No significant changes.
-
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.
-
Fix a panic in OTLP label setting when receiving on non-compliant attribute array values {pull}13950[13950]
-
Map OpenTelemetry instrumentation scope to
Service.Framework.*
for all signal types. {pull}13903[13903]
-
Fix broken APM Agents configuration cache when there are more than 100 agent config entries {pull}13958[13958]
-
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 value0.0.0.0
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]
-
Support OpenTelemetry Java agent inferred span links {pull}13852[13852]
-
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]
-
Integration custom ingest pipelines
logs-apm.integration@custom
,metrics-apm.integration@custom
, andtraces-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]
-
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 deriveservice.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 to3
. {pull}13620[13620]