Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Remove inlined function PC resolution for stack traces, remove DWARF caching #32166

Conversation

grantseltzer
Copy link
Member

@grantseltzer grantseltzer commented Dec 13, 2024

What does this PR do?

  • Removes the map and its usage that contains the inlined function PCs, which will no longer be in stack traces. This resolves a huge amount of memory usage.

  • Removes caching of DWARF information for each inspected binary. This resolves a large memory leak.

  • Filters return parameters from instrumentation for now.

Motivation

We noticed a large amount of memory being used by Go DI within the system probe and determined the program counter map was a primary culprit.

We also were caching the DWARF information for each binary. That data was only being used when a new probe was being created which is infrequent enough where it's not worth storing (and never clearing).

Describe how you validated your changes

Working on producing local profiling comparison.


Here is an example of events of the sample not_inlined() function:

After:

{
  "service": "go-di-sample-service",
  "message": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.not_inlined {\"entry\":{}}",
  "ddsource": "dd_debugger",
  "ddtags": "",
  "logger": {
    "name": "",
    "method": ""
  },
  "debugger": {
    "snapshot": {
      "id": "8aad8538-b8ee-11ef-b1a5-001c42659120",
      "timestamp": 1734052014013,
      "language": "go",
      "probe": {
        "id": "e504163d-f367-4522-8905-fe8bc34eb975",
        "location": {
          "method": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.not_inlined"
        }
      },
      "captures": {
        "entry": {}
      },
      "stack": [
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.not_inlined",
          "lineNumber": 51
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.call_inlined_func_chain",
          "lineNumber": 30
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.ExecuteStackAndInlining",
          "lineNumber": 56
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/sample_service/sample_service.go",
          "function": "main.main",
          "lineNumber": 19
        }
      ]
    }
  },
  "duration": 0
}

Before:

{
  "service": "go-di-sample-service",
  "message": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.not_inlined {\"entry\":{}}",
  "ddsource": "dd_debugger",
  "ddtags": "",
  "logger": {
    "name": "",
    "method": ""
  },
  "debugger": {
    "snapshot": {
      "id": "c517407a-b8ee-11ef-a5da-001c42659120",
      "timestamp": 1734052112014,
      "language": "go",
      "probe": {
        "id": "e504163d-f367-4522-8905-fe8bc34eb975",
        "location": {
          "method": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.not_inlined"
        }
      },
      "captures": {
        "entry": {}
      },
      "stack": [
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.not_inlined",
          "lineNumber": 51
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.inline_me_3 [inlined in github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.call_inlined_func_chain]",
          "lineNumber": 41
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.inline_me_2 [inlined in github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.call_inlined_func_chain]",
          "lineNumber": 36
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.inline_me_1 [inlined in github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.call_inlined_func_chain]",
          "lineNumber": 31
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.call_inlined_func_chain",
          "lineNumber": 30
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/stacktraces.go",
          "function": "github.com/DataDog/datadog-agent/pkg/dynamicinstrumentation/testutil/sample.ExecuteStackAndInlining",
          "lineNumber": 56
        },
        {
          "fileName": "/home/vagrant/datadog-agent/pkg/dynamicinstrumentation/testutil/sample/sample_service/sample_service.go",
          "function": "main.main",
          "lineNumber": 19
        }
      ]
    }
  },
  "duration": 0
}

@grantseltzer grantseltzer added component/system-probe team/dynamic-instrumentation Dynamic Instrumentation qa/done QA done before merge and regressions are covered by tests labels Dec 13, 2024
@grantseltzer grantseltzer requested review from a team as code owners December 13, 2024 18:13
@github-actions github-actions bot added medium review PR review might take time and removed component/system-probe labels Dec 13, 2024
@agent-platform-auto-pr
Copy link
Contributor

agent-platform-auto-pr bot commented Dec 13, 2024

Package size comparison

Comparison with ancestor f483cc4c0dffec6f7bcd4824dc8df65c15957e27

Diff per package
package diff status size ancestor threshold
datadog-agent-amd64-deb 0.00MB 1265.94MB 1265.94MB 140.00MB
datadog-iot-agent-amd64-deb 0.00MB 113.28MB 113.28MB 10.00MB
datadog-dogstatsd-amd64-deb 0.00MB 78.52MB 78.52MB 10.00MB
datadog-heroku-agent-amd64-deb 0.00MB 502.50MB 502.50MB 70.00MB
datadog-agent-x86_64-rpm 0.00MB ⚠️ 1275.18MB 1275.18MB 140.00MB
datadog-agent-x86_64-suse 0.00MB ⚠️ 1275.18MB 1275.18MB 140.00MB
datadog-iot-agent-x86_64-rpm 0.00MB ⚠️ 113.35MB 113.35MB 10.00MB
datadog-iot-agent-x86_64-suse 0.00MB ⚠️ 113.35MB 113.35MB 10.00MB
datadog-dogstatsd-x86_64-rpm -0.00MB 78.59MB 78.59MB 10.00MB
datadog-dogstatsd-x86_64-suse -0.00MB 78.59MB 78.59MB 10.00MB
datadog-agent-arm64-deb -0.00MB 1001.02MB 1001.02MB 140.00MB
datadog-iot-agent-arm64-deb -0.00MB 108.76MB 108.76MB 10.00MB
datadog-dogstatsd-arm64-deb 0.00MB 55.74MB 55.74MB 10.00MB
datadog-agent-aarch64-rpm -0.00MB 1010.23MB 1010.24MB 140.00MB
datadog-iot-agent-aarch64-rpm -0.00MB 108.83MB 108.83MB 10.00MB

Decision

⚠️ Warning

@agent-platform-auto-pr
Copy link
Contributor

agent-platform-auto-pr bot commented Dec 13, 2024

Test changes on VM

Use this command from test-infra-definitions to manually test this PR changes on a VM:

inv aws.create-vm --pipeline-id=51775098 --os-family=ubuntu

Note: This applies to commit 580cbf7

Copy link

cit-pr-commenter bot commented Dec 13, 2024

Regression Detector

Regression Detector Results

Metrics dashboard
Target profiles
Run ID: e8e9ef55-71b9-4141-88b3-031e78379439

Baseline: f93202d
Comparison: 580cbf7
Diff

Optimization Goals: ✅ No significant changes detected

Fine details of change detection per experiment

perf experiment goal Δ mean % Δ mean % CI trials links
tcp_syslog_to_blackhole ingress throughput +0.26 [+0.20, +0.32] 1 Logs
quality_gate_idle memory utilization +0.25 [+0.22, +0.29] 1 Logs bounds checks dashboard
file_tree memory utilization +0.18 [+0.04, +0.31] 1 Logs
file_to_blackhole_0ms_latency_http1 egress throughput +0.05 [-0.81, +0.91] 1 Logs
file_to_blackhole_500ms_latency egress throughput +0.03 [-0.75, +0.81] 1 Logs
file_to_blackhole_0ms_latency egress throughput +0.01 [-0.88, +0.91] 1 Logs
file_to_blackhole_1000ms_latency_linear_load egress throughput +0.01 [-0.46, +0.48] 1 Logs
file_to_blackhole_0ms_latency_http2 egress throughput -0.00 [-0.83, +0.83] 1 Logs
tcp_dd_logs_filter_exclude ingress throughput -0.00 [-0.01, +0.01] 1 Logs
uds_dogstatsd_to_api ingress throughput -0.01 [-0.13, +0.11] 1 Logs
file_to_blackhole_100ms_latency egress throughput -0.04 [-0.69, +0.61] 1 Logs
file_to_blackhole_300ms_latency egress throughput -0.05 [-0.68, +0.59] 1 Logs
file_to_blackhole_1000ms_latency egress throughput -0.05 [-0.84, +0.73] 1 Logs
quality_gate_idle_all_features memory utilization -0.46 [-0.54, -0.38] 1 Logs bounds checks dashboard
quality_gate_logs % cpu utilization -0.82 [-4.02, +2.38] 1 Logs
uds_dogstatsd_to_api_cpu % cpu utilization -1.01 [-1.69, -0.33] 1 Logs

Bounds Checks: ❌ Failed

perf experiment bounds_check_name replicates_passed links
file_to_blackhole_0ms_latency_http2 lost_bytes 7/10
file_to_blackhole_0ms_latency lost_bytes 10/10
file_to_blackhole_0ms_latency memory_usage 10/10
file_to_blackhole_0ms_latency_http1 lost_bytes 10/10
file_to_blackhole_0ms_latency_http1 memory_usage 10/10
file_to_blackhole_0ms_latency_http2 memory_usage 10/10
file_to_blackhole_1000ms_latency memory_usage 10/10
file_to_blackhole_1000ms_latency_linear_load memory_usage 10/10
file_to_blackhole_100ms_latency lost_bytes 10/10
file_to_blackhole_100ms_latency memory_usage 10/10
file_to_blackhole_300ms_latency lost_bytes 10/10
file_to_blackhole_300ms_latency memory_usage 10/10
file_to_blackhole_500ms_latency lost_bytes 10/10
file_to_blackhole_500ms_latency memory_usage 10/10
quality_gate_idle memory_usage 10/10 bounds checks dashboard
quality_gate_idle_all_features memory_usage 10/10 bounds checks dashboard
quality_gate_logs lost_bytes 10/10
quality_gate_logs memory_usage 10/10

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, bounds check memory_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 lost_bytes: 10/10 replicas passed. Gate passed.
  • quality_gate_idle_all_features, bounds check memory_usage: 10/10 replicas passed. Gate passed.

