Draft
Conversation
sky333999
force-pushed
the
otel-bump-v0.143.0
branch
from
5aac084 to
295f99c
Compare
sky333999
added
the
ready for testing
sky333999
force-pushed
the
otel-bump-v0.143.0
branch
from
7840f62 to
b58af0b
Compare
Update CloudWatch Agent dependencies to use OpenTelemetry Collector Contrib v0.143.0 from the amazon-contributing fork. Changes: - AWS fork components: Updated to v0.0.0-20260206202803-61da50e53977 (latest commit from otel-bump-v0.143.0 branch) - Upstream contrib components: Updated to v0.143.0 - Collector core stable APIs: Updated to v1.49.0 - Collector core experimental APIs: Updated to v0.143.0 - OTel SDK: Updated to v1.39.0 - Collector semconv: Updated to v0.128.0 This update includes upstream v0.143.0 changes and AWS-specific customizations from the amazon-contributing fork.
Update target_allocator.go for otel-contrib v0.143.0 breaking changes: - TargetAllocator field changed from pointer to configoptional.Optional - promslog.AllowedLevel renamed to promslog.Level - TLSSetting field renamed to TLS - targetallocator package exposed via public API in AWS OTel contrib fork
…anges
- labels.Labels: len(ls)/range ls -> ls.Len()/ls.Range() (metrics_receiver.go)
- labels.Labels: []labels.Label{} -> labels.FromStrings() (metrics_receiver_test.go)
- scrape.Target.DiscoveredLabels: now requires *labels.Builder arg (metrics_type_handler.go)
- promslog.AllowedLevel -> promslog.Level with NewLevel() constructor (start.go, metrics_receiver_test.go)
- promslog.AllowedFormat -> promslog.Format (start.go, metrics_receiver_test.go)
- TLSSetting -> TLS on proxy Config (awsxray/translator.go)
- TLSSetting -> TLS + configoptional.Optional on GRPC/HTTP server configs (otlp/translator.go)
- configauth.Authentication -> configauth.Config + configoptional.Some (prometheusremotewrite/translator.go)
- TargetAllocator: pointer -> configoptional.Optional with HasValue()/Get() (prometheus/translator.go)
- GRPC/HTTP: pointer -> configoptional.Optional with HasValue()/Get()/None() (otlp/translator.go)
- Updated all corresponding test assertions
- MetricMetadata.Metric -> MetricMetadata.MetricFamily - scrape.NewTarget: replace (labels, discoveredLabels, params) with (labels, scrapeConfig, tLabels, tgLabels) - Remove unused url.Values params from test mocks
…rID guard on TLS setup
…ending_queue fields - Set factories.Telemetry = otelconftelemetry.NewFactory() in defaultcomponents.Factories() (required by collector core v0.143.0) - Remove obsolete 'blocking' and 'enabled' fields from sending_queue in 12 golden sample config YAML files. The queue config struct was replaced by queuebatch.Config which dropped these fields; queue enablement is now controlled via configoptional.Optional presence.
otlpreceiver.Config.GRPC and .HTTP are now configoptional.Optional[T] initialized in the Default flavor by CreateDefaultConfig(). Default.Get() returns nil, so the previous code using .Get() silently skipped endpoint and TLS assignment. Replaced .Get() with .GetOrInsertDefault() which promotes Default → Some and returns a mutable pointer. Extracted TLS config into a shared variable that defaults to None (zero value) when no cert/key files are provided. Upstream PRs: - open-telemetry/opentelemetry-collector#13044 (configoptional module) - open-telemetry/opentelemetry-collector#13119 (otlpreceiver migration) - open-telemetry/opentelemetry-collector#13252 (configgrpc TLS migration)
Prometheus v0.308.1 added an internal 'loaded' boolean to promconfig.Config. GetScrapeConfigs() now returns an error if the config was not loaded through promconfig.Load() or Reload(). Our translator manually copies GlobalConfig, ScrapeConfigs, and TracingConfig from a parsed PromConfig, bypassing Load(), so the loaded flag was never set. Added cfg.PrometheusConfig.Reload() after the manual field copy. Reload() runs the config through Load() internally, setting the loaded flag and populating new default fields like ScrapeFallbackProtocol. Upstream PR: prometheus/prometheus#15637
debugexporter.Config gained new QueueConfig defaults in v0.143.0 (QueueSize=1000, NumConsumers=10, FlushTimeout=200ms, etc.). The test used a hardcoded struct literal that didn't include these fields. Instead of switching to field-level assertions (which would hide future default changes), build the expected config from CreateDefaultConfig() and only override the 2 fields the translator sets (Verbosity, SamplingThereafter). Full struct comparison is preserved.
