Refactoring streams and storage

.mvn/wrapper/MavenWrapperDownloader.java

@@ -17,9 +17,6 @@ Licensed to the Apache Software Foundation (ASF) under one
 under the License.
 */
 
-import java.net.*;
-import java.io.*;
-import java.nio.channels.*;
 import java.util.Properties;
 
 public class MavenWrapperDownloader {

DESIGN.md

@@ -2,95 +2,129 @@
 
 ## Goals
 
-* Provide a fast and reliable storage that enable extensibility via Kafka Topics.
+* Provide a fast and reliable storage that enable extensibility via Kafka topics.
 * Provide full storage functionality via streaming aggregations (e.g., dependency graph).
 * Create a processing space where additional enrichment can be plugged in into the processing 
 pipeline.
 * Remove need for additional storage when Kafka is available.
+* More focused on supporting processing than storage: traces and dependency links are emitted 
+downstream to support metrics aggregation. Storage is currently supported but in a single node.
 
-### Zipkin Storage Component
+## Kafka Zipkin Storage
 
-A Zipkin Storage component has the following internal parts:
+Storage is composed by 3 main components: 
 
-* `Builder`: which configures if
-    - `strictTraceId(boolean strictTraceId)`
-    - `searchEnabled(boolean searchEnabled)`
-    - `autocompleteKeys(List<String> keys)`
-    - `autocompleteTtl(int autocompleteTtl)`
-    - `autocompleteCardinality(int autocompleteCardinality)`
-* `SpanStore`: main component
-    - `Call<List<List<Span>>> getTraces(QueryRequest request);`
-    - `Call<List<Span>> getTrace(String traceId);`
-    - `Call<List<String>> getServiceNames();`
-    - `Call<List<String>> getSpanNames(String serviceName);`
-    - `Call<List<DependencyLink>> getDependencies(long endTs, long lookback);`
-* `SpanConsumer`: which ingest spans
-    - `Call<Void> accept(List<Span> spans)`
-* `QueryRequest`: which includes
-    - `String serviceName, spanName;`
-    - `Map<String, String> annotationQuery;`
-    - `Long minDuration, maxDuration;`
-    - `long endTs, lookback;`
-    - `int limit;`
+- Span Consumer: repartition of collected span batches into individual spans keyed by `traceId`
+- Span Aggregation: stream processing of spans into aggregated traces and then into dependency links.
+- Span Store: building local state stores to support search and query API.
 
-### Kafka Zipkin Storage
+And it is supported by 3 main Kafka topics:
 
-#### `KafkaSpanStore`
+- `zipkin-spans`: Topic where list of spans indexed by trace Id are stored.
+- `zipkin-trace`: Topic where aggregated traces are stored.
+- `zipkin-dependency`: Topic where dependency links per trace are stored. 
 
-Span Store is expecting Spans to be stored in topics partitioned by `TraceId`.
+### Kafka Span Consumer
 
-> These can be created by Span Consumer, or can be **enriched** by other Stream Processors, outside of
-Zipkin Server.
+This component processes collected span batches (via HTTP, Kafka, ActiveMQ, etc), 
+take each element and re-indexed them by `traceId` on the "spans" topic.
 
-Kafka Span Store will need to support different kind of queries:
+This component is currently compensating how `KafkaSender` (part of [Zipkin-Reporter](https://github.com/openzipkin/zipkin-reporter-java))
+is reporting spans to Kafka, by grouping spans into batches and sending them to a un-keyed
+Kafka topic.
 
+Component source code: [KafkaSpanConsumer.java](storage/src/main/java/zipkin2/storage/kafka/KafkaSpanConsumer.java)
 
-##### Get Service Names/Get Span Names
+### Stream Processing
 
-Service name to Span names pairs are indexed by aggregating spans.
+#### Span Aggregation
 
-##### Get Trace/Find Traces
+"Partitioned" Spans are processed to produced two aggregated streams: `Traces` and `Dependencies`.
 
-When search requests are received, span index is used to search for trace ids. After a list is 
-retrieved, trace DAG is retrieved from trace state store.
+**Traces**: 
 
-##### Get Dependencies
+Spans are grouped by ID and stored on a local
+[Session window](https://kafka.apache.org/23/javadoc/org/apache/kafka/streams/kstream/SessionWindows.html),
+where the `traceId` becomes the token, and session inactivity gap 
+(i.e. period of time without receiving a span with the same session) 
+defines if a trace is still active or not. This is evaluated on the next span received on the stream--
+regardless of incoming `traceId`. If session window is closed, a trace message is emitted to the 
+traces topic.
 