@grantseltzer grantseltzer changed the title Add config for inlined functions in program counter resolution for Go DI Add config for inlined functions in program counter resolution for Go DI, remove DWARF caching Dec 13, 2024
pkg/dynamicinstrumentation/ditypes/config.go Outdated Show resolved Hide resolved
pkg/dynamicinstrumentation/di.go Outdated Show resolved Hide resolved
@github-actions github-actions bot added long review PR is complex, plan time to review it and removed medium review PR review might take time labels Dec 21, 2024
@grantseltzer grantseltzer changed the title Add config for inlined functions in program counter resolution for Go DI, remove DWARF caching Remove inlined function PC resolution for stack traces, remove DWARF caching Dec 24, 2024
@agent-platform-auto-pr
Copy link
Contributor

agent-platform-auto-pr bot commented Dec 24, 2024

[Fast Unit Tests Report]

On pipeline 51775098 (CI Visibility). The following jobs did not run any unit tests:

Jobs:
  • tests_windows-x64

If you modified Go files and expected unit tests to run in these jobs, please double check the job logs. If you think tests should have been executed reach out to #agent-devx-help

@grantseltzer
Copy link
Member Author

I've removed inline function resolution altogether.

@grantseltzer grantseltzer force-pushed the grantseltzer/DEBUG-3207-add-config-for-inlined-functions-in-program-counter-resolution branch from 5c42da2 to 580cbf7 Compare December 24, 2024 17:04
@agent-platform-auto-pr
Copy link
Contributor

Uncompressed package size comparison

Comparison with ancestor f93202d5b1c2065d3b9ab0b50ba833c30467466c

Diff per package
package diff status size ancestor threshold
datadog-dogstatsd-amd64-deb 0.00MB 78.57MB 78.57MB 10.00MB
datadog-dogstatsd-x86_64-rpm 0.00MB 78.65MB 78.65MB 10.00MB
datadog-dogstatsd-x86_64-suse 0.00MB 78.65MB 78.65MB 10.00MB
datadog-dogstatsd-arm64-deb 0.00MB 55.77MB 55.77MB 10.00MB
datadog-heroku-agent-amd64-deb 0.00MB 505.17MB 505.17MB 70.00MB
datadog-iot-agent-amd64-deb 0.00MB 113.34MB 113.34MB 10.00MB
datadog-iot-agent-x86_64-rpm 0.00MB 113.41MB 113.41MB 10.00MB
datadog-iot-agent-x86_64-suse 0.00MB 113.41MB 113.41MB 10.00MB
datadog-iot-agent-arm64-deb 0.00MB 108.81MB 108.81MB 10.00MB
datadog-iot-agent-aarch64-rpm 0.00MB 108.88MB 108.88MB 10.00MB
datadog-agent-amd64-deb -0.00MB 1190.77MB 1190.77MB 140.00MB
datadog-agent-x86_64-rpm -0.00MB 1200.06MB 1200.06MB 140.00MB
datadog-agent-x86_64-suse -0.00MB 1200.06MB 1200.06MB 140.00MB
datadog-agent-arm64-deb -0.01MB 935.06MB 935.06MB 140.00MB
datadog-agent-aarch64-rpm -0.01MB 944.33MB 944.33MB 140.00MB

Decision

✅ Passed

Copy link
Collaborator

@cimi cimi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍 Thanks!

@grantseltzer grantseltzer requested review from brycekahle and removed request for brycekahle December 24, 2024 20:31
@grantseltzer
Copy link
Member Author

/merge

@dd-devflow
Copy link

dd-devflow bot commented Dec 24, 2024

Devflow running: /merge

View all feedbacks in Devflow UI.


2024-12-24 20:32:11 UTC ℹ️ MergeQueue: pull request added to the queue

The median merge time in main is 34m.


2024-12-24 21:27:16 UTC ℹ️ MergeQueue: This merge request was merged

@dd-mergequeue dd-mergequeue bot merged commit 272716f into main Dec 24, 2024
298 checks passed
@dd-mergequeue dd-mergequeue bot deleted the grantseltzer/DEBUG-3207-add-config-for-inlined-functions-in-program-counter-resolution branch December 24, 2024 21:27
@github-actions github-actions bot added this to the 7.62.0 milestone Dec 24, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
changelog/no-changelog long review PR is complex, plan time to review it qa/done QA done before merge and regressions are covered by tests team/dynamic-instrumentation Dynamic Instrumentation
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants