctx := context.Background()
k.parser, _ = parsers.NewInfluxParser()
go k.receiver(ctx, &acc)
consumer.Inject(saramaMsg(invalidMsg))
acc.WaitError(1)
assert.Equal(t, acc.NFields(), 0)
}
// Test that overlong messages are dropped
func TestDropOverlongMsg(t *testing.T) {
const maxMessageLen = 64 * 1024
k, consumer := newTestKafka()
k.MaxMessageLen = maxMessageLen
acc := testutil.Accumulator{}
ctx := context.Background()
overlongMsg := strings.Repeat("v", maxMessageLen+1)
go k.receiver(ctx, &acc)
consumer.Inject(saramaMsg(overlongMsg))
acc.WaitError(1)
assert.Equal(t, acc.NFields(), 0)
}
// Test that the parser parses kafka messages into points
func TestRunParserAndGather(t *testing.T) {
k, consumer := newTestKafka()
acc := testutil.Accumulator{}
ctx := context.Background()
k.parser, _ = parsers.NewInfluxParser()
go k.receiver(ctx, &acc)
consumer.Inject(saramaMsg(testMsg))
acc.Wait(1)
acc.GatherError(k.Gather)
assert.Equal(t, acc.NFields(), 1)
acc.AssertContainsFields(t, "cpu_load_short",
map[string]interface{}{"value": float64(23422)})
}
// Test that the parser parses kafka messages into points
func TestRunParserAndGatherGraphite(t *testing.T) {
k, consumer := newTestKafka()
acc := testutil.Accumulator{}
ctx := context.Background()
k.parser, _ = parsers.NewGraphiteParser("_", []string{}, nil)
go k.receiver(ctx, &acc)
consumer.Inject(saramaMsg(testMsgGraphite))
acc.Wait(1)
acc.GatherError(k.Gather)
assert.Equal(t, acc.NFields(), 1)
acc.AssertContainsFields(t, "cpu_load_short_graphite",
map[string]interface{}{"value": float64(23422)})
}
// Test that the parser parses kafka messages into points
func TestRunParserAndGatherJSON(t *testing.T) {
k, consumer := newTestKafka()
acc := testutil.Accumulator{}
ctx := context.Background()
k.parser, _ = parsers.NewParser(&parsers.Config{
DataFormat: "json",
MetricName: "kafka_json_test",
}) })
go k.receiver(ctx, &acc) }
consumer.Inject(saramaMsg(testMsgJSON)) }
func TestStartStop(t *testing.T) {
cg := &FakeConsumerGroup{errors: make(chan error)}
plugin := &KafkaConsumer{
ConsumerCreator: &FakeCreator{ConsumerGroup: cg},
}
err := plugin.Init()
require.NoError(t, err)
var acc testutil.Accumulator
err = plugin.Start(&acc)
require.NoError(t, err)
plugin.Stop()
}
type FakeConsumerGroupSession struct {
ctx context.Context
}
func (s *FakeConsumerGroupSession) Claims() map[string][]int32 {
panic("not implemented")
}
func (s *FakeConsumerGroupSession) MemberID() string {
panic("not implemented")
}
func (s *FakeConsumerGroupSession) GenerationID() int32 {
panic("not implemented")
}
func (s *FakeConsumerGroupSession) MarkOffset(topic string, partition int32, offset int64, metadata string) {
panic("not implemented")
}
func (s *FakeConsumerGroupSession) ResetOffset(topic string, partition int32, offset int64, metadata string) {
panic("not implemented")
}
func (s *FakeConsumerGroupSession) MarkMessage(msg *sarama.ConsumerMessage, metadata string) {
}
func (s *FakeConsumerGroupSession) Context() context.Context {
return s.ctx
}
type FakeConsumerGroupClaim struct {
messages chan *sarama.ConsumerMessage
}
func (c *FakeConsumerGroupClaim) Topic() string {
panic("not implemented")
}
func (c *FakeConsumerGroupClaim) Partition() int32 {
panic("not implemented")
}
func (c *FakeConsumerGroupClaim) InitialOffset() int64 {
panic("not implemented")
}
func (c *FakeConsumerGroupClaim) HighWaterMarkOffset() int64 {
panic("not implemented")
}