-After `spans` are aggregated into traces, traces are processed to collect dependencies. 
-Dependencies changelog are stored in a Kafka topic to be be stored as materialized view on 
-Zipkin instances.
+**Dependencies**
 
-### Stream processors
+Once `traces` are emitted downstream as part of the initial processing, dependency links are evaluated
+on each trace, and emitted the dependencies topic for further metric aggregation.
 
-#### Trace Aggregation Stream Processor
+Kafka Streams topology:
 
-This is the main processors that take incoming spans and aggregate them into:
+![trace aggregation](docs/trace-aggregation-topology.png)
 
-- Traces
-- Dependencies
+#### Trace Store Stream
 
-![service aggregation](docs/service-aggregation-stream.png)
+This component build local stores from state received on `spans` Kafka topic 
+for traces, service names and autocomplete tags. 
 
-![dependency aggregation](docs/dependency-aggregation-stream.png)
+Kafka Streams source code: [TraceStoreTopologySupplier](storage/src/main/java/zipkin2/storage/kafka/streams/TraceStoreTopologySupplier.java)
 
-#### Store Stream Processors
+Kafka Streams topology:
 
-Global tables for traces, service names and dependencies to be available on local state.
+![trace store](docs/trace-store-topology.png)
 
-![trace store](docs/trace-store-stream.png)
+#### Dependency Store
 
-![service store](docs/service-store-stream.png)
+This component build local store from state received on `dependency` Kafka topic.
 
-![dependency store](docs/dependency-store-stream.png)
+It builds a 1 minute time-window when counts calls and errors.
 
-#### Index Stream Processor
+Kafka Streams source code: [DependencyStoreTopologySupplier](storage/src/main/java/zipkin2/storage/kafka/streams/DependencyStoreTopologySupplier.java)
 
-Custom processor to full-text indexing of traces using Lucene as back-end.
+Kafka Streams topology:
 
-![span index](docs/span-index-stream.png)
+![dependency store](docs/dependency-store-topology.png)
 
-#### Retention Stream Processor
+### Kafka Span Store
 
-This is the processor that keeps track of trace timestamps for cleanup.
+This component supports search and query APIs on top of local state stores build by the Store 
+Kafka Streams component.
 
-![trace retention](docs/trace-retention-stream.png)
+Component source code: [KafkaSpanStore.java](storage/src/main/java/zipkin2/storage/kafka/KafkaSpanStore.java)
+
+#### Get Service Names/Get Span Names/Get Remote Service Names
+
+These queries are supported by service names indexed stores built from `spans` Kafka topic.
+
+Store names:
+
+- `zipkin-service-names`: key/value store with service name as key and value.
+- `zipkin-span-names`: key/value store with service name as key and span names list as value.
+- `zipkin-remote-service-names`: key/value store with service name as key and remote service names as value.
+
+#### Get Trace/Find Traces
+
+These queries are supported by two key value stores: 
+
+- `zipkin-traces`: indexed by `traceId`, contains span list status received from `spans` Kafka topic.
+- `zipkin-traces-by-timestamp`: list of trace IDs indexed by `timestamp`.
+
+`GetTrace` query is supported by `zipkin-traces` store.
+`FindTraces` query is supported by both: When receiving a query request time range is used to get
+trace IDs, and then query request is tested on each trace to build a response.
+
+#### Get Dependencies
+
+This query is supported 1-minute windowed store from `DependencyStoreStream`.
+
+When a request is received, time range is used to pick valid windows and join counters.
+
+Windowed store:
+
+- `zipkin-dependencies`.
+
+### Kafka Autocomplete Tags
+
+#### Get Keys/Get Values
+
+Supported by a key-value containing list of values valid for `autocompleteKeys`.
+
+- `zipkin-autocomplete-tags`: key-value store.

Dockerfile

@@ -12,34 +12,49 @@
 # the License.
 #
 
-FROM openjdk:8
+FROM alpine
 
-ARG KAFKA_STORAGE_VERSION=0.1.1
+ENV ZIPKIN_REPO https://repo1.maven.org/maven2
+ENV ZIPKIN_VERSION 2.16.1
+ENV KAFKA_STORAGE_VERSION 0.4.1-SNAPSHOT
 
-ENV ZIPKIN_REPO https://jcenter.bintray.com
-ENV ZIPKIN_VERSION 2.12.6
-ENV ZIPKIN_LOGGING_LEVEL INFO
+WORKDIR /zipkin
+
+RUN apk add unzip curl --no-cache && \
+    curl -SL $ZIPKIN_REPO/io/zipkin/zipkin-server/$ZIPKIN_VERSION/zipkin-server-$ZIPKIN_VERSION-exec.jar > zipkin-server.jar && \
+    # don't break when unzip finds an extra header https://github.com/openzipkin/zipkin/issues/1932
+    unzip zipkin-server.jar ; \
+    rm zipkin-server.jar
+
+COPY autoconfigure/target/zipkin-autoconfigure-storage-kafka-${KAFKA_STORAGE_VERSION}-module.jar BOOT-INF/lib/kafka-module.jar
+RUN unzip -o BOOT-INF/lib/kafka-module.jar lib/* -d BOOT-INF
+
+FROM gcr.io/distroless/java:11-debug
 
 # Use to set heap, trust store or other system properties.
 ENV JAVA_OPTS -Djava.security.egd=file:/dev/./urandom
+
+RUN ["/busybox/sh", "-c", "adduser -g '' -D zipkin"]
+
 # Add environment settings for supported storage types
+ENV STORAGE_TYPE kafka
+
+COPY --from=0 /zipkin/ /zipkin/
 WORKDIR /zipkin
 
-RUN curl -SL $ZIPKIN_REPO/io/zipkin/java/zipkin-server/$ZIPKIN_VERSION/zipkin-server-${ZIPKIN_VERSION}-exec.jar > zipkin.jar
+# TODO haven't found a better way to mount libs from custom storage. issue #28
+#COPY autoconfigure/target/zipkin-autoconfigure-storage-kafka-${KAFKA_STORAGE_VERSION}-module.jar kafka-module.jar
+#ENV MODULE_OPTS -Dloader.path='BOOT-INF/lib/kafka-module.jar,BOOT-INF/lib/kafka-module.jar!/lib' -Dspring.profiles.active=kafka
+ENV MODULE_OPTS -Dspring.profiles.active=kafka
+
+RUN ["/busybox/sh", "-c", "ln -s /busybox/* /bin"]
 
-ADD storage/target/zipkin-storage-kafka-${KAFKA_STORAGE_VERSION}.jar zipkin-storage-kafka.jar
-ADD autoconfigure/target/zipkin-autoconfigure-storage-kafka-${KAFKA_STORAGE_VERSION}-module.jar zipkin-autoconfigure-storage-kafka.jar
+ENV KAFKA_STORAGE_DIR /data
+RUN mkdir /data  && chown zipkin /data
+VOLUME /data
 
-ENV STORAGE_TYPE=kafkastore
+USER zipkin
 
-EXPOSE 9410 9411
+EXPOSE 9411
 
-CMD exec java \
-    ${JAVA_OPTS} \
-    -Dloader.path='zipkin-storage-kafka.jar,zipkin-autoconfigure-storage-kafka.jar' \
-    -Dspring.profiles.active=kafkastore \
-    -Dcom.linecorp.armeria.annotatedServiceExceptionVerbosity=all \
-    -Dcom.linecorp.armeria.verboseExceptions=true \
-    -cp zipkin.jar \
-    org.springframework.boot.loader.PropertiesLauncher \
-    --logging.level.zipkin2=${ZIPKIN_LOGGING_LEVEL}
+ENTRYPOINT ["/busybox/sh", "-c", "exec java ${MODULE_OPTS} ${JAVA_OPTS} -cp . org.springframework.boot.loader.PropertiesLauncher"]

Makefile

@@ -3,23 +3,33 @@ all: build
 
 OPEN := 'xdg-open'
 MAVEN := './mvnw'
-VERSION := '0.3.3-SNAPSHOT'
+VERSION := '0.4.1-SNAPSHOT'
+IMAGE_NAME := 'jeqo/zipkin-kafka'
 
