Revert "Revert "[WP] Fix isolation re-apply mechanism"" and fix the related test

[theop-dd]

Reverts #55598 with the required fixes.

Fix a race condition in TestRawPacketRouterSelFlipOnRulesetReload

This race condition caused CI delays and intermittent test failures.

The isolation reapplication change moved the raw_packet_router_sel flip out of ApplyRuleSet. The flip now occurs later, in the eBPF event-reader goroutine, after the snapshot has been replayed.

However, reloadPolicies() returns as soon as a no-op event is queued. The test then immediately reads the selector and expects it to have already flipped.

Depending on the timing, this could result in either of the following:

• The snapshot has not yet triggered the flip, causing the assertion to fail.
• The test starts tearing down while the snapshot is still being replayed. StopReaders then holds the manager lock while handleEvent attempts to call GetMap, causing a deadlock. As a result, the test suite does not exit and the CI job eventually times out.

This issue only affected tests. In production, the selector is still flipped correctly, isolations remain on the active buffer during reloads, and the probe is not stopped while a snapshot is being replayed.

The fix ensures that snapshot replay completes before reloadPolicies() returns. It also moves the selector flip outside the event-reader callback, preventing teardown from deadlocking.

The underlying deadlock is related to ebpf-manager and was addressed for this use case in DataDog/ebpf-manager#288. This PR bumps the ebpf-manager version to include that fix.

Use separate eBPF collections for each raw-packet router buffer

This issue did not affect CI, but it also needed to be fixed. Each raw-packet router buffer now uses its own eBPF collection.

[github-actions]

AI review by Codex (OpenAI) - workflow run

Found one correctness issue.

Overall verdict: patch is incorrect because a failed filter installation can still activate an incomplete router buffer.

[github-actions]

Do not flip when setupRawPacketProgs fails. That function can fail after partially updating the inactive buffer; activating it could leave network-isolation filters missing or incomplete. Return the setup error and keep the current buffer active.

[datadog-datadog-us1-prod]

🎯 Code Coverage (details)
• Patch Coverage: 0.00%
• Overall Coverage: 55.29% (-0.05%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 2c618e3 | Docs | View more details | Give us feedback!}

[dd-octo-sts]

Files inventory check summary

File checks results against ancestor 14a1e606:

Results for datadog-agent_7.85.0~devel.git.15.2c618e3.pipeline.135976808-1_amd64.deb:

No change detected

Results for datadog-iot-agent_7.85.0~devel.git.15.2c618e3.pipeline.135976808-1_amd64.deb:

No change detected

[cit-pr-commenter-54b7da]

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Job ID: 861ea148-c355-4aee-9865-26f80eb956e6

Baseline: 14a1e60
Comparison: 2c618e3
Diff

Optimization Goals: ✅ No significant changes detected

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI	trials	links
➖	dsd_uds_10mb_3k_timestamped_contexts_memory	memory utilization	+0.28	[+0.06, +0.50]	1	Logs
➖	quality_gate_idle_all_features	memory utilization	+0.16	[+0.11, +0.21]	1	Logs bounds checks dashboard
➖	quality_gate_idle	memory utilization	+0.04	[+0.00, +0.08]	1	Logs bounds checks dashboard
➖	quality_gate_security_no_fs_load	memory utilization	-0.26	[-0.33, -0.19]	1	Logs bounds checks dashboard
➖	quality_gate_security_mean_fs_load	memory utilization	-0.27	[-0.30, -0.23]	1	Logs bounds checks dashboard
➖	quality_gate_private_action_runner	memory utilization	-0.27	[-0.40, -0.14]	1	Logs bounds checks dashboard
➖	quality_gate_security_idle	memory utilization	-0.28	[-0.32, -0.24]	1	Logs bounds checks dashboard
➖	dsd_uds_10mb_3k_timestamped_contexts_cpu	% cpu utilization	-0.45	[-0.69, -0.21]	1	Logs
➖	quality_gate_logs	% cpu utilization	-0.55	[-1.43, +0.33]	1	Logs bounds checks dashboard
➖	quality_gate_metrics_logs	memory utilization	-0.69	[-0.92, -0.46]	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 ≤ 5	bounds checks dashboard
✅	quality_gate_idle	memory_usage	10/10	175.24MiB ≤ 179MiB	bounds checks dashboard
✅	quality_gate_idle	total_bytes_received	10/10	750.69KiB ≤ 819.20KiB	bounds checks dashboard
✅	quality_gate_idle_all_features	intake_connections	10/10	3 ≤ 5	bounds checks dashboard
✅	quality_gate_idle_all_features	memory_usage	10/10	473.65MiB ≤ 537MiB	bounds checks dashboard
✅	quality_gate_idle_all_features	total_bytes_received	10/10	1.13MiB ≤ 1.25MiB	bounds checks dashboard
✅	quality_gate_logs	intake_connections	10/10	18 ≤ 40	bounds checks dashboard
✅	quality_gate_logs	memory_usage	10/10	212.24MiB ≤ 228MiB	bounds checks dashboard
✅	quality_gate_logs	missed_bytes	10/10	0B = 0B	bounds checks dashboard
✅	quality_gate_logs	total_bytes_received	10/10	263.15MiB ≤ 292MiB	bounds checks dashboard
✅	quality_gate_metrics_logs	cpu_usage	10/10	364.21 ≤ 2000	bounds checks dashboard
✅	quality_gate_metrics_logs	intake_connections	10/10	16 ≤ 40	bounds checks dashboard
✅	quality_gate_metrics_logs	memory_usage	10/10	446.42MiB ≤ 455MiB	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	71.95MiB ≤ 75MiB	bounds checks dashboard
✅	quality_gate_security_idle	cpu_usage	10/10	31.54 ≤ 100	bounds checks dashboard
✅	quality_gate_security_idle	memory_usage	10/10	328.08MiB ≤ 355MiB	bounds checks dashboard
✅	quality_gate_security_mean_fs_load	cpu_usage	10/10	64.64 ≤ 200	bounds checks dashboard
✅	quality_gate_security_mean_fs_load	memory_usage	10/10	302.83MiB ≤ 335MiB	bounds checks dashboard
✅	quality_gate_security_no_fs_load	cpu_usage	10/10	25.19 ≤ 100	bounds checks dashboard
✅	quality_gate_security_no_fs_load	memory_usage	10/10	310.12MiB ≤ 345MiB	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_idle_all_features, bounds check total_bytes_received: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, bounds check intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_idle_all_features, 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_security_idle, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_private_action_runner, 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, bounds check memory_usage: 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_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 intake_connections: 10/10 replicas passed. Gate passed.
• quality_gate_metrics_logs, bounds check missed_bytes: 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_logs, bounds check memory_usage: 10/10 replicas passed. Gate passed.
• quality_gate_logs, bounds check missed_bytes: 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 intake_connections: 10/10 replicas passed. Gate passed.

