benchmark-prs.yml

name: PR Benchmarks

on: pull_request

env:
  CARGO_INCREMENTAL: '0'
  RUST_BACKTRACE: 1
  CLIENT_DATA_PATH: /home/runner/.local/share/safe/client
  NODE_DATA_PATH: /home/runner/.local/share/safe/node

jobs:
  benchmark-cli:
    name: Compare sn_cli benchmarks to main
    # right now only ubuntu, running on multiple systems would require many pushes...\
    # perhaps this can be done with one consolidation action in the future, pulling down all results and pushing
    # once to the branch..
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - uses: dtolnay/rust-toolchain@stable
        with:
          components: rustfmt, clippy

      - uses: Swatinem/rust-cache@v2
        continue-on-error: true

      ########################
      ### Setup            ###
      ########################
      - run: cargo install cargo-criterion

      - name: install ripgrep
        run: sudo apt-get -y install ripgrep

      - name: Download 95mb file to be uploaded with the safe client
        shell: bash
        run: wget https://sn-node.s3.eu-west-2.amazonaws.com/the-test-data.zip

      # As normal user won't care much about initial client startup,
      # but be more alerted on communication speed during transmission.
      # Meanwhile the criterion testing code includes the client startup as well,
      # it will be better to execute bench test with `local-discovery`,
      # to make the measurement results reflect speed improvement or regression more accurately.
      - name: Build sn bins
        run: cargo build --release --bins --features local-discovery
        timeout-minutes: 30

      - name: Start a local network
        uses: maidsafe/sn-local-testnet-action@main
        env:
          SN_LOG: "all"
        with:
          action: start
          interval: 2000
          node-path: target/release/safenode
          faucet-path: target/release/faucet
          platform: ubuntu-latest
          build: true

      - name: Check SAFE_PEERS was set
        shell: bash
        run: echo "The SAFE_PEERS variable has been set to $SAFE_PEERS"

      #########################
      ### Upload large file ###
      #########################

      - name: Fund cli wallet
        shell: bash
        run: target/release/safe --log-output-dest=data-dir wallet get-faucet 127.0.0.1:8000
        env:
          SN_LOG: "all"

      - name: Start a client instance to compare memory usage
        shell: bash
        run: target/release/safe --log-output-dest=data-dir files upload the-test-data.zip --retry-strategy quick
        env:
          SN_LOG: "all"

      - name: Cleanup uploaded_files folder to avoid pollute download benchmark
        shell: bash
        run: rm -rf $CLIENT_DATA_PATH/uploaded_files

      ###########################
      ### Client Mem Analysis ###
      ###########################

      - name: Check client memory usage
        shell: bash
        run: |
          client_peak_mem_limit_mb="1024" # mb
          client_avg_mem_limit_mb="512" # mb

          peak_mem_usage=$(
            rg '"memory_used_mb":[^,]*' $CLIENT_DATA_PATH/logs --glob safe.* -o --no-line-number --no-filename |
            awk -F':' '/"memory_used_mb":/{print $2}' |
            sort -n |
            tail -n 1
          )
          echo "Peak memory usage: $peak_mem_usage MB"
          if (( $(echo "$peak_mem_usage > $client_peak_mem_limit_mb" | bc -l) )); then
            echo "Client peak memory usage exceeded threshold: $client_peak_mem_limit_mb MB"
            exit 1
          fi

          total_mem=$(
            rg '"memory_used_mb":[^,]*' $CLIENT_DATA_PATH/logs --glob safe.* -o --no-line-number --no-filename |
            awk -F':' '/"memory_used_mb":/ {sum += $2} END {printf "%.0f\n", sum}'
          )
          num_of_times=$(
            rg "\"memory_used_mb\"" $CLIENT_DATA_PATH/logs --glob safe.* -c --stats |
            rg "(\d+) matches" |
            rg "\d+" -o
          )
          echo "num_of_times: $num_of_times"
          echo "Total memory is: $total_mem"
          average_mem=$(($total_mem/$(($num_of_times))))
          echo "Average memory is: $average_mem"

          if (( $(echo "$average_mem > $client_avg_mem_limit_mb" | bc -l) )); then
            echo "Client average memory usage exceeded threshold: $client_avg_mem_limit_mb MB"
            exit 1
          fi
          # Write the client memory usage to a file
          echo '[
              {
                  "name": "client-peak-memory-usage-during-upload",
                  "value": '$peak_mem_usage',
                  "unit": "MB"
              },
              {
                  "name": "client-average-memory-usage-during-upload",
                  "value": '$average_mem',
                  "unit": "MB"
              }
          ]' > client_memory_usage.json

      - name: check client_memory_usage.json
        shell: bash
        run: cat client_memory_usage.json

      - name: Alert for client memory usage
        uses: benchmark-action/github-action-benchmark@v1
        with:
          name: 'Memory Usage of Client during uploading large file'
          tool: 'customSmallerIsBetter'
          output-file-path: client_memory_usage.json
          # Where the previous data file is stored
          external-data-json-path: ./cache/client-mem-usage.json
          # Workflow will fail when an alert happens
          fail-on-alert: true
          # GitHub API token to make a commit comment
          github-token: ${{ secrets.GITHUB_TOKEN }}
          # Enable alert commit comment
          comment-on-alert: true
          # 200% regression will result in alert
          alert-threshold: '200%'
          # Enable Job Summary for PRs
          summary-always: true

      ########################
      ### Benchmark        ###
      ########################
      - name: Bench `safe` cli
        shell: bash
        # Criterion outputs the actual bench results to stderr "2>&1 tee output.txt" takes stderr,
        # passes to tee which displays it in the terminal and writes to output.txt
        run: |
          cargo criterion --features=local-discovery --message-format=json 2>&1 -p sn_cli | tee -a output.txt
          cat output.txt | rg benchmark-complete | jq -s 'map({
          name: (.id | split("/"))[-1],
          unit: "MiB/s",
          value: ((if .throughput[0].unit == "KiB/s" then (.throughput[0].per_iteration / (1024*1024*1024)) else (.throughput[0].per_iteration / (1024*1024)) end) / (.mean.estimate / 1e9))
          })' > files-benchmark.json
        timeout-minutes: 15

      - name: Confirming the number of files uploaded and downloaded during the benchmark test
        shell: bash
        run: |
          ls -l $CLIENT_DATA_PATH
          ls -l $CLIENT_DATA_PATH/uploaded_files
          ls -l $CLIENT_DATA_PATH/safe_files

      - name: Store benchmark result
        uses: benchmark-action/github-action-benchmark@v1
        with:
          # What benchmark tool the output.txt came from
          tool: 'customBiggerIsBetter'
          output-file-path: files-benchmark.json
          # Where the previous data file is stored
          external-data-json-path: ./cache/benchmark-data.json
          # Workflow will fail when an alert happens
          fail-on-alert: true
          # GitHub API token to make a commit comment
          github-token: ${{ secrets.GITHUB_TOKEN }}
          # Enable alert commit comment
          comment-on-alert: true
          # 200% regression will result in alert
          alert-threshold: '200%'
          # Enable Job Summary for PRs
          summary-always: true

      - name: Start a client to carry out download to output the logs
        shell: bash
        run: target/release/safe --log-output-dest=data-dir files download --retry-strategy quick

      - name: Start a client to simulate criterion upload
        shell: bash
        run: |
          ls -l target/release
          target/release/safe --log-output-dest=data-dir files upload target/release/faucet --retry-strategy quick

      #########################
      ### Stop Network      ###
      #########################

      - name: Stop the local network
        if: always()
        uses: maidsafe/sn-local-testnet-action@main
        with:
          action: stop
          log_file_prefix: safe_test_logs_benchmark
          platform: ubuntu-latest
          build: true

      - name: Upload Faucet folder
        uses: actions/upload-artifact@main
        with:
          name: faucet_folder
          path: /home/runner/.local/share/safe/test_faucet
        continue-on-error: true
        if: always()

      #########################
      ### Node Mem Analysis ###
      #########################

      # The large file uploaded will increase node's peak mem usage a lot
      - name: Check node memory usage
        shell: bash
        run: |
          node_peak_mem_limit_mb="250" # mb
          peak_mem_usage=$(
            rg '"memory_used_mb":[^,]*' $NODE_DATA_PATH/*/logs/* -o --no-line-number --no-filename |
            awk -F':' '/"memory_used_mb":/{print $2}' |
            sort -n |
            tail -n 1
          )

          echo "Memory usage: $peak_mem_usage MB"
          if (( $(echo "$peak_mem_usage > $node_peak_mem_limit_mb" | bc -l) )); then
            echo "Node memory usage exceeded threshold: $peak_mem_usage MB"
            exit 1
          fi
          # Write the node memory usage to a file
          echo '[
              {
                  "name": "node-memory-usage-through-safe-benchmark",
                  "value": '$peak_mem_usage',
                  "unit": "MB"
              }
          ]' > node_memory_usage.json

      - name: check node_memory_usage.json
        shell: bash
        run: cat node_memory_usage.json

      - name: Alert for node memory usage
        uses: benchmark-action/github-action-benchmark@v1
        with:
          tool: 'customSmallerIsBetter'
          output-file-path: node_memory_usage.json
          # Where the previous data file is stored
          external-data-json-path: ./cache/node-mem-usage.json
          # Workflow will fail when an alert happens
          fail-on-alert: true
          # GitHub API token to make a commit comment
          github-token: ${{ secrets.GITHUB_TOKEN }}
          # Enable alert commit comment
          comment-on-alert: true
          # Comment on the PR
          comment-always: true
          # 200% regression will result in alert
          alert-threshold: '200%'
          # Enable Job Summary for PRs
          summary-always: true

      ###########################################
      ### Swarm_driver handling time Analysis ###
      ###########################################