 .PHONY: run
 run: build zipkin-local
 
 .PHONY: run-docker
 run-docker: build docker-build docker-up
 
+.PHONY: kafka-topics
+kafka-topics:
+	docker-compose exec kafka-zookeeper /busybox/sh /kafka/bin/kafka-run-class.sh kafka.admin.TopicCommand \
+		--zookeeper localhost:2181 --create --topic zipkin-spans --partitions 1 --replication-factor 1 --if-not-exists
+	docker-compose exec kafka-zookeeper /busybox/sh /kafka/bin/kafka-run-class.sh kafka.admin.TopicCommand \
+		--zookeeper localhost:2181 --create --topic zipkin-trace --partitions 1 --replication-factor 1 --if-not-exists
+	docker-compose exec kafka-zookeeper /busybox/sh /kafka/bin/kafka-run-class.sh kafka.admin.TopicCommand \
+		--zookeeper localhost:2181 --create --topic zipkin-dependency --partitions 1 --replication-factor 1 --if-not-exists
+
 .PHONY: docker-build
 docker-build:
-	TAG=${VERSION} \
-	docker-compose build
+	docker build -t ${IMAGE_NAME}:latest .
+	docker build -t ${IMAGE_NAME}:${VERSION} .
 
 .PHONY: docker-push
 docker-push: docker-build
-	TAG=${VERSION} \
-	docker-compose push
+	docker push ${IMAGE_NAME}:latest
+	docker push ${IMAGE_NAME}:${VERSION}
 
 .PHONY: docker-up
 docker-up:
@@ -33,7 +43,7 @@ docker-down:
 
 .PHONY: docker-kafka-up
 docker-kafka-up:
-	docker-compose up -d kafka zookeeper
+	docker-compose up -d kafka-zookeeper
 
 .PHONY: license-header
 license-header:
@@ -49,22 +59,35 @@ test: build
 
 .PHONY: zipkin-local
 zipkin-local:
-	STORAGE_TYPE=kafkastore \
+	STORAGE_TYPE=kafka \
+	KAFKA_BOOTSTRAP_SERVERS=localhost:19092 \
 	java \
-	-Dloader.path='storage/target/zipkin-storage-kafka-${VERSION}.jar,autoconfigure/target/zipkin-autoconfigure-storage-kafka-${VERSION}-module.jar' \
-	-Dspring.profiles.active=kafkastore \
+	-Dloader.path='autoconfigure/target/zipkin-autoconfigure-storage-kafka-${VERSION}-module.jar,autoconfigure/target/zipkin-autoconfigure-storage-kafka-${VERSION}-module.jar!/lib' \
+	-Dspring.profiles.active=kafka \
 	-cp zipkin.jar \
 	org.springframework.boot.loader.PropertiesLauncher
 
 .PHONY: get-zipkin
 get-zipkin:
 	curl -sSL https://zipkin.io/quickstart.sh | bash -s
 
+.PHONY: zipkin-test-multi
+zipkin-test-multi:
+	curl -s https://raw.githubusercontent.com/openzipkin/zipkin/master/zipkin-lens/testdata/netflix.json | \
+	curl -X POST -s localhost:9411/api/v2/spans -H'Content-Type: application/json' -d @- ; \
+	${OPEN} 'http://localhost:9412/zipkin/?lookback=custom&startTs=1'
+	sleep 61
+	curl -s https://raw.githubusercontent.com/openzipkin/zipkin/master/zipkin-lens/testdata/messaging.json | \
+	curl -X POST -s localhost:9411/api/v2/spans -H'Content-Type: application/json' -d @- ; \
+
 .PHONY: zipkin-test
 zipkin-test:
-	curl -s https://raw.githubusercontent.com/openzipkin/zipkin/master/zipkin-ui/testdata/netflix.json | \
+	curl -s https://raw.githubusercontent.com/openzipkin/zipkin/master/zipkin-lens/testdata/netflix.json | \
 	curl -X POST -s localhost:9411/api/v2/spans -H'Content-Type: application/json' -d @- ; \
 	${OPEN} 'http://localhost:9411/zipkin/?lookback=custom&startTs=1'
+	sleep 61
+	curl -s https://raw.githubusercontent.com/openzipkin/zipkin/master/zipkin-lens/testdata/messaging.json | \
+	curl -X POST -s localhost:9411/api/v2/spans -H'Content-Type: application/json' -d @- ; \
 
 .PHONY: release
 release: