[procmgr] add Windows LogonUser spawn when supervisor token does not match by jose-manuel-almaza · Pull Request #55834 · DataDog/datadog-agent

jose-manuel-almaza · 2026-09-03T14:41:17Z

What does this PR do?

Rebases Windows agent-profile LogonUser spawn onto main after #56015 (supervisor-token inherit).

Windows spawn router:

Privileged (datadog-agent-process) and agent inherit (supervisor already runs as the installed agent account): CreateProcessW in the supervisor's security context. Job attached at create via PROC_THREAD_ATTRIBUTE_JOB_LIST.
Agent logon (supervisor token does not match): LogonUser + CreateProcessAsUserW, with the user profile kept loaded until the child exits. Job attached the same way.

Windows no longer uses Tokio Command::spawn. CreateProcessAsUserW is not used for inherit: that API needs SeIncreaseQuotaPrivilege, which the installed agent account does not hold.

Default MSI (process manager as ddagentuser) still inherits. No behavior change for those children. LogonUser is for a future LocalSystem supervisor (after A1 + #55921).

Unix behavior is unchanged.

Motivation

If dd-procmgrd runs as LocalSystem, agent-profile children must log on as the installed agent account. This must land before the MSI switches the process manager service account.

Review follow-up from the previous stack: both privilege levels use CreateProcess* and pass the job at create time, instead of Tokio spawn plus post-assign.

Describe how you validated your changes

cargo test -p dd-procmgrd --features test-helpers --lib
cargo test -p dd-procmgrd --test e2e --features test-helpers reload_
Windows cfg(windows) spawn paths need CI (not compiled on macOS)

Additional Notes

Password retrieval is not in this PR. LocalSystem spawn needs [procmgr] read installer LSA password for agent spawn logon #55921 (installer LSA).
Privileged children still run as the supervisor. Real LocalSystem process-agent is A5.
History was rewritten onto current main after [procmgr] resolve Windows spawn identity and inherit supervisor token #56015. Previous tip remains recoverable on GitHub.

cit-pr-commenter-54b7da · 2026-09-03T15:37:27Z

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Job ID: 57417f9d-ab55-4960-a3e7-00468fe410a0

Baseline: 6179283
Comparison: a832dce
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	+0.65	[-0.20, +1.50]	1	Logs bounds checks dashboard
➖	dsd_uds_10mb_3k_timestamped_contexts_cpu	% cpu utilization	+0.61	[+0.38, +0.85]	1	Logs
➖	quality_gate_security_mean_fs_load	memory utilization	+0.56	[+0.51, +0.60]	1	Logs bounds checks dashboard
➖	quality_gate_metrics_logs	memory utilization	-0.06	[-0.29, +0.17]	1	Logs bounds checks dashboard
➖	quality_gate_idle	memory utilization	-0.12	[-0.16, -0.08]	1	Logs bounds checks dashboard
➖	quality_gate_private_action_runner	memory utilization	-0.18	[-0.31, -0.06]	1	Logs bounds checks dashboard
➖	quality_gate_security_idle	memory utilization	-0.22	[-0.25, -0.19]	1	Logs bounds checks dashboard
➖	quality_gate_security_no_fs_load	memory utilization	-0.33	[-0.40, -0.25]	1	Logs bounds checks dashboard
➖	quality_gate_idle_all_features	memory utilization	-0.37	[-0.41, -0.34]	1	Logs bounds checks dashboard
➖	dsd_uds_10mb_3k_timestamped_contexts_memory	memory utilization	-1.48	[-1.69, -1.27]	1	Logs

