Please refer to the 7.82.1 tag on integrations-core for the list of changes on the Core Checks
Bug Fixes
Windows: Fixed an issue where an explicit DDAGENTUSERKEEPRIGHTS or DDAGENTUSERNAME value passed as an install argument to a Fleet Automation-triggered Windows Agent install/upgrade could be silently overridden by a stale fallback value (res
Fix an issue where GPU monitoring could trigger a kernel panic on multi-GPU nodes with Hopper/Blackwell GPUs.
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Adds support for `runtime.gomemlimit` in vertical scaling recommendations. When the backend recommendation includes a `Gomemlimit` value for a container, the cluster agent now: 1. Parses `Ru
### What does this PR do? Adds a macOS implementation of the `thermal` corecheck, which until now existed only for Windows. The check is darwin-only (`//go:build darwin`) and is registered alongside the other core checks in `pkg/commonchecks/corechecks.go:154`. It collects two independent classes of signal: - **Hardware temperatures via AppleSMC.** `thermal_
Data Observability query actions now dispatch remote configuration payloads to MySQL checks as well as Postgres and SAP HANA. The component uses one supported-integration allowlist for startup detection and instance matching, so unrelated integrations cannot activate the subscription or receive query configs. Closes https://linear.app/datadog/issue/DOIO-171
### What does this PR do? Gates the NVML workloadmeta collector on `gpu.enabled`. ### Motivation Avoid NVML collection when GPU monitoring is disabled. ### Describe how you validated your changes Added unit coverage for disabled startup. ### Additional Notes
### What does this PR do? Tightens how system-probe's runtime-compiled eBPF object cache is created and consumed: - Adds `secureRuntimeDir`, which creates the runtime-compiler output directory root-only (`0700`) and verifies that it — and every ancestor up to the filesystem root — is a real directory owned by root and not writable by other users. The sticky
### What does this PR do? Changes how the auto multi-line timestamp detector turns a token-transition match into a probability. `MatchProbability` marks every token transition in a line as `1` (the graph has that edge) or `-1` (it does not), runs Kadane's algorithm to find the highest-*sum* run, and then reports the **average over that run** as the probabili
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? In errortracking log handler, pass the stacktrace pcs via context value rather than log record attributes. ### Motivation Log record attributes are meant to be readable and rendered, while c
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? This PR fully eliminates usage of `visual_studio.bzl` that was just pinning version of a local MSVC installation. #54718 introduced a fully hermetic MSVC toolchain, as well as MSBuild tool a
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Create a typed file provisioner. To use an environment already provisioned inside a test ### Motivation We want to separate provisioning from test execution, that part gives us a first step
> **Note:** This PR was created by Claude. ### What does this PR do? Adds the e2e test suite for Agent Workload Balancing that was called out as needed in #54652's original plan, and adds the one missing CLI surface found while auditing that plan for completeness. - `TestWorkloadBalancingRunningMetrics` / `TestWorkloadBalancingAddedToRCListeners` (`workloadb
### What does this PR do? Runs the USM and protocol-classification test suites against the **fentry** connection tracer, which they previously never exercised. Fentry was absent from `usmtestutil.SupportedBuildModes`, and four separate gates short-circuited on it even when it was selected: | Gate | Old behavior on fentry | |---|---| | `httpSupported()` / `ht
### What does this PR do? Two independent allocation reductions on the trace-agent receive/decode path. **1. Intern strings directly from the payload (v0.4 decode).** `parseStringBytesRef` materialized a Go string for every value just to look it up in the payload's string table, discarding it on a hit. **2. Reserve `bytes.MinRead` beyond the body when buffer
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Ensures that calls to `GetFieldValues` are serialized as that is not a thread safe API. ### Motivation #incident-59170 - `nvlink_fields` collector might mark some ports as unsupported. This
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Adds `gpu_nvlink_capable` and `gpu_nvlink_version` tags to GPU metrics. ### Motivation Knowing whether a GPU is NVlink-capable and the version of the NVlink system or not is useful to ensure
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? ### Motivation ### Describe how you validated your changes ### Additional Notes
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Connects configurable workload target resolution to DatadogInstrumentation checks and logs. - Adds `instrumentation_crd_controller.custom_workload_targets` configuration. - Starts the Pod wo
