gohai/cpu: collect L1/L2/L3 cache sizes and CPU packages on Darwin#48312
gohai/cpu: collect L1/L2/L3 cache sizes and CPU packages on Darwin#48312rahulkaukuntla wants to merge 5 commits intomainfrom
Conversation
Add `getSysctlUint64Int64` (via `hw.l1dcachesize`, `hw.l2cachesize`, `hw.l3cachesize`) and populate `CPUPkgs` via `hw.packages` on Darwin. `getSysctlOptional` handles both ENOENT and EINVAL so Apple Silicon (which returns EINVAL for hw.l3cachesize) degrades gracefully to ErrNotCollectable rather than propagating an unexpected error. Fixes AGENTCFG-47 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
github-actions
bot
added
the
short review
Based on Pierre Gimalac's ticket comments: - cpu: fall back to hw.cpufamily when machdep.cpu.family returns ENOENT (Apple Silicon doesn't expose machdep.cpu.family) - cpu: compute CacheSizeKB from L1+L2+L3 bytes / 1024, mirroring Linux ARM64 behavior - platform: populate HardwarePlatform from uname.Machine on Darwin (equivalent to `uname -i` on Linux) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
github-actions
bot
added
medium review
|
@codex review |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 1cad362905
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| machine := utils.StringFromBytes(uname.Machine[:]) | ||
| platformInfo.Machine = utils.NewValue(machine) | ||
| // HardwarePlatform mirrors Machine on Darwin (equivalent to uname -i) | ||
| platformInfo.HardwarePlatform = utils.NewValue(machine) |
There was a problem hiding this comment.
When the agent runs as an x86_64 process under Rosetta on Apple Silicon, fillPlatformInfo() intentionally rewrites Machine to arm64, but HardwarePlatform is now initialized from uname.Machine and never updated in that Rosetta branch. This leaves contradictory architecture fields (machine=arm64 but hardware_platform=x86_64) and can misreport host architecture to downstream metadata consumers that rely on hardware_platform.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Good catch — fixed in 05ff40a. The Rosetta override block now also sets HardwarePlatform = "arm64" alongside Machine and Processor, keeping all three architecture fields consistent.
…lock When running as an x86_64 process under Rosetta 2, fillPlatformInfo already overrides Machine to "arm64" to reflect actual hardware. HardwarePlatform, now populated from uname.Machine in updateUnameInfo, was not updated in that same block, leaving Machine="arm64" but HardwarePlatform="x86_64" — contradictory metadata downstream. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Files inventory check summaryFile checks results against ancestor 17ae85ff: Results for datadog-agent_7.79.0~devel.git.108.cb149d2.pipeline.104210333-1_amd64.deb:No change detected |
Static quality checks✅ Please find below the results from static quality gates 30 successful checks with minimal change (< 2 KiB)
On-wire sizes (compressed)
|
Regression DetectorRegression Detector ResultsMetrics dashboard Baseline: 17ae85f Optimization Goals: ✅ No significant changes detected
|
| perf | experiment | goal | Δ mean % | Δ mean % CI | trials | links |
|---|---|---|---|---|---|---|
| ➖ | docker_containers_cpu | % cpu utilization | -0.16 | [-3.21, +2.89] | 1 | Logs |
Fine details of change detection per experiment
| perf | experiment | goal | Δ mean % | Δ mean % CI | trials | links |
|---|---|---|---|---|---|---|
| ➖ | ddot_logs | memory utilization | +0.21 | [+0.15, +0.26] | 1 | Logs |
| ➖ | otlp_ingest_logs | memory utilization | +0.19 | [+0.09, +0.29] | 1 | Logs |
| ➖ | quality_gate_idle_all_features | memory utilization | +0.15 | [+0.12, +0.19] | 1 | Logs bounds checks dashboard |
| ➖ | ddot_metrics_sum_delta | memory utilization | +0.13 | [-0.04, +0.31] | 1 | Logs |
| ➖ | quality_gate_logs | % cpu utilization | +0.11 | [-1.53, +1.74] | 1 | Logs bounds checks dashboard |
| ➖ | ddot_metrics | memory utilization | +0.06 | [-0.12, +0.25] | 1 | Logs |
| ➖ | quality_gate_idle | memory utilization | +0.06 | [+0.01, +0.11] | 1 | Logs bounds checks dashboard |
| ➖ | ddot_metrics_sum_cumulativetodelta_exporter | memory utilization | +0.03 | [-0.19, +0.26] | 1 | Logs |
| ➖ | file_to_blackhole_500ms_latency | egress throughput | +0.03 | [-0.36, +0.42] | 1 | Logs |
| ➖ | file_to_blackhole_1000ms_latency | egress throughput | +0.02 | [-0.41, +0.44] | 1 | Logs |
| ➖ | file_to_blackhole_100ms_latency | egress throughput | +0.02 | [-0.07, +0.10] | 1 | Logs |
| ➖ | uds_dogstatsd_to_api_v3 | ingress throughput | +0.01 | [-0.19, +0.20] | 1 | Logs |
| ➖ | tcp_dd_logs_filter_exclude | ingress throughput | -0.00 | [-0.11, +0.11] | 1 | Logs |
| ➖ | uds_dogstatsd_to_api | ingress throughput | -0.00 | [-0.20, +0.20] | 1 | Logs |
| ➖ | file_tree | memory utilization | -0.01 | [-0.07, +0.04] | 1 | Logs |
| ➖ | file_to_blackhole_0ms_latency | egress throughput | -0.07 | [-0.52, +0.39] | 1 | Logs |
| ➖ | otlp_ingest_metrics | memory utilization | -0.07 | [-0.22, +0.09] | 1 | Logs |
| ➖ | tcp_syslog_to_blackhole | ingress throughput | -0.10 | [-0.23, +0.02] | 1 | Logs |
| ➖ | docker_containers_cpu | % cpu utilization | -0.16 | [-3.21, +2.89] | 1 | Logs |
| ➖ | ddot_metrics_sum_cumulative | memory utilization | -0.31 | [-0.46, -0.17] | 1 | Logs |
| ➖ | uds_dogstatsd_20mb_12k_contexts_20_senders | memory utilization | -0.45 | [-0.51, -0.38] | 1 | Logs |
| ➖ | quality_gate_metrics_logs | memory utilization | -0.78 | [-1.02, -0.55] | 1 | Logs bounds checks dashboard |
| ➖ | docker_containers_memory | memory utilization | -2.01 | [-2.18, -1.84] | 1 | Logs |
Bounds Checks: ✅ Passed
| perf | experiment | bounds_check_name | replicates_passed | observed_value | links |
|---|---|---|---|---|---|
| ✅ | docker_containers_cpu | simple_check_run | 10/10 | 728 ≥ 26 | |
| ✅ | docker_containers_memory | memory_usage | 10/10 | 277.18MiB ≤ 370MiB | |
| ✅ | docker_containers_memory | simple_check_run | 10/10 | 609 ≥ 26 | |
| ✅ | file_to_blackhole_0ms_latency | memory_usage | 10/10 | 0.19GiB ≤ 1.20GiB | |
| ✅ | file_to_blackhole_0ms_latency | missed_bytes | 10/10 | 0B = 0B | |
| ✅ | file_to_blackhole_1000ms_latency | memory_usage | 10/10 | 0.23GiB ≤ 1.20GiB | |
| ✅ | file_to_blackhole_1000ms_latency | missed_bytes | 10/10 | 0B = 0B | |
| ✅ | file_to_blackhole_100ms_latency | memory_usage | 10/10 | 0.19GiB ≤ 1.20GiB | |
| ✅ | file_to_blackhole_100ms_latency | missed_bytes | 10/10 | 0B = 0B | |
| ✅ | file_to_blackhole_500ms_latency | memory_usage | 10/10 | 0.21GiB ≤ 1.20GiB | |
| ✅ | file_to_blackhole_500ms_latency | missed_bytes | 10/10 | 0B = 0B | |
| ✅ | quality_gate_idle | intake_connections | 10/10 | 3 = 3 | bounds checks dashboard |
| ✅ | quality_gate_idle | memory_usage | 10/10 | 174.45MiB ≤ 175MiB | bounds checks dashboard |
| ✅ | quality_gate_idle_all_features | intake_connections | 10/10 | 2 ≤ 3 | bounds checks dashboard |
| ✅ | quality_gate_idle_all_features | memory_usage | 10/10 | 509.98MiB ≤ 550MiB | bounds checks dashboard |
| ✅ | quality_gate_logs | intake_connections | 10/10 | 4 ≤ 6 | bounds checks dashboard |
| ✅ | quality_gate_logs | memory_usage | 10/10 | 205.18MiB ≤ 220MiB | bounds checks dashboard |
| ✅ | quality_gate_logs | missed_bytes | 10/10 | 0B = 0B | bounds checks dashboard |
| ✅ | quality_gate_metrics_logs | cpu_usage | 10/10 | 330.83 ≤ 2000 | bounds checks dashboard |
| ✅ | quality_gate_metrics_logs | intake_connections | 10/10 | 4 ≤ 6 | bounds checks dashboard |
| ✅ | quality_gate_metrics_logs | memory_usage | 10/10 | 399.00MiB ≤ 475MiB | bounds checks dashboard |
| ✅ | quality_gate_metrics_logs | missed_bytes | 10/10 | 0B = 0B | 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_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_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 cpu_usage: 10/10 replicas passed. Gate passed.
- quality_gate_metrics_logs, bounds check missed_bytes: 10/10 replicas passed. Gate passed.
- quality_gate_idle, 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_logs, 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.
rahulkaukuntla
added
the
qa/done
dustmop
left a comment
dustmop
left a comment
There was a problem hiding this comment.
LGTM, with one small nitpick
pkg/gohai/cpu/cpu_darwin.go
Outdated
|
|
||
| // type returned by sysctl is uint64, stored as uint64; treats EINVAL as ErrNotCollectable | ||
| // because some keys (e.g. hw.l3cachesize) return EINVAL on hardware that lacks the feature | ||
| func getSysctlUint64Int64(key string) utils.Value[uint64] { |
There was a problem hiding this comment.
I'm curious why you're naming this function with both Uint64 and Int64? Why not just call it getSysctlInt64?
There was a problem hiding this comment.
Good point — renamed to getSysctlInt64. The storage type is already conveyed by the return type Value[uint64], so the Uint64 prefix was redundant.
The storage type is already conveyed by the return type Value[uint64]. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2 participants
Summary
Addresses all Darwin gaps identified in AGENTCFG-47 (Pierre Gimalac's comments):
pkg/gohai/cpu(Darwin):CacheSizeL1Bytes,CacheSizeL2Bytes,CacheSizeL3Bytesviahw.l1dcachesize,hw.l2cachesize,hw.l3cachesizeCPUPkgsviahw.packagesCacheSizeKBas sum of available cache levels / 1024 (mirrors Linux ARM64)hw.cpufamilyforFamilywhenmachdep.cpu.familyreturns ENOENT (Apple Silicon)pkg/gohai/platform(Darwin):HardwarePlatformfromuname.Machine(equivalent touname -ion Linux)Apple Silicon graceful degradation:
getSysctlOptionalmaps bothENOENTandEINVALtoErrNotCollectable— needed becausehw.l3cachesizereturnsEINVALon Apple Silicon (key exists in MIB but hardware has no discrete L3)Familyfalls back frommachdep.cpu.family(Intel) tohw.cpufamily(all Macs)Not addressed (noted as "not very interesting for customers" per ticket):
cpu.CPUNumaNodeson Darwin — macOS doesn't expose NUMA topology/sysscanning logic from the ARM64 implTest plan
go test -tags test ./pkg/gohai/cpu/... ./pkg/gohai/platform/...— all pass on Apple Silicondda inv linter.go --targets=./pkg/gohai/cpu,./pkg/gohai/platform— 0 issuesFixes AGENTCFG-47
🤖 Generated with Claude Code