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	172.46MiB ≤ 179MiB	bounds checks dashboard
✅	quality_gate_idle	total_bytes_received	10/10	747.75KiB ≤ 819.20KiB	bounds checks dashboard
✅	quality_gate_idle_all_features	intake_connections	10/10	4 ≤ 5	bounds checks dashboard
✅	quality_gate_idle_all_features	memory_usage	10/10	522.04MiB ≤ 537MiB	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	212.98MiB ≤ 228MiB	bounds checks dashboard
✅	quality_gate_logs	missed_bytes	10/10	0B = 0B	bounds checks dashboard
✅	quality_gate_logs	total_bytes_received	10/10	263.43MiB ≤ 292MiB	bounds checks dashboard
✅	quality_gate_metrics_logs	cpu_usage	10/10	364.30 ≤ 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	433.10MiB ≤ 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	72.98MiB ≤ 75MiB	bounds checks dashboard
✅	quality_gate_security_idle	cpu_usage	10/10	31.40 ≤ 100	bounds checks dashboard
✅	quality_gate_security_idle	memory_usage	10/10	325.35MiB ≤ 355MiB	bounds checks dashboard
✅	quality_gate_security_mean_fs_load	cpu_usage	10/10	62.02 ≤ 200	bounds checks dashboard
✅	quality_gate_security_mean_fs_load	memory_usage	10/10	306.12MiB ≤ 335MiB	bounds checks dashboard
✅	quality_gate_security_no_fs_load	cpu_usage	10/10	22.81 ≤ 100	bounds checks dashboard
✅	quality_gate_security_no_fs_load	memory_usage	10/10	307.63MiB ≤ 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:

Its estimated |Δ mean %| ≥ 5.00%, indicating the change is big enough to merit a closer look.
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.
Its configuration does not mark it "erratic".

CI Pass/Fail Decision

✅ Passed. All Quality Gates passed.

quality_gate_metrics_logs, bounds check missed_bytes: 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 intake_connections: 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_private_action_runner, 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_security_no_fs_load, bounds check memory_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_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_idle, bounds check intake_connections: 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 intake_connections: 10/10 replicas passed. Gate passed.
quality_gate_logs, bounds check total_bytes_received: 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_mean_fs_load, bounds check memory_usage: 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 total_bytes_received: 10/10 replicas passed. Gate passed.
quality_gate_idle_all_features, bounds check intake_connections: 10/10 replicas passed. Gate passed.

dd-octo-sts · 2026-09-03T15:48:57Z

Files inventory check summary

File checks results against ancestor 61792831:

Results for datadog-agent_7.85.0~devel.git.163.36a715f.pipeline.136848129-1_amd64.deb:

No change detected

Results for datadog-iot-agent_7.85.0~devel.git.163.36a715f.pipeline.136848129-1_amd64.deb:

No change detected

dd-octo-sts · 2026-09-04T08:25:08Z

Static quality checks

✅ Please find below the results from static quality gates
Comparison made with ancestor 6179283
📊 Static Quality Gates Dashboard
🔗 SQG Job

Successful checks

Info

	Quality gate	Change	Size (prev → curr → max)
✅	agent_msi	-53.33 KiB (0.01% reduction, +0.37% of buffer)	651.653 → 651.601 → 665.630
✅	docker_agent_amd64	+7.56 KiB (0.00% increase, -6.99% of buffer)	840.014 → 840.022 → 840.120
✅	docker_agent_arm64	+3.58 KiB (0.00% increase, -0.11% of buffer)	836.004 → 836.007 → 839.280
✅	docker_agent_jmx_amd64	+7.57 KiB (0.00% increase, -1.83% of buffer)	1030.927 → 1030.934 → 1031.330
✅	docker_agent_jmx_arm64	+3.59 KiB (0.00% increase, -0.11% of buffer)	1015.696 → 1015.699 → 1018.970
✅	docker_cluster_agent_amd64	+4.49 KiB (0.00% increase, -0.54% of buffer)	212.881 → 212.885 → 213.700
✅	docker_host_profiler_amd64	+10.66 KiB (0.00% increase, -0.09% of buffer)	307.729 → 307.739 → 319.490
✅	docker_host_profiler_arm64	+6.68 KiB (0.00% increase, -0.06% of buffer)	318.910 → 318.917 → 330.690

25 successful checks with minimal change (< 2 KiB)

	Quality gate	Current Size
✅	agent_deb_amd64	781.093 MiB
✅	agent_deb_amd64_fips	715.756 MiB
✅	agent_heroku_amd64	316.222 MiB
✅	agent_rpm_amd64	781.077 MiB
✅	agent_rpm_amd64_fips	715.740 MiB
✅	agent_rpm_arm64	752.022 MiB
✅	agent_rpm_arm64_fips	692.195 MiB
✅	agent_suse_amd64	781.077 MiB
✅	agent_suse_amd64_fips	715.740 MiB
✅	agent_suse_arm64	752.022 MiB
✅	agent_suse_arm64_fips	692.195 MiB
✅	docker_cluster_agent_arm64	225.905 MiB
✅	docker_cws_instrumentation_amd64	7.439 MiB
✅	docker_cws_instrumentation_arm64	6.877 MiB
✅	docker_dogstatsd_amd64	39.597 MiB
✅	docker_dogstatsd_arm64	37.686 MiB
✅	dogstatsd_deb_amd64	30.343 MiB
✅	dogstatsd_deb_arm64	28.351 MiB
✅	dogstatsd_rpm_amd64	30.343 MiB
✅	dogstatsd_suse_amd64	30.343 MiB
✅	iot_agent_deb_amd64	46.669 MiB
✅	iot_agent_deb_arm64	43.286 MiB
✅	iot_agent_deb_armhf	44.108 MiB
✅	iot_agent_rpm_amd64	46.669 MiB
✅	iot_agent_suse_amd64	46.668 MiB

pgimalac

Any chance this could be split into several smaller PRs ?
4000 lines is not really reviewable tbh...

jose-manuel-almaza · 2026-09-07T11:28:51Z

Any chance this could be split into several smaller PRs ? 4000 lines is not really reviewable tbh...

@pgimalac
Fair point on size.
Most of the diff is the Windows LogonUser path (~2,500 lines). The rest is plumbing this sits on: ProcessHandle, SpawnRequest, and unix spawn code extracted from process.rs, plus the exit watcher fold-in from #55874. L$ and _SC_ are already split out (#55921 and RFC hold).
If the size is blocking review, I can split into a no-behavior-change refactor PR first, then the Windows LogonUser work on top
wdyt?

pgimalac · 2026-09-07T11:30:02Z

@jose-manuel-almaza yeah would be nice if possible, currently it's a lot of changes at once and I don't have context on most of that so it's not easy to review

jose-manuel-almaza · 2026-09-07T14:05:45Z

@pgimalac Split is done.

[procmgr] extract spawn behind platform::spawn_child_handle #55944: spawn abstraction refactor (no behavior change on either platform). That is where the unix spawn extraction and exit watcher fold-in live.
This PR: LogonUser on Windows only, ~3k lines, one commit, stacked on [procmgr] extract spawn behind platform::spawn_child_handle #55944 for review.

The platform/unix thread is resolved by that split: those files are no longer in this diff. Happy to walk through either PR if useful.

### What does this PR do? Extracts spawn behind `platform::spawn_child_handle()` and introduces `ProcessHandle` and `SpawnRequest`. Unix and Windows still use `Command::spawn`. Successful `ManagedProcess::spawn` starts the exit watcher itself, so the manager no longer calls `spawn` then `spawn_watcher` as two steps. Spawn behavior is unchanged. ### Motivation Reviewer-requested split of #55834. This lands the cross-platform spawn plumbing first. Windows LogonUser agent-profile spawn stays in #55834. ### Describe how you validated your changes - `cargo test -p dd-procmgrd` - `cargo clippy -p dd-procmgrd --all-targets -- -D warnings` - `cargo test -p dd-procmgrd --test e2e --features test-helpers reload_` - `cargo test -p dd-procmgrd --test e2e --features test-helpers list` - `cargo test -p dd-procmgrd --test e2e --features test-helpers describe` - `dda inv test --targets=./pkg/procmgr/coat/...` Bazel `//pkg/procmgr/rust:dd-procmgrd_test` is linux/windows only (skipped locally). ### Additional Notes Mechanical refactor only. No LogonUser, no password read, no change to privileged spawn. Keep #55834 open. After this PR merges, rebase #55834 onto `main` so that diff is LogonUser-only. Co-authored-by: josemanuel.almaza <josemanuel.almaza@datadoghq.com>

### What does this PR do? Splits the monolithic `platform/windows.rs` (~684 lines) into focused modules: `child_env`, `console`, `job_object`, `process`, `runtime_user`, and `wide`. Public API and spawn behavior are unchanged. Windows still uses `Command::spawn` via `spawn_child_handle(name, config)`. ### Motivation #55834 was too large to review in one pass. #55944 extracted the cross-platform spawn plumbing first. This PR reorganizes the Windows platform code only, so follow-up work on identity resolution and LogonUser spawn can land in smaller, reviewable PRs. ### Describe how you validated your changes - `cargo clippy -p dd-procmgrd --all-targets -- -D warnings` - `cargo test -p dd-procmgrd` Bazel `//pkg/procmgr/rust:dd-procmgrd_test` on Windows amd64 (CI). ### Additional Notes Mechanical refactor only. No LogonUser, no credential types, no identity modules, no change to the `ManagedProcess` spawn path. **Review map:** 1. `platform/windows/mod.rs`: re-exports and install/registry helpers (same surface as old `windows.rs`) 2. `child_env.rs`, `console.rs`, `job_object.rs`, `runtime_user.rs`, `wide.rs`: pure moves from #55944 3. `process.rs`: `terminate_process_by_pid` (used by `send_force_kill`) 4. `platform/windows/spawn.rs`: unchanged `Command::spawn` path Stacks on #55944. After that PR merges, retarget base to `main`. Co-authored-by: josemanuel.almaza <josemanuel.almaza@datadoghq.com>

…#56015) ## Summary Adds Windows spawn identity resolution (`SpawnCredential`, installed-user registry/SID lookup) and agent-profile `CreateProcessW` when dd-procmgrd already runs as the installed agent account (supervisor-token inherit). `Privileged` profile keeps `Command::spawn`. Agent profile fails when the supervisor token does not match the installed account, with no wrong-identity fallback. Expands `ProcessHandle` on Windows for supervised Win32 children (suspended spawn, job assignment, restricted stdio inheritance). Preserves registry install-account display in list/describe (`registry_domain` vs `logon_domain`). ## Motivation #55834 was too large to review in one pass. #55944 and #56010 landed spawn plumbing and platform structure. This PR adds identity and the supervisor-inherit spawn path so the LogonUser work in #55834 can focus on the remaining credential cases. ## Describe how you validated your changes - `cargo test -p dd-procmgrd` - `cargo clippy -p dd-procmgrd --all-targets -- -D warnings` - reload / list / describe e2e (test-helpers) - pending Windows CI - COAT tests - pending Windows CI Bazel Windows CI. ## Additional Notes **Review map:** 1. `local_agent_account.rs`, `token_identity.rs`, `sid.rs`: identity resolution 2. `spawn/managed.rs`: inherit vs explicit error when supervisor token does not match installed account 3. `spawn/credential.rs`: supervisor-only token duplication 4. `handle.rs` + `spawn/suspended.rs`: supervised child lifecycle 5. `process.rs`: `ManagedProcess` credential field + unified `spawn_child_handle(self)` Follows #56010 (merged). No LogonUser, no installer password read. Co-authored-by: josemanuel.almaza <josemanuel.almaza@datadoghq.com>

Wire agent-profile LogonUser spawn on top of CreateProcessW inherit. Privileged and same-account agent children share that inherit path, with the job attached at CreateProcess time. LogonUser remains for agent children when the supervisor token does not match. Default MSI still inherits as the installed agent account.

Compact SupervisorAccount and PasswordLogon destructuring so Bazel rustfmt checks pass in CI.

Attach JOB_LIST before HANDLE_LIST in STARTUPINFOEX. Use CREATE_NO_WINDOW only for hidden children. Remove dead from_tokio_child.

GitLab CI runs dd-procmgrd inside a job object, so create-time JOB_LIST needs CREATE_BREAKAWAY_FROM_JOB. On ERROR_INVALID_PARAMETER, fall back to post-create nested job assignment and fix the Clippy items-after-test-module lint in win32.rs.

Apply rustfmt to job_object.rs and inherit_supervisor.rs so the dd-procmgrd-lib and dd-procmgrd_test rustfmt checks pass in CI.

When create-time JOB_LIST is rejected (parent in a foreign job), retry with CREATE_NEW_CONSOLE so spawn works in no-console CI runners.

Detect foreign job parents with IsProcessInJob and use the pre-JOB_LIST suspended create path (assign + ResumeThread) instead of create-time JOB_LIST. Replaces the ineffective CREATE_NEW_CONSOLE-only fallback.