func (c *FakeConsumerGroupClaim) Messages() <-chan *sarama.ConsumerMessage {
return c.messages
}
func TestConsumerGroupHandler_Lifecycle(t *testing.T) {
acc := &testutil.Accumulator{}
parser := &value.ValueParser{MetricName: "cpu", DataType: "int"}
cg := NewConsumerGroupHandler(acc, 1, parser)
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
session := &FakeConsumerGroupSession{
ctx: ctx,
}
var claim FakeConsumerGroupClaim
var err error
err = cg.Setup(session)
require.NoError(t, err)
cancel()
err = cg.ConsumeClaim(session, &claim)
require.NoError(t, err)
err = cg.Cleanup(session)
require.NoError(t, err)
}
func TestConsumerGroupHandler_ConsumeClaim(t *testing.T) {
acc := &testutil.Accumulator{}
parser := &value.ValueParser{MetricName: "cpu", DataType: "int"}
cg := NewConsumerGroupHandler(acc, 1, parser)
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
session := &FakeConsumerGroupSession{ctx: ctx}
claim := &FakeConsumerGroupClaim{
messages: make(chan *sarama.ConsumerMessage, 1),
}
err := cg.Setup(session)
require.NoError(t, err)
claim.messages <- &sarama.ConsumerMessage{
Topic: "telegraf",
Value: []byte("42"),
}
go func() {
err = cg.ConsumeClaim(session, claim)
require.NoError(t, err)
}()
acc.Wait(1) acc.Wait(1)
cancel()
acc.GatherError(k.Gather) err = cg.Cleanup(session)
require.NoError(t, err)
assert.Equal(t, acc.NFields(), 2) expected := []telegraf.Metric{
acc.AssertContainsFields(t, "kafka_json_test", testutil.MustMetric(
"cpu",
map[string]string{},
map[string]interface{}{ map[string]interface{}{
"a": float64(5), "value": 42,
"b_c": float64(6), },
time.Now(),
),
}
testutil.RequireMetricsEqual(t, expected, acc.GetTelegrafMetrics(), testutil.IgnoreTime())
}
func TestConsumerGroupHandler_Handle(t *testing.T) {
tests := []struct {
name string
maxMessageLen int
topicTag string
msg *sarama.ConsumerMessage
expected []telegraf.Metric
}{
{
name: "happy path",
msg: &sarama.ConsumerMessage{
Topic: "telegraf",
Value: []byte("42"),
},
expected: []telegraf.Metric{
testutil.MustMetric(
"cpu",
map[string]string{},
map[string]interface{}{
"value": 42,
},
time.Now(),
),
},
},
{
name: "message to long",
maxMessageLen: 4,
msg: &sarama.ConsumerMessage{
Topic: "telegraf",
Value: []byte("12345"),
},
expected: []telegraf.Metric{},
},
{
name: "parse error",
msg: &sarama.ConsumerMessage{
Topic: "telegraf",
Value: []byte("not an integer"),
},
expected: []telegraf.Metric{},
},
{
name: "add topic tag",
topicTag: "topic",
msg: &sarama.ConsumerMessage{
Topic: "telegraf",
Value: []byte("42"),
},
expected: []telegraf.Metric{
testutil.MustMetric(
"cpu",
map[string]string{
"topic": "telegraf",
},
map[string]interface{}{
"value": 42,
},
time.Now(),
),
},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
acc := &testutil.Accumulator{}
parser := &value.ValueParser{MetricName: "cpu", DataType: "int"}
cg := NewConsumerGroupHandler(acc, 1, parser)
cg.MaxMessageLen = tt.maxMessageLen
cg.TopicTag = tt.topicTag
ctx := context.Background()
session := &FakeConsumerGroupSession{ctx: ctx}
cg.Reserve(ctx)
cg.Handle(session, tt.msg)
testutil.RequireMetricsEqual(t, tt.expected, acc.GetTelegrafMetrics(), testutil.IgnoreTime())
}) })
}
func saramaMsg(val string) *sarama.ConsumerMessage {
return &sarama.ConsumerMessage{
Key: nil,
Value: []byte(val),
Offset: 0,
Partition: 0,
}
}
func saramaMsgWithTopic(val string, topic string) *sarama.ConsumerMessage {
return &sarama.ConsumerMessage{
Key: nil,
Value: []byte(val),
Offset: 0,
Partition: 0,
Topic: topic,
} }
} }