fix(logs): Isolate partial container logs by stream

[soberpeach]

What does this PR do?

This PR prevents partial container log fragments from stdout and stderr from being combined into the same logical message.

It propagates normalized stream identity from the CRI and Docker JSON-file parsers, replaces the decoder partial-line parser single accumulator with per-stream state, and gives each stream independent timeout and truncation handling. Parsers without stream identity continue to use the original default accumulator.

Motivation

Fixes #54873.

CRI records contain both a stream and a partial/final flag, but the decoder previously accumulated every partial fragment in one shared buffer. A complete stdout record arriving between two stderr fragments could therefore be included in the stderr message, and the inverse ordering had the same problem. Docker JSON-file logs used the same partial-line path and had the same structural issue.

Describe how you validated your changes

• Added a regression matrix covering CRI stderr interrupted by stdout, the inverse ordering, both streams being partial simultaneously, and Docker JSON-file interleaving.
• Added focused coverage for independent stream deadlines and independent size-limit truncation.
• Ran a repeatable local E2E harness on Colima with a kind/containerd node. The raw runtime file contained the test sequence stderr P, stdout F, stderr P, stderr F. agent stream-logs emitted stdout separately and reconstructed both stderr fragments into one logical message. The harness passed again through its reusable --skip-build path. The Agent namespace used a private-network-only egress policy and a dummy API key.

Checkpointing consideration

The file tailer currently advances its decoded checkpoint by each emitted message's RawDataLen. That field therefore represents both a logical message's source-byte size and positional progress through the file.

Per-stream buffering can make logical emission order differ from source-file order. For example, when a partial stderr CRI record is followed by a complete stdout record, stdout can now be emitted while the earlier stderr record remains buffered. If the Agent terminates abruptly in that window, the checkpoint is calculated from the emitted stdout bytes without their original file position. Depending on record sizes, it can land within either buffered or already-emitted CRI data, so restart may re-read a record, start in the middle of one, or fail to replay the buffered fragment. A graceful decoder shutdown flushes the remaining stream buffers, after which the total byte count is correct.

Eliminating that crash-window risk requires decoupling source-position acknowledgement from RawDataLen (or holding completed messages until all preceding source ranges are complete). That is a broader file-tailer/decoder pipeline change than the stream-isolation fix and should include restart-at-interleaving-boundary coverage. This PR keeps RawDataLen accurate for each reconstructed logical message and does not claim to solve that separate checkpoint-ordering problem.

[github-actions]

AI review by Codex (OpenAI) - workflow run

Patch is incorrect. Per-stream buffering breaks file checkpoint accounting when streams emit out of physical file order, risking replay, malformed records, or log loss after restart.

[github-actions]

[P1] Preserve physical-order checkpoint accounting. RawDataLen advances the file tailer's checkpoint for every emitted message, but assigning only this stream's bytes means an interleaved later record can emit before bytes already buffered from another stream. If the Agent stops between those emissions, the saved offset points into an earlier CRI/Docker record. Restart can then replay, misparse, or lose logs. The decoder needs offset accounting independent from per-stream message aggregation.

[datadog-datadog-prod-us1-2]

🎯 Code Coverage (details)
• Patch Coverage: 89.53%
• Overall Coverage: 52.57% (-0.14%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: d672ace | Docs | View more details | Give us feedback!}

[dd-octo-sts]

Files inventory check summary

File checks results against ancestor e318062a:

Results for datadog-agent_7.84.0~devel.git.432.d672ace.pipeline.134061774-1_amd64.deb:

No change detected

Results for datadog-iot-agent_7.84.0~devel.git.432.d672ace.pipeline.134061774-1_amd64.deb:

No change detected

[dd-octo-sts]

Static quality checks

✅ Please find below the results from static quality gates
Comparison made with ancestor e318062
📊 Static Quality Gates Dashboard
🔗 SQG Job