Apply rustfmt import ordering across the Windows job-object spawn files and silence clippy on the post-assign spawn helpers so Bazel rustfmt and clippy checks pass in CI.

Use sleep_test_config with a short stop_timeout and a 10s Windows sleep so ping-based test children force-kill quickly instead of hanging past the 60s per-test limit.

Use wait_for_stop after request_stop in process tests so ping-based sleep children escalate to force-kill instead of waiting for natural exit.

PowerShell is not always on PATH in the Windows Bazel CI container, so test_shutdown_all_sigkill_on_timeout could fail immediately on spawn. Use ping for stubborn-child helpers and shorten Windows TEST_SLEEP_SECS.

Restore a longer Windows sleep child duration so ping does not exit before request_stop under CI scheduling jitter, which left processes in Exited instead of Stopped.

jose-manuel-almaza added the qa/done QA done before merge and regressions are covered by tests label Sep 3, 2026

This comment has been minimized.

Sign in to view

jose-manuel-almaza marked this pull request as ready for review September 4, 2026 16:01

jose-manuel-almaza requested review from a team as code owners September 4, 2026 16:01

jose-manuel-almaza requested review from clarkb7 and pgimalac September 4, 2026 16:01

jose-manuel-almaza force-pushed the jose/procmgr-windows-agent-spawn branch from b868ff7 to a7295e2 Compare September 4, 2026 20:01

Base automatically changed from jose/procmgr-spawn-profile-api to main September 6, 2026 17:10

gh-worker-dd-mergequeue-cf854d Bot requested review from a team as code owners September 6, 2026 17:10

jose-manuel-almaza force-pushed the jose/procmgr-windows-agent-spawn branch from a7295e2 to 2e58f19 Compare September 7, 2026 05:33

dd-octo-sts Bot added internal Identify a non-fork PR team/agent-runtimes labels Sep 7, 2026

github-actions Bot added the long review PR is complex, plan time to review it label Sep 7, 2026

dd-octo-sts Bot added the team/windows-products label Sep 7, 2026

jose-manuel-almaza removed request for a team September 7, 2026 05:39

jose-manuel-almaza added the ask-review Ask in slack required teams to review this PR label Sep 7, 2026

pgimalac reviewed Sep 7, 2026

View reviewed changes

Comment thread pkg/procmgr/rust/src/platform/unix/spawn/managed.rs Outdated

jose-manuel-almaza mentioned this pull request Sep 7, 2026

[procmgr] extract spawn behind platform::spawn_child_handle #55944

Merged

jose-manuel-almaza changed the base branch from main to jose/procmgr-spawn-abstraction September 7, 2026 13:42

jose-manuel-almaza force-pushed the jose/procmgr-windows-agent-spawn branch from faec628 to 9e1930e Compare September 7, 2026 14:01

jose-manuel-almaza requested a review from pgimalac September 8, 2026 10:36

jose-manuel-almaza force-pushed the jose/procmgr-spawn-abstraction branch from 90103cb to d68f3ff Compare September 8, 2026 14:01

jose-manuel-almaza mentioned this pull request Sep 8, 2026

[procmgr] split Windows platform into focused modules #56010

Merged

pgimalac reviewed Sep 8, 2026

View reviewed changes

jose-manuel-almaza mentioned this pull request Sep 8, 2026

[procmgr] resolve Windows spawn identity and inherit supervisor token #56015

Merged

clarkb7 approved these changes Sep 8, 2026

View reviewed changes

Comment thread pkg/procmgr/rust/src/spawn/request.rs

Comment thread pkg/procmgr/rust/src/platform/windows/child_env.rs

Comment thread pkg/procmgr/rust/src/platform/windows/spawn/managed.rs Outdated

Base automatically changed from jose/procmgr-spawn-abstraction to main September 8, 2026 22:20

jose-manuel-almaza force-pushed the jose/procmgr-windows-agent-spawn branch from 0968953 to 52195ed Compare September 10, 2026 15:34

jose-manuel-almaza changed the title ~~[procmgr] spawn agent-profile children with LogonUser on Windows~~ [procmgr] add Windows LogonUser spawn when supervisor token does not match Sep 10, 2026

jose-manuel-almaza added 2 commits September 10, 2026 18:13

fix(procmgr): satisfy rustfmt in user_profile match arms

aa01d03

Compact SupervisorAccount and PasswordLogon destructuring so Bazel rustfmt checks pass in CI.

fix(procmgr): fix Windows spawn error 87 with create-time JOB_LIST

36ac90a

Attach JOB_LIST before HANDLE_LIST in STARTUPINFOEX. Use CREATE_NO_WINDOW only for hidden children. Remove dead from_tokio_child.

jose-manuel-almaza force-pushed the jose/procmgr-windows-agent-spawn branch from 7414868 to 36ac90a Compare September 10, 2026 16:29

jose-manuel-almaza added 9 commits September 10, 2026 18:51

style(procmgr): rustfmt Windows spawn sources

c554ca0

Apply rustfmt to job_object.rs and inherit_supervisor.rs so the dd-procmgrd-lib and dd-procmgrd_test rustfmt checks pass in CI.

fix(procmgr): use CREATE_NEW_CONSOLE on Windows spawn fallback

1d83e78

When create-time JOB_LIST is rejected (parent in a foreign job), retry with CREATE_NEW_CONSOLE so spawn works in no-console CI runners.

fix(procmgr): spawn under GitLab Windows job objects

0c00ea6

Detect foreign job parents with IsProcessInJob and use the pre-JOB_LIST suspended create path (assign + ResumeThread) instead of create-time JOB_LIST. Replaces the ineffective CREATE_NEW_CONSOLE-only fallback.

style(procmgr): rustfmt and clippy Windows spawn sources

f3408d8

Apply rustfmt import ordering across the Windows job-object spawn files and silence clippy on the post-assign spawn helpers so Bazel rustfmt and clippy checks pass in CI.

fix(procmgr): keep Windows stop tests under Tokio timeout

bb00981

Use sleep_test_config with a short stop_timeout and a 10s Windows sleep so ping-based test children force-kill quickly instead of hanging past the 60s per-test limit.

fix(procmgr): force-kill ping children in Windows stop unit tests

a832dce

Use wait_for_stop after request_stop in process tests so ping-based sleep children escalate to force-kill instead of waiting for natural exit.

fix(procmgr): use ping instead of powershell in Windows shutdown tests

36a715f

PowerShell is not always on PATH in the Windows Bazel CI container, so test_shutdown_all_sigkill_on_timeout could fail immediately on spawn. Use ping for stubborn-child helpers and shorten Windows TEST_SLEEP_SECS.

fix(procmgr): avoid ping race in Windows stop unit tests

103c8f4

Restore a longer Windows sleep child duration so ping does not exit before request_stop under CI scheduling jitter, which left processes in Exited instead of Stopped.

Uh oh!

Conversation

jose-manuel-almaza commented Sep 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What does this PR do?

Motivation

Describe how you validated your changes

Additional Notes

Uh oh!

This comment has been minimized.

cit-pr-commenter-54b7da Bot commented Sep 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Regression Detector

Regression Detector Results

Optimization Goals: ✅ No significant changes detected

Fine details of change detection per experiment

Bounds Checks: ✅ Passed

Explanation

CI Pass/Fail Decision

Uh oh!

dd-octo-sts Bot commented Sep 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Files inventory check summary

Results for datadog-agent_7.85.0~devel.git.163.36a715f.pipeline.136848129-1_amd64.deb:

Results for datadog-iot-agent_7.85.0~devel.git.163.36a715f.pipeline.136848129-1_amd64.deb:

Uh oh!

dd-octo-sts Bot commented Sep 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Static quality checks

Info

Uh oh!

pgimalac left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jose-manuel-almaza commented Sep 7, 2026

Uh oh!

pgimalac commented Sep 7, 2026

Uh oh!

jose-manuel-almaza commented Sep 7, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

jose-manuel-almaza commented Sep 3, 2026 •

edited

Loading

cit-pr-commenter-54b7da Bot commented Sep 3, 2026 •

edited

Loading

dd-octo-sts Bot commented Sep 3, 2026 •

edited

Loading

dd-octo-sts Bot commented Sep 4, 2026 •

edited

Loading