Three default value changes in the PRW exporter golden test fixture:
- headers: {} → null (type changed to configopaque.MapList, no longer
initialized to empty map)
- force_attempt_http2: true (new field in confighttp.ClientConfig,
added in v0.132.0; matches Go http.DefaultTransport behavior)
- protobuf_message: prometheus.WriteRequest (new field controlling
Remote Write v1 vs v2; default preserves existing v1 behavior)
All three changes are cosmetic/serialization — no functional impact.
filterprocessor.Config in v0.143.0 gained unexported ottl.Factory pointer maps (resourceFunctions, dataPointFunctions, etc.) that are populated during CreateDefaultConfig(). Two independently-created default configs have identical map keys but different pointer addresses, making full struct comparison via require.Equal always fail. Compare all 6 exported config fields individually instead: ErrorMode, Metrics, Logs, Spans, Traces, Profiles.
Prometheus v0.308.1 changed several default values in GlobalConfig and ScrapeConfig (MetricNameValidationScheme, ScrapeNativeHistograms, ScrapeProtocols, etc.). The test used hardcoded struct literals that didn't include these new defaults. Instead of updating struct literals or switching to field-level assertions, load the expected config through promconfig.Load() using the same test YAML. This populates all defaults identically to the translator's Reload() path, preserving full struct comparison. Behavioral note: MetricNameValidationScheme changed from 0 (zero value string in v0.300.1, treated as UTF-8 at runtime) to 2 (explicit UTF8Validation in v0.308.1). No actual behavior change — UTF-8 was already the effective default.
Regenerated all tocwconfig golden files to match new serialized output
from OTel Collector v0.143.0 and Prometheus v0.308.1 component defaults.
Key change patterns (all cosmetic/serialization unless noted):
- exclude_service_attributes: false (new resourcetotelemetry field)
- profiles: {} (new filter/transform processor section)
- keep_alives_enabled: true (new HTTP server config field)
- cors/enforcement_policy/server_parameters: {} → null (serialization)
- tls: null (TLS field now explicitly serialized when absent)
- fail_on_missing_metadata: false (new k8s metadata field)
- force_attempt_http2: true (new confighttp.ClientConfig field)
- cloud.availability_zone: enabled: false (new resource attribute)
- New resource attributes for akamai/digitalocean cloud providers
⚠️ BEHAVIORAL CHANGE: max_staleness: 0s → 1h0m0s in cumulative-to-delta
processor. Metrics not seen for 1 hour will have state evicted. See
OTEL_BUMP_v0.143.0_SUMMARY.md for full impact analysis.
…figs Prometheus v0.308.1 changed model.ValidationScheme from a string enum (zero value fell through to UTF-8) to an int enum (zero value 0 = UnsetValidation panics in IsValidLabelName). The 3 relabel.Config structs created programmatically in relabelScrapeConfigs() and metricNameRelabelConfigs bypass Prometheus's YAML unmarshaling + ScrapeConfig.Validate() pipeline, leaving NameValidationScheme at the zero value. This causes a panic in the scrape pool goroutine when PopulateLabels calls relabel.ProcessBuilder on the custom configs. Set NameValidationScheme: model.UTF8Validation on all 3 configs.
… metric names The OTel v0.143.0 bump includes upstream commit 430ea5940 which graduated the pkg.translator.prometheus.NormalizeName feature gate to Stable and removed it from the code path. Previously, the CWA fork had reverted this gate to Alpha (PR amazon-contributing/opentelemetry-collector-contrib#29), which meant add_metric_suffixes: true (the factory default) was silently ignored. With the gate removed, the setting now takes full effect, causing type/unit suffixes to be appended to metric names exported to AMP (e.g. CPU_USAGE_IDLE -> CPU_USAGE_IDLE_unit, my_gauge -> my_gauge_ratio). Fix: explicitly set AddMetricSuffixes = false in the CWA PRW translator to preserve backward-compatible metric naming. This is the upstream- recommended migration path per opentelemetry-collector-contrib#26488. Updated translator test expectation and 6 sample config golden files.
sky333999
force-pushed
the
otel-bump-v0.143.0
branch
from
62ecf8e to
fa69d48
Compare
Contributor
|
This PR was marked stale due to lack of activity. |
The OTel contrib v0.143.0 bump rewrote the EKS detector in the resourcedetectionprocessor with two detection paths: detectFromIMDS (standard managed nodes) and detectFromAPI (fallback when IMDS is unavailable, e.g. EKS Auto, Fargate). The detectFromAPI path requires the Kubernetes node name to query the node's providerID. It reads this from Config.NodeFromEnvVar, which was previously unset, causing the API-only path to fail with 'can't get K8s Instance Metadata; node name is empty'. Set node_from_env_var: HOST_NAME in the embedded resource detection config. HOST_NAME is already set in all CWAgent K8s deployments via the Kubernetes downward API (spec.nodeName). On non-K8s environments, the env var is empty and the EKS detector gracefully degrades. Updated translator test and 6 golden YAML sample configs.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description of the issue
Describe the problem or feature in addition to a link to the issues.
Description of changes
How does this change address the problem?
License
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.
Tests
Describe what tests you have done.
Requirements
Before commiting your code, please do the following steps.
make fmtandmake fmt-shmake lintIntegration Tests
To run integration tests against this PR, add the
ready for testinglabel.