Release version 0.14.0

v0.14.0 Beta

🚀 New components 🚀

• otlphttp exporter which implements OTLP over HTTP protocol.

🛑 Breaking changes 🛑

• Rename consumer.TraceConsumer to consumer.TracesConsumer #1974
• Rename component.TraceReceiver to component.TracesReceiver #1975
• Rename component.TraceProcessor to component.TracesProcessor #1976
• Rename component.TraceExporter to component.TracesExporter #1975
• Deprecate NopExporter, add NopConsumer (#1972)
• Deprecate SinkExporter, add SinkConsumer (#1973)
• Move tailsampling processor to contrib (#2012)
• Remove NewAttributeValueSlice (#2028) and mark NewAttributeValue as deprecated (#2022)
• Remove pdata.StringValue (#2021)
• Remove pdata.InitFromAttributeMap, use CopyTo if needed (#2042)
• Remove SetMapVal and SetArrayVal for pdata.AttributeValue (#2039)

💡 Enhancements 💡

• zipkin exporter: Add queue retry to zipkin (#1971)
• prometheus exporter: Add send_timestamps option (#1951)
• filter processor: Add expr pdata.Metric filtering support (#1940, #1996)
• attribute processor: Add log support (#1934)
• logging exporter: Add index for histogram buckets count (#2009)
• otlphttp exporter: Add correct handling of server error responses (#2016)
• prometheusremotewrite exporter:
  • Add user agent header to outgoing http request (#2000)
  • Convert histograms to cumulative (#2049)
  • Return permanent errors (#2053)
  • Add external labels (#2044)
• hostmetrics receiver: Use scraper controller (#1949)
• Change Span/Trace ID to be byte array (#2001)
• Add simple metrics helper to facilitate building pdata.Metrics in receivers (#1540)
• Improve diagnostic logging for exporters (#2020)
• Add obsreport to receiverhelper scrapers (#1961)
• Update OTLP to 0.6.0 and use the new Span Status code (#2031)
• Add support of partial requests for logs and metrics to the exporterhelper (#2059)

🧰 Bug fixes 🧰

• logging exporter: Added array serialization (#1994)
• zipkin receiver: Allow receiver to parse string tags (#1893)
• batch processor: Fix shutdown race (#1967)
• Guard for nil data points (#2055)

Release version 0.13.0

v0.13.0 Beta

🛑 Breaking changes 🛑

• Host metric system.disk.time renamed to system.disk.operation_time (#1887)
• Use consumer for sender interface, remove unnecessary receiver address from Runner (#1941)
• Enable sending queue by default in all exporters configured to use it (#1924)
• Removed groupbytraceprocessor (#1891)
• Remove ability to configure collection interval per scraper (#1947)

💡 Enhancements 💡

• Host Metrics receiver now reports both system.disk.io_time and system.disk.operation_time (#1887)
• Match spans against the instrumentation library and resource attributes (#928)
• Add receiverhelper for creating flexible "scraper" metrics receiver (#1886, #1890, #1945, #1946)
• Migrate tailsampling processor to new OTLP-based internal data model and add Composite Sampler (#1894)
• Metadata Generator: Change Metrics fields to implement an interface with new methods (#1912)
• Add unmarshalling for pdata.Traces (#1948)
• Add debug-level message on error for jaeger exporter (#1964)

🧰 Bug fixes 🧰

• Fix bug where the service does not correctly start/stop the log exporters (#1943)
• Fix Queued Retry Unusable without Batch Processor (#1813) - (#1930)
• prometheus receiver: Log error message when process_start_time_seconds gauge is missing (#1921)
• Fix trace jaeger conversion to internal traces zero time bug (#1957)
• Fix panic in otlp traces to zipkin (#1963)
• Fix OTLP/HTTP receiver's path to be /v1/traces (#1979)

Release version 0.12.0

🚀 New components 🚀

• configauth package with the auth settings that can be used by receivers (#1807, #1808, #1809, #1810)
• perfcounters package that uses perflib for host metrics receiver (#1835, #1836, #1868, #1869, #1870)

💡 Enhancements 💡

• Remove queued_retry and enable otlp metrics receiver in default config (#1823, #1838)
• Add limit_percentage and spike_limit_percentage options to memorylimiter processor (#1622)
• hostmetrics receiver:
  • Collect additional labels from partitions in the filesystems scraper (#1858)
  • Add filters for mount point and filesystem type (#1866)
• Add cloud.provider semantic conventions (#1865)
• attribute processor: Add log support (#1783)
• Deprecate OpenCensus-based internal data structures (#1843)
• Introduce SpanID data type, not yet used in Protobuf messages ($1854, #1855)
• Enable otlp trace by default in the released docker image (#1883)
• tailsampling processor: Combine batches of spans into a single batch (#1864)
• filter processor: Update to use pdata (#1885)
• Allow MSI upgrades (#1914)

🧰 Bug fixes 🧰

• prometheus receiver: Print a more informative message about 'up' metric value (#1826)
• Use custom data type and custom JSON serialization for traceid (#1840)
• Skip creation of redundant nil resource in translation from OC if there are no combined metrics (#1803)
• tailsampling processor: Only send to next consumer once (#1735)
• Report Windows pagefile usage in bytes (#1837)
• Fix issue where Prometheus SD config cannot be parsed (#1877)

Release version 0.11.0

v0.11.0 Beta

🛑 Breaking changes 🛑

• Rename service.Start() to Run() since it's a blocking call
• Fix slice Append to accept by value the element in pdata
• Change CreateTraceProcessor and CreateMetricsProcessor to use the same parameter order as receivers/logs processor and exporters.
• Prevent accidental use of LogsToOtlp and LogsFromOtlp and the OTLP data structs (#1703)
• Remove SetType from configmodels, ensure all registered factories set the type in config (#1798)
• Move process telemetry to service/internal (#1794)

💡 Enhancements 💡

• Add map and array attribute value type support (#1656)
• Add authentication support to kafka (#1632)
• Implement InstrumentationLibrary translation to jaeger (#1645)
• Add public functions to export pdata to ExportXServicesRequest Protobuf bytes (#1741)
• Expose telemetry level in the configtelemetry (#1796)
• Add configauth package (#1807)
• Add config to docker image (#1792)

🧰 Bug fixes 🧰

• Use zap int argument for int values instead of conversion (#1779)
• Add support for gzip encoded payload in OTLP/HTTP receiver (#1581)
• Return proto status for OTLP receiver when failed (#1788)

Release version 0.10.0

v0.10.0 Beta

🛑 Breaking changes 🛑

• Update OTLP to v0.5.0, incompatible metrics protocol.
• Remove support for propagating summary metrics in OtelCollector.
  • This is a temporary change, and will affect mostly OpenCensus users who use metrics.

💡 Enhancements 💡

• Support zipkin proto in kafka receiver (#1646)
• Prometheus Remote Write Exporter supporting Cortex (#1577, #1643)
• Add deployment environment semantic convention (#1722)
• Add logs support to batch and resource processors (#1723, #1729)

🧰 Bug fixes 🧰

• Identify config error when expected map is other value type (#1641)
• Fix Kafka receiver closing ready channel multiple times (#1696)
• Fix a panic issue while processing Zipkin spans with an empty service name (#1742)
• Zipkin Receiver: Always set the endtime (#1750)

Release version 0.9.0

v0.9.0 Beta

🛑 Breaking changes 🛑

• Remove old base factories:
  • ReceiverFactoryBase (#1583)
  • ProcessorFactoryBase (#1596)
  • ExporterFactoryBase (#1630)
• Remove logs factories and merge with normal factories (#1569)
• Remove reconnection_delay from OpenCensus exporter (#1516)
• Remove ConsumerOld interfaces (#1631)

🚀 New components 🚀

• prometheusremotewrite exporter (WIP): Send metrics data in Prometheus TimeSeries format to Cortex or any Prometheus (#1544)
• kafka receiver: Receive traces from Kafka (#1410)

💡 Enhancements 💡

• kafka exporter: Enable queueing, retry, timeout (#1455)
• Add Headers field in HTTPClientSettings (#1552)
• Change OpenCensus receiver (#1556) and exporter (#1571) to the new interfaces
• Add semantic attribute for telemetry.auto.version (#1578)
• Add uptime and RSS memory self-observability metrics (#1549)
• Support conversion for OpenCensus SameProcessAsParentSpan (#1629)
• Access application version in components (#1559)
• Make Kafka payload encoding configurable (#1584)

🧰 Bug fixes 🧰

• Stop further processing if filterprocessor filters all data (#1500)
• processscraper: Use same scrape time for all data points coming from same process (#1539)
• Ensure that time conversion for 0 returns nil timestamps or Time where IsZero returns true (#1550)
• Fix multiple exporters panic (#1563)
• Allow attribute processor for external use (#1574)
• Do not duplicate filesystem metrics for devices with many mount points (#1617)

Release version 0.8.0

v0.8.0 Beta

🚀 New components 🚀

• groupbytrace processor that waits for a trace to be completed (#1362)

💡 Enhancements 💡

• Migrate zipkin receiver/exporter to the new interfaces (#1484)
• Migrate prometheus receiver/exporter to the new interfaces (#1477, #1515)
• Add new FactoryUnmarshaler support to all components, deprecate old way (#1468)
• Update fileexporter to write data in OTLP (#1488)
• Add extension factory helper (#1485)
• Host scrapers: Use same scrape time for all data points coming from same source (#1473)
• Make logs SeverityNumber publicly available (#1496)
• Add recently included conventions for k8s and container resources (#1519)
• Add new config StartTimeMetricRegex to prometheus receiver (#1511)
• Convert Zipkin receiver and exporter to use OTLP (#1446)

🧰 Bug fixes 🧰

• Infer OpenCensus resource type based on OpenTelemetry's semantic conventions (#1462)
• Fix log adapter in prometheus receiver (#1493)
• Avoid frequent errors for process telemetry on Windows (#1487)

Release version 0.7.0

v0.7.0 Beta

🚀 New components 🚀

• Receivers
  • fluentfoward runs a TCP server that accepts events via the Fluent Forward protocol (#1173)
• Exporters
  • kafka exports traces to Kafka (#1439)
• Extensions
  • Experimental fluenetbit facilitates running a FluentBit subprocess of the collector (#1381)

💡 Enhancements

• Updated golang/protobuf from v1.3.5 to v1.4.2 (#1308)
• Updated opencensus-proto from v0.2.1 to v0.3.0 (#1308)
• Added round_robin balancer_name as an option to gRPC client settings (#1353)
• hostmetrics receiver
  • Switch to using perf counters to get disk io metrics on Windows (#1340)
  • Add device filter for file system (#1379) and disk (#1378) scrapers
  • Record process physical & virtual memory stats separately (#1403)
  • Scrape system.disk.time on Windows (#1408)
  • Add disk.pending_operations metric (#1428)
  • Add network interface label to network metrics (#1377)
• Add exporterhelper (#1351) and processorhelper (#1359) factories
• Update OTLP to latest version (#1384)
• Disable timeout, retry on failure and sending queue for logging exporter (#1400)
• Add support for retry and sending queue for jaeger exporter (#1401)
• Add batch size bytes metric to batch processor (#1270)
• otlp receiver: Add Log Support (#1444)
• Allow to configure read/write buffer sizes for http Client (#1447)
• Update DB conventions to latest and add exception conventions (#1452)

🧰 Bug fixes 🧰

• Fix resource processor for old metrics (#1412)
• jaeger receiver: Do not try to stop if failed to start. Collector service will do that (#1434)

OpenTelemetry Collector v0.6.1 (Beta)

Docker: No changes compare to v0.6.0
GoLang: Fixes #1418

OpenTelemetry Collector v0.6.0 (Beta)

🎉 OpenTelemetry Collector v0.6.0 (Beta) 🎉

🛑 Breaking changes 🛑

• Renamed the metrics generated by hostmetrics receiver to match the (currently still pending) OpenTelemetry system metric conventions (#1261) (#1269)
• Removed vmmetrics receiver (#1282)
• Removed cpu scraper report_per_cpu config option (#1326)

💡 Enhancements 💡

• Added disk merged (#1267) and process count (#1268) metrics to hostmetrics
• Log metric data points in logging exporter (#1258)
• Changed the batch processor to not ignore the errors returned by the exporters (#1259)
• Build and publish MSI (#1153) and DEB/RPM packages (#1278, #1335)
• Added batch size metric to batch processor (#1241)
• Added log support for memorylimiter processor (#1291) and logging exporter (#1298)
• Always add tags for observability, other metrics may use them (#1312)
• Added metrics support (#1313) and allow partial retries in queued_retry processor (#1297)
• Update resource processor: introduce attributes config parameter to specify actions on attributes similar to attributes processor, old config interface is deprecated (#1315)
• Update memory state labels for non-Linux OSs (#1325)
• Ensure tcp connection value is provided for all states, even when count is 0 (#1329)
• Set batch processor channel size to num cpus (#1330)
• Add send_batch_max_size config parameter to batch processor enforcing hard limit on batch size (#1310)
• Add support for including a per-RPC authentication to gRPC settings (#1250)

🧰 Bug fixes 🧰

• Fixed OTLP waitForReady, not set from config (#1254)
• Fixed all translation diffs between OTLP and Jaeger (#1222)
• Disabled process scraper for any non Linux/Windows OS (#1328)

📦 Release bits 📦

• Docker: docker pull otel/opentelemetry-collector:0.6.0
• Binary checksums: Generated with shasum -a 256 * and are present in the checksums.txt file