      - name: Check swarm_driver handling time
        shell: bash
        run: |
          num_of_times=$(
            rg "SwarmCmd handled in [0-9.]+ms:" $NODE_DATA_PATH/*/logs/* --glob safe.* -c --stats |
            rg "(\d+) matches" |
            rg "\d+" -o
          )
          echo "Number of long cmd handling times: $num_of_times"
          total_long_handling_ms=$(
            rg "SwarmCmd handled in [0-9.]+ms:" $NODE_DATA_PATH/*/logs/* --glob safe.* -o --no-line-number --no-filename |
            awk -F' |ms:' '{sum += $4} END {printf "%.0f\n", sum}'
          )
          echo "Total cmd long handling time is: $total_long_handling_ms ms"
          average_handling_ms=$(($total_long_handling_ms/$(($num_of_times))))
          echo "Average cmd long handling time is: $average_handling_ms ms"
          total_long_handling=$(($total_long_handling_ms))
          total_num_of_times=$(($num_of_times))
          num_of_times=$(
            rg "SwarmEvent handled in [0-9.]+ms:" $NODE_DATA_PATH/*/logs/* --glob safe.* -c --stats |
            rg "(\d+) matches" |
            rg "\d+" -o
          )
          echo "Number of long event handling times: $num_of_times"
          total_long_handling_ms=$(
            rg "SwarmEvent handled in [0-9.]+ms:" $NODE_DATA_PATH/*/logs/* --glob safe.* -o --no-line-number --no-filename |
            awk -F' |ms:' '{sum += $4} END {printf "%.0f\n", sum}'
          )
          echo "Total event long handling time is: $total_long_handling_ms ms"
          average_handling_ms=$(($total_long_handling_ms/$(($num_of_times))))
          echo "Average event long handling time is: $average_handling_ms ms"
          total_long_handling=$(($total_long_handling_ms+$total_long_handling))
          total_num_of_times=$(($num_of_times+$total_num_of_times))
          average_handling_ms=$(($total_long_handling/$(($total_num_of_times))))
          echo "Total swarm_driver long handling times is: $total_num_of_times"
          echo "Total swarm_driver long handling duration is: $total_long_handling ms"
          echo "Total average swarm_driver long handling duration is: $average_handling_ms ms"
          total_num_of_times_limit_hits="20000" # hits
          total_long_handling_limit_ms="170000" # ms
          average_handling_limit_ms="20" # ms
          if (( $(echo "$total_num_of_times > $total_num_of_times_limit_hits" | bc -l) )); then
            echo "Swarm_driver long handling times exceeded threshold: $total_num_of_times hits"
            exit 1
          fi
          if (( $(echo "$total_long_handling > $total_long_handling_limit_ms" | bc -l) )); then
            echo "Swarm_driver total long handling duration exceeded threshold: $total_long_handling ms"
            exit 1
          fi
          if (( $(echo "$average_handling_ms > $average_handling_limit_ms" | bc -l) )); then
            echo "Swarm_driver average long handling time exceeded threshold: $average_handling_ms ms"
            exit 1
          fi

          # Write the node memory usage to a file
          echo '[
              {
                  "name": "swarm_driver long handling times",
                  "value": '$total_num_of_times',
                  "unit": "hits"
              },
              {
                  "name": "swarm_driver long handling total_time",
                  "value": '$total_long_handling',
                  "unit": "ms"
              },
              {
                  "name": "swarm_driver average long handling time",
                  "value": '$average_handling_ms',
                  "unit": "ms"
              }
          ]' > swarm_driver_long_handlings.json

      - name: check swarm_driver_long_handlings.json
        shell: bash
        run: cat swarm_driver_long_handlings.json

      - name: Alert for swarm_driver long handlings
        uses: benchmark-action/github-action-benchmark@v1
        with:
          tool: 'customSmallerIsBetter'
          output-file-path: swarm_driver_long_handlings.json
          # Where the previous data file is stored
          external-data-json-path: ./cache/swarm_driver_long_handlings.json
          # Workflow will fail when an alert happens
          fail-on-alert: true
          # GitHub API token to make a commit comment
          github-token: ${{ secrets.GITHUB_TOKEN }}
          # Enable alert commit comment
          comment-on-alert: true
          # Comment on the PR
          comment-always: true
          # 200% regression will result in alert
          alert-threshold: '200%'
          # Enable Job Summary for PRs
          summary-always: true

  benchmark-cash:
    name: Compare sn_transfer benchmarks to main
    # right now only ubuntu, running on multiple systems would require many pushes...\
    # perhaps this can be done with one consolidation action in the future, pulling down all results and pushing
    # once to the branch..
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - uses: dtolnay/rust-toolchain@stable
        with:
          components: rustfmt, clippy

      - uses: Swatinem/rust-cache@v2
        continue-on-error: true

      ########################
      ### Setup            ###
      ########################
      - run: cargo install cargo-criterion

      - name: install ripgrep
        run: sudo apt-get -y install ripgrep



      ########################
      ### Benchmark        ###
      ########################
      - name: Bench `sn_transfers`
        shell: bash
        # Criterion outputs the actual bench results to stderr "2>&1 tee output.txt" takes stderr,
        # passes to tee which displays it in the terminal and writes to output.txt
        run: |
          cargo criterion --message-format=json 2>&1 -p sn_transfers | tee -a output.txt
          cat output.txt