Successful checks

Info

	Quality gate	Change	Size (prev → curr → max)
✅	agent_deb_amd64	+8.24 KiB (0.00% increase, -0.22% of buffer)	760.911 → 760.919 → 764.600
✅	agent_deb_amd64_fips	+8.24 KiB (0.00% increase, -0.37% of buffer)	713.693 → 713.701 → 715.840
✅	agent_heroku_amd64	+8.0 KiB (0.00% increase, -0.12% of buffer)	312.817 → 312.825 → 319.080
✅	agent_rpm_amd64	+8.24 KiB (0.00% increase, -0.22% of buffer)	760.895 → 760.903 → 764.570
✅	agent_rpm_amd64_fips	+8.24 KiB (0.00% increase, -0.37% of buffer)	713.677 → 713.685 → 715.840
✅	agent_rpm_arm64	+4.43 KiB (0.00% increase, -1.36% of buffer)	736.811 → 736.816 → 737.130
✅	agent_rpm_arm64_fips	+4.43 KiB (0.00% increase, -0.47% of buffer)	692.907 → 692.912 → 693.820
✅	agent_suse_amd64	+8.24 KiB (0.00% increase, -0.22% of buffer)	760.895 → 760.903 → 764.570
✅	agent_suse_amd64_fips	+8.24 KiB (0.00% increase, -0.37% of buffer)	713.677 → 713.685 → 715.840
✅	agent_suse_arm64	+4.43 KiB (0.00% increase, -1.36% of buffer)	736.811 → 736.816 → 737.130
✅	agent_suse_arm64_fips	+4.43 KiB (0.00% increase, -0.47% of buffer)	692.907 → 692.912 → 693.820
✅	docker_agent_amd64	+11.18 KiB (0.00% increase, -1.75% of buffer)	819.514 → 819.525 → 820.140
✅	docker_agent_arm64	+63.55 KiB (0.01% increase, -15.19% of buffer)	820.481 → 820.544 → 820.890
✅	docker_agent_jmx_amd64	+26.36 KiB (0.00% increase, -5.28% of buffer)	1010.412 → 1010.438 → 1010.900
✅	docker_agent_jmx_arm64	+209.14 KiB (0.02% increase, -37.93% of buffer)	1000.031 → 1000.236 → 1000.570
✅	docker_cluster_agent_amd64	+10.2 KiB (0.00% increase, -1.32% of buffer)	211.376 → 211.386 → 212.130
✅	docker_cluster_agent_arm64	+66.32 KiB (0.03% increase, -8.50% of buffer)	224.458 → 224.523 → 225.220
✅	docker_host_profiler_amd64	-4.19 KiB (0.00% reduction, +0.04% of buffer)	306.802 → 306.798 → 317.700
✅	docker_host_profiler_arm64	+51.95 KiB (0.02% increase, -0.46% of buffer)	318.045 → 318.096 → 328.970
✅	iot_agent_deb_amd64	+4.0 KiB (0.01% increase, -0.37% of buffer)	46.503 → 46.506 → 47.550
✅	iot_agent_deb_arm64	+4.0 KiB (0.01% increase, -0.37% of buffer)	43.159 → 43.163 → 44.220
✅	iot_agent_deb_armhf	+4.0 KiB (0.01% increase, -0.37% of buffer)	43.953 → 43.957 → 45.020
✅	iot_agent_rpm_amd64	+4.0 KiB (0.01% increase, -0.37% of buffer)	46.503 → 46.507 → 47.550
✅	iot_agent_suse_amd64	+4.0 KiB (0.01% increase, -0.37% of buffer)	46.502 → 46.506 → 47.550

8 successful checks with minimal change (< 2 KiB)

	Quality gate	Current Size
✅	docker_cws_instrumentation_amd64	7.443 MiB
✅	docker_cws_instrumentation_arm64	6.877 MiB
✅	docker_dogstatsd_amd64	39.483 MiB
✅	docker_dogstatsd_arm64	37.623 MiB
✅	dogstatsd_deb_amd64	30.224 MiB
✅	dogstatsd_deb_arm64	28.255 MiB
✅	dogstatsd_rpm_amd64	30.224 MiB
✅	dogstatsd_suse_amd64	30.224 MiB

[cit-pr-commenter-54b7da]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Job ID: c65bfec5-e9e7-4389-87f8-2994368c0183

Baseline: e318062
Comparison: b8923ac
Diff

Optimization Goals: ✅ No significant changes detected

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
➖	quality_gate_logs	% cpu utilization	+3.67	[+2.78, +4.57]	1	Logs bounds checks dashboard
➖	quality_gate_security_no_fs_load	memory utilization	+0.59	[+0.50, +0.67]	1	Logs bounds checks dashboard
➖	quality_gate_security_idle	memory utilization	+0.47	[+0.42, +0.52]	1	Logs bounds checks dashboard
➖	dsd_uds_10mb_3k_timestamped_contexts_cpu	% cpu utilization	+0.41	[+0.17, +0.65]	1	Logs
➖	quality_gate_idle	memory utilization	+0.30	[+0.25, +0.34]	1	Logs bounds checks dashboard
➖	dsd_uds_10mb_3k_timestamped_contexts_memory	memory utilization	+0.28	[+0.07, +0.49]	1	Logs
➖	quality_gate_security_mean_fs_load	memory utilization	+0.06	[+0.02, +0.09]	1	Logs bounds checks dashboard
➖	quality_gate_idle_all_features	memory utilization	-0.04	[-0.08, -0.01]	1	Logs bounds checks dashboard
➖	quality_gate_private_action_runner	memory utilization	-0.75	[-0.87, -0.63]	1	Logs bounds checks dashboard
➖	quality_gate_metrics_logs	memory utilization	-1.68	[-1.91, -1.46]	1	Logs bounds checks dashboard

Bounds Checks: ✅ Passed

perf	experiment	bounds_check_name	replicates_passed	observed_value	links
✅	quality_gate_idle	intake_connections	10/10	4 = 4	bounds checks dashboard
✅	quality_gate_idle	memory_usage	10/10	175.97MiB ≤ 178MiB	bounds checks dashboard
✅	quality_gate_idle	total_bytes_received	10/10	745.16KiB ≤ 819.20KiB	bounds checks dashboard
✅	quality_gate_idle_all_features	intake_connections	10/10	4 = 4	bounds checks dashboard
✅	quality_gate_idle_all_features	memory_usage	10/10	524.96MiB ≤ 538MiB	bounds checks dashboard
✅	quality_gate_idle_all_features	total_bytes_received	10/10	1.14MiB ≤ 1.25MiB	bounds checks dashboard
✅	quality_gate_logs	intake_connections	10/10	18 ≤ 40	bounds checks dashboard
✅	quality_gate_logs	memory_usage	10/10	210.98MiB ≤ 229MiB	bounds checks dashboard
✅	quality_gate_logs	missed_bytes	10/10	0B = 0B	bounds checks dashboard
✅	quality_gate_logs	total_bytes_received	10/10	263.47MiB ≤ 292MiB	bounds checks dashboard
✅	quality_gate_metrics_logs	cpu_usage	10/10	413.11 ≤ 2000	bounds checks dashboard
✅	quality_gate_metrics_logs	intake_connections	10/10	19 ≤ 40	bounds checks dashboard
✅	quality_gate_metrics_logs	memory_usage	10/10	397.78MiB ≤ 453MiB	bounds checks dashboard
✅	quality_gate_metrics_logs	missed_bytes	10/10	0B = 0B	bounds checks dashboard
✅	quality_gate_metrics_logs	total_bytes_received	10/10	0.94GiB ≤ 1.04GiB	bounds checks dashboard
✅	quality_gate_private_action_runner	memory_usage	10/10	71.91MiB ≤ 76MiB	bounds checks dashboard
✅	quality_gate_security_idle	cpu_usage	10/10	28.74 ≤ 100	bounds checks dashboard
✅	quality_gate_security_idle	memory_usage	10/10	328.15MiB ≤ 335MiB	bounds checks dashboard
✅	quality_gate_security_mean_fs_load	cpu_usage	10/10	63.57 ≤ 200	bounds checks dashboard
✅	quality_gate_security_mean_fs_load	memory_usage	10/10	302.46MiB ≤ 314MiB	bounds checks dashboard
✅	quality_gate_security_no_fs_load	cpu_usage	10/10	22.69 ≤ 100	bounds checks dashboard
✅	quality_gate_security_no_fs_load	memory_usage	10/10	315.94MiB ≤ 343MiB	bounds checks dashboard

Explanation

Confidence level: 90.00%
Effect size tolerance: |Δ mean %| ≥ 5.00%

Performance changes are noted in the perf column of each table:

• ✅ = significantly better comparison variant performance
• ❌ = significantly worse comparison variant performance
• ➖ = no significant change in performance

A regression test is an A/B test of target performance in a repeatable rig, where "performance" is measured as "comparison variant minus baseline variant" for an optimization goal (e.g., ingress throughput). Due to intrinsic variability in measuring that goal, we can only estimate its mean value for each experiment; we report uncertainty in that value as a 90.00% confidence interval denoted "Δ mean % CI".

For each experiment, we decide whether a change in performance is a "regression" -- a change worth investigating further -- if all of the following criteria are true:

1. Its estimated |Δ mean %| ≥ 5.00%, indicating the change is big enough to merit a closer look.

2. Its 90.00% confidence interval "Δ mean % CI" does not contain zero, indicating that if our statistical model is accurate, there is at least a 90.00% chance there is a difference in performance between baseline and comparison variants.

3. Its configuration does not mark it "erratic".

CI Pass/Fail Decision

✅ Passed. All Quality Gates passed.

• quality_gate_logs, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_logs, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_logs, bounds check missed_bytes: 10/10 replicas passed. Gate passed.
• quality_gate_logs, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_idle, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_idle, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_idle, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_security_no_fs_load, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_security_no_fs_load, bounds check cpu_usage: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check cpu_usage: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check missed_bytes: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_security_mean_fs_load, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_security_mean_fs_load, bounds check cpu_usage: 10/10 replicas passed. Gate passed.
• quality_gate_security_idle, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_security_idle, bounds check cpu_usage: 10/10 replicas passed. Gate passed.
• quality_gate_private_action_runner, bounds check memory_usage: 10/10 replicas passed. Gate passed.