Skip to content

8.15.asciidoc

Latest commit

 

History

History
111 lines (79 loc) · 5.37 KB

8.15.asciidoc

File metadata and controls

111 lines (79 loc) · 5.37 KB
Raw

APM version 8.15

APM version 8.15.5

No significant changes.

APM version 8.15.4

Bug fixes

  • Rollover for all APM data streams is now executing correctly in Elasticsearch 8.15.4, fixing a known issue present in 8.15.0, 8.15.1, 8.15.2 and 8.15.3. The fix is applied through elasticsearch#116219.

APM version 8.15.3

Bug fixes

  • Fix a panic in OTLP label setting when receiving on non-compliant attribute array values {pull}13950[13950]

Added

  • Map OpenTelemetry instrumentation scope to Service.Framework.* for all signal types. {pull}13903[13903]

APM version 8.15.2

Bug fixes

  • Fix broken APM Agents configuration cache when there are more than 100 agent config entries {pull}13958[13958]

APM version 8.15.1

Bug fixes

  • Fix fallback memory limit check conversion from Bytes to Gigabytes {pull}13838[13838]

  • Fix Elasticsearch apm-data plugin to fallback to ILM policy for data streams using default APM integration and not explicitly migrated to data stream lifecycle. More details are available in the APM known issues page {pull}13918[13918]

  • Remove from logs the error reason returned by Elasticsearch for x_content_parse_exception error {pull}13853[13853]

  • Fix ingestion failure when ip resource attribute has value 0.0.0.0 in OpenTelemetry events from Jaeger {pull}13852[13852]

  • Fix possible nil pointer dereference when parsing stacktraces in OpenTelemetry events {pull}13852[13852]

  • Fix missing tracking in instrumentation metrics for certain request-level status codes from Elasticsearch bulk responses {pull}13853[13853]

Added

  • Support OpenTelemetry Java agent inferred span links {pull}13852[13852]

APM version 8.15.0

Bug fixes

  • Avoid data race due to reuse of bytes.Buffer in ES bulk requests {pull}13155[13155]

  • APM Server now relies on the Elasticsearch apm-data plugin’s index templates, which reverts some unsafe uses of flattened field types {pull}12066[12066]

  • Add error.id to jaeger errors {pull}13196[13196]

  • Process host.ip resource attribute for OTel {pull}13614[13614]

  • Fix a bug where indexed OpenTelemetry events and their aggregated metrics may contain incorrect label values from other events when the affected OpenTelemetry event has resource attribute values that are slices. This bug may also cause a panic with a message like "panic: runtime error: slice bounds out of range [-4:]" from apm-aggregation library. {pull}13822[13822]

  • Fix unmapped or dropped APM Server monitoring metrics in APM Server self-monitoring, Stack monitoring, Metricbeat, and Elastic Agent integration {issue}13475[13475]

Breaking Changes

  • Integration custom ingest pipelines logs-apm.integration@custom, metrics-apm.integration@custom, and traces-apm.integration@custom are no longer supported. Any existing pipelines above will need to be moved to the more fine-grained *-apm*@custom ingest pipelines {pull}12066[12066]

Added

  • Add elasticsearch.flushed.uncompressed.bytes metric {pull}13155[13155]

  • APM Server now relies on the Elasticsearch apm-data plugin’s index templates, removing the requirement to install the APM integration package {pull}12066[12066]

  • Switch the default lifecycle management to leverage DSL instead of ILM, the change is executed as a part of Elasticsearch apm-data plugin migration {pull}12066[12066]

  • Upgraded bundled APM Java agent attacher CLI to version 1.50.0 {pull}13326[13326]

  • Enable Kibana curated UIs to work with hostmetrics from OpenTelemetry’s hostmetricsreceiver {pull}13196[13196]

  • Add require data stream to bulk index requests {pull}13398[13398]

  • Support self-instrumentation when in managed mode by getting tracing configs via reloader {pull}13514[13514] {pull}13653[13653] {pull}13691[13691] {pull}13790[13790]

  • Add mapping for OpenTelemetry attribute messaging.destination.name to derive service.target correctly {pull}13472[13472]

  • APM Server now automatically retries document-level 429s from Elasticsearch to avoid dropping data. output.elasticsearch.max_retries now controls both request-level and document-level retries, and defaults to 3. {pull}13620[13620]