[dd-octo-sts]

Static quality checks

✅ Please find below the results from static quality gates
Comparison made with ancestor 14a1e60
📊 Static Quality Gates Dashboard
🔗 SQG Job

Successful checks

Info

	Quality gate	Change	Size (prev → curr → max)
✅	agent_deb_amd64	+17.42 KiB (0.00% increase, -0.62% of buffer)	768.464 → 768.481 → 771.220
✅	agent_deb_amd64_fips	+12.79 KiB (0.00% increase, -0.42% of buffer)	715.052 → 715.064 → 718.020
✅	agent_rpm_amd64	+17.42 KiB (0.00% increase, -0.62% of buffer)	768.448 → 768.465 → 771.190
✅	agent_rpm_amd64_fips	+12.79 KiB (0.00% increase, -0.42% of buffer)	715.036 → 715.048 → 718.020
✅	agent_rpm_arm64	+12.91 KiB (0.00% increase, -0.32% of buffer)	740.392 → 740.405 → 744.370
✅	agent_rpm_arm64_fips	+16.91 KiB (0.00% increase, -0.38% of buffer)	691.525 → 691.542 → 695.880
✅	agent_suse_amd64	+17.42 KiB (0.00% increase, -0.62% of buffer)	768.448 → 768.465 → 771.190
✅	agent_suse_amd64_fips	+12.79 KiB (0.00% increase, -0.42% of buffer)	715.036 → 715.048 → 718.020
✅	agent_suse_arm64	+12.91 KiB (0.00% increase, -0.32% of buffer)	740.392 → 740.405 → 744.370
✅	agent_suse_arm64_fips	+16.91 KiB (0.00% increase, -0.38% of buffer)	691.525 → 691.542 → 695.880
✅	docker_agent_amd64	+19.09 KiB (0.00% increase, -1.88% of buffer)	827.427 → 827.446 → 828.420
✅	docker_agent_arm64	+14.58 KiB (0.00% increase, -0.35% of buffer)	824.427 → 824.442 → 828.440
✅	docker_agent_jmx_amd64	+19.07 KiB (0.00% increase, -1.44% of buffer)	1018.340 → 1018.358 → 1019.630
✅	docker_agent_jmx_arm64	+14.57 KiB (0.00% increase, -0.35% of buffer)	1004.120 → 1004.134 → 1008.130
✅	docker_host_profiler_amd64	+3.35 KiB (0.00% increase, -0.03% of buffer)	307.574 → 307.578 → 319.480
✅	docker_host_profiler_arm64	+3.36 KiB (0.00% increase, -0.03% of buffer)	318.791 → 318.795 → 330.690

17 successful checks with minimal change (< 2 KiB)

	Quality gate	Current Size
✅	agent_heroku_amd64	316.125 MiB
✅	agent_msi	647.746 MiB
✅	docker_cluster_agent_amd64	212.686 MiB
✅	docker_cluster_agent_arm64	225.718 MiB
✅	docker_cws_instrumentation_amd64	7.439 MiB
✅	docker_cws_instrumentation_arm64	6.877 MiB
✅	docker_dogstatsd_amd64	39.557 MiB
✅	docker_dogstatsd_arm64	37.686 MiB
✅	dogstatsd_deb_amd64	30.303 MiB
✅	dogstatsd_deb_arm64	28.315 MiB
✅	dogstatsd_rpm_amd64	30.303 MiB
✅	dogstatsd_suse_amd64	30.303 MiB
✅	iot_agent_deb_amd64	46.651 MiB
✅	iot_agent_deb_arm64	43.280 MiB
✅	iot_agent_deb_armhf	44.097 MiB
✅	iot_agent_rpm_amd64	46.651 MiB
✅	iot_agent_suse_amd64	46.650 MiB

[github-actions]

AI review by Codex (OpenAI) - workflow run

Patch is incorrect. The test remains timing-dependent and does not reliably verify that snapshot replay and buffer flipping completed.

[github-actions]

A fixed sleep preserves the race: slow CI can still read the selector before replay completes, while every run now incurs a one-second delay. More importantly, this does not establish the claimed teardown safety because reloadPolicies still returns before the callback finishes. Add an explicit completion signal and wait for it deterministically.