generated: '2026-07-19'
method: searched
source: https://github.com/intel/gprofiler/blob/master/README.md
name: gProfiler
description: >-
  gProfiler is Granulate's (Intel) open-source system-wide continuous profiler.
  It runs as a single agent/executable (or container) on each host, combining
  multiple sampling profilers (perf, py-spy, async-profiler, rbspy, PHP/.NET/
  Node.js modes) into one flamegraph, optionally uploading results to the
  Granulate Performance Studio. The binary is catalogued in packages/.
license: Apache-2.0
install:
  - method: executable
    command: >-
      curl -sSL https://github.com/intel/gprofiler/releases/latest/download/gprofiler -o gprofiler && chmod +x gprofiler
  - method: docker
    command: docker pull granulate/gprofiler:latest
usage: sudo ./gprofiler [options]
flags:
  runtime_modes:
    - "--python-mode"
    - "--java-mode"
    - "--ruby-mode"
    - "--php-mode"
    - "--nodejs-mode"
    - "--dotnet-mode"
    - "--perf-mode"
    - "--no-java"
    - "--no-python"
    - "--no-ruby"
    - "--no-php"
    - "--no-dotnet"
  execution:
    - "--continuous"
    - "--profiling-duration"
    - "--profiling-frequency"
    - "--rotating-output"
    - "--output-dir"
    - "--no-flamegraph"
    - "--pids"
    - "--pid-file"
    - "--log-file"
  upload:
    - "--upload-results"
    - "--token"
    - "--service-name"
    - "--server-host"
  metrics:
    - "--enable-hw-metrics-collection"
    - "--disable-metrics-collection"
    - "--disable-metadata-collection"
    - "--perfspect-path"
key_flows:
  - name: Continuous profiling with upload to Performance Studio
    command: >-
      docker run --name granulate-gprofiler -d --restart=on-failure:10
      --pid=host --userns=host --privileged granulate/gprofiler:latest
      -cu --token="$TOKEN" --service-name="$SERVICE"
  - name: One-off local flamegraph
    command: sudo ./gprofiler -o /tmp/results --output-dir /tmp/results