This issue lists Renovate updates and detected dependencies. Read the [Dependency Dashboard](https://docs.renovatebot.com/key-concepts/dashboard/) docs to learn more.<br>[View this repository on the Mend.io Web Portal](https://developer.mend.io/github/DataDog/datadog-agent). ## Deprecations / Replacements > [!WARNING] The following dependencies are either de
`service.namespace` is a stable OpenTelemetry semantic convention, and the natural way to distinguish multiple deployments of the same service that share a `service.name`. It doesn't reach Datadog metrics as a tag. In `pkg/opentelemetry-mapping-go/otlp/attributes/attributes.go`, `TagsFromAttributes` promotes resource attributes through `coreMapping`, `kubern
### What does this PR do? For debian packaging, we need to include a file of md5 sums of all the files in the data payload as the file md5sums in the control payload. This makes that available as a rule. I'll plug it into a wrapper for pkg_deb in a follow-up. ### Describe how you validated your changes - Tests - reading the generated code. It's pretty simple
### What does this PR do? Make all extension-related converter features behave like `datadog` and `dogtel`: in the case where the relevant extension is defiled by the user but not added to `service.pipeline`, add this instance instead of defining a `/dd-autoconfigured` one ### Motivation Unify the user experience [OTAGENT-1127](https://datadoghq.atlassian.ne
### What does this PR do? Adds the shared Go infrastructure that AWS MicroVM cloud-service support depends on. This is the first in a stack of PRs that together deliver full MicroVM observability. ### Motivation The Datadog Agent's serverless-init supports multiple cloud platforms (Cloud Run, Container Apps, etc.) via a `CloudService` abstraction. AWS Lambda
## Summary This adds the MicroVM lifecycle HTTP server itself: an `http.Server` listening on the configured port, handling the six lifecycle hooks the AWS MicroVM platform sends — `/ready`, `/validate`, `/run`, `/resume`, `/suspend`, and `/terminate` — in **standalone** mode, i.e. the agent answers each hook itself rather than forwarding it to a user applica
## Summary - Adds a local Docker-based Kind provisioner (`testing/provisioners/local/kubernetes`) as a cheaper alternative to the existing AWS-VM-backed Kind provisioner — no dedicated EC2 VM needed for most Kind-based E2E tests. - Brings the local provisioner to feature parity with the AWS one: fakeintake options (memory/retention/dddev-forwarding), dogstat
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Introduces a shared pool of macOS EC2 Dedicated Hosts for the E2E test framework, so macOS test runs stop provisioning (and tearing down) a brand-new Dedicated Host + instance every time. -
### What does this PR do? Adds a default-disabled prepared-startup primitive for the core Agent and direct trace Agent. Each process constructs its Fx graph, identifies its own Pod and DaemonSet through the local kubelet Pod list, and waits without starting Fx hooks when an older sibling Pod exists. It publishes `prepared` so native DaemonSet surge can consi
…s-proxy and upload it too <!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? ### Motivation ### Describe how you validated your changes ### Additional Notes
## Summary Points at the `linux_test_only` image and crosstool-ng toolchains published by the `chouquette/toolchains_mass` branch in `datadog-agent-buildimages` (new toolchain build+publish CI stage), to confirm the agent still builds with them before that branch merges. - `.gitlab-ci.yml`: `CI_IMAGE_LINUX` → test-only image from that branch's pipeline - `ba
### What does this PR do? The Cluster Agent's metrics server (`0.0.0.0:metrics_port`, default `5000`) served the entire `http.DefaultServeMux`, exposing the `pprof` and `expvar` debug endpoints (registered via blank imports in `cmd/cluster-agent/main.go`) unauthenticated to anything that can reach the pod. This routes only `/metrics` on the public mux and ga
### What does this PR do? Bumps `dd-sensitive-data-scanner` to `0.1.0-20260807-dcc653f6f465` and, in the `datasecurity` rust check, drops the scanner and clears dd-sds' global regex caches once scanning finishes. ### Motivation Keep the check's idle memory footprint near zero between runs by releasing the scanner and its cached compiled regexes as soon as th
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Adds a new `MACOS_CORECHECKS` list to `tasks/core_checks.py` (and its byte-identical mirror `cmd/agent/dist/core_checks.bzl`), moves `battery` and `wlan` out of `AGENT_CORECHECKS` into it, a
### What does this PR do? Adds the process-lifecycle boundary for `par-control`: - Uses the shared `dd-procmgr-client` connector for local IPC on Linux and Windows. - Loads the minimal configuration needed to gate split mode. - Exposes one-shot executor lifecycle operations to start or adopt the executor, inspect liveness and terminal state, and stop it. - I
### What does this PR do? Adds the effective configuration and identity bootstrap layer for `par-control`: - Adds a Go helper subcommand that exports the Agent's resolved configuration. - Loads split-runner configuration from Rust. - Resolves local and Fleet-managed settings consistently. - Loads and persists runner identity. - Supports self-enrollment boots
### What does this PR do? Defines the local control-to-executor channel and implements both ends of it: - Adds the executor gRPC protocol and generated bindings. - Exposes executor health and readiness. - Synchronizes workflow signing keys. - Streams action dispatch outcomes. - Adds the Rust mTLS client and transport support. - Keeps the shared terminal `Out
### What does this PR do? Composes the process manager, effective configuration, OPMS client, and executor channel into the production `par-control` loop: - Adds bounded task concurrency and retry policy. - Leaves the executor stopped while idle and starts it only after a task is dequeued. - Starts task heartbeats at dequeue and continues them through execut
### What does this PR do? Teaches the existing Go runner and Fx component to stand down when split mode is enabled on supported Linux and Windows host deployments, while retaining the executor subcommand used by `par-control`. Container deployments do not yet launch the replacement topology. Official Agent containers are detected through `configenv.IsContain
### What does this PR do? Sets `SkipSettingGRPCLogger: true` on the `otelcol.CollectorSettings` for the otel-agent and host-profiler collectors, so `otelcol.(*Collector).Run` no longer calls the non-mutex-protected `grpclog.SetLogger` while other gRPC clients (e.g. remote-config/remote-tagger) are active in the same process. ### Motivation Fix a race, found
> **Note:** This PR was created by Claude. Stacked on #54652 — review that one first. ### What does this PR do? Emits `datadog.agent.workload_balancing.running` once per NDM workload balancing group the Agent knows about, tagged with: - `workload_balancing_group:<id>` - `workload_balancing_state:active|standby|unmanaged` It follows the shape of the existing
### What does this PR do? Upgrades the Agent's embedded CPython interpreter from **3.13.14** to **3.13.15** (patch version update). - Updated `omnibus/config/software/python3.rb` with the new version - Updated `deps/cpython/cpython.MODULE.bazel` with the new version and SHA256 - Updated `test/new-e2e/tests/agent-platform/common/agent_behaviour.go` with the e
### What does this PR do? Modifies the auto multi-line timestamp detector's scoring so that a line opening with a long enough run of matching tokens is scored on that run alone. Adds regression cases for both sides of the default threshold. ### Motivation Consecutive IIS W3C access-log records were being concatenated into a single log. W3C extended records a
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Bump google.golang.org/grpc to v1.82.1 ### Motivation ### Describe how you validated your changes ### Additional Notes
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Remove unused `SyncCapture` code in log package. ### Motivation This is currently dead code. Discussed with the team which introduced this and it has been refactored so this is not needed an
### What does this PR do? Add telemetry to dogstatsd http server ### Motivation Track usage and performance of the new component ### Describe how you validated your changes Unit tests. Run agent with dogstatsd http enabled and metric filter configured. Check agent telemetry to include expected metrics. ### Additional Notes
## What does this PR do? Adds `//go:embed` directives for `tmpl/gen/oci-nocap/*.service` and `tmpl/gen/debrpm-nocap/*.service`, and the matching Bazel `embedsrcs` entries, so the `-nocap` systemd unit templates ship inside the installer binary. Adds `embed_test.go`, which enumerates units through the `embed.FS` and asserts `GetSystemdUnit` resolves every uni
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Remove bazel strptime_cgo_testlib override. ### Motivation Fixed by https://github.com/open-telemetry/opentelemetry-collector-contrib/pull/49516. ### Describe how you validated your changes
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? Due to a framework issue, the fakeintakes might still be in use (ie. an agent is still sending payloads to them) when the test starts, which makes the test flaky. We added a detection logic
### What does this PR do? Moves the `dd-sds` (`dd-sensitive-data-scanner`) dependency from the `datasecurity` check crate into the shared `[workspace.dependencies]` in the root `Cargo.toml`. The crate now references it via `dd_sds.workspace = true`. ### Motivation Centralize the pinned version and feature set so future check crates share a single source of t
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !--> ### What does this PR do? ### Motivation ### Describe how you validated your changes ### Additional Notes