Module profiling

This binary is used for profiling with valgrind and kcachegrind.

Requirements

valgrind and kcachegrind.

On Ubuntu you can install them with:

sudo apt install valgrind kcachegrind

NOTICE: valgrind executes the program you wan to profile and waits until it ends. Since the tracker is a service and does not end the profiling binary accepts an arguments with the duration you want to run the tracker, so that it terminates automatically after that period of time.

Run profiling

To run the profiling you have to:

1. Build and run the tracker for profiling.
2. Run the aquatic UDP load test tool to start collecting data in the tracker.

Build and run the tracker for profiling:

RUSTFLAGS='-g' cargo build --release --bin profiling \
  && export TORRUST_TRACKER_CONFIG_TOML_PATH="./share/default/config/tracker.udp.benchmarking.toml" \
  && valgrind \
    --tool=callgrind \
    --callgrind-out-file=callgrind.out \
    --collect-jumps=yes \
    --simulate-cache=yes \
    ./target/release/profiling 60

The output should be something like:

RUSTFLAGS='-g' cargo build --release --bin profiling \
   && export TORRUST_TRACKER_CONFIG_TOML_PATH="./share/default/config/tracker.udp.benchmarking.toml" \
   && valgrind \
     --tool=callgrind \
     --callgrind-out-file=callgrind.out \
     --collect-jumps=yes \
     --simulate-cache=yes \
     ./target/release/profiling 60

   Compiling torrust-tracker v3.0.0-alpha.12-develop (/home/developer/Documents/git/committer/me/github/torrust/torrust-tracker)
    Finished `release` profile [optimized + debuginfo] target(s) in 1m 15s
==122801== Callgrind, a call-graph generating cache profiler
==122801== Copyright (C) 2002-2017, and GNU GPL'd, by Josef Weidendorfer et al.
==122801== Using Valgrind-3.19.0 and LibVEX; rerun with -h for copyright info
==122801== Command: ./target/release/profiling 60
==122801==
--122801-- warning: L3 cache found, using its data for the LL simulation.
==122801== For interactive control, run 'callgrind_control -h'.
Loading configuration file: `./share/default/config/tracker.udp.benchmarking.toml` ...
Torrust successfully shutdown.
==122801==
==122801== Events    : Ir Dr Dw I1mr D1mr D1mw ILmr DLmr DLmw
==122801== Collected : 1160654816 278135882 247755311 24453652 12650490 16315690 10932 2481624 4832145
==122801==
==122801== I   refs:      1,160,654,816
==122801== I1  misses:       24,453,652
==122801== LLi misses:           10,932
==122801== I1  miss rate:          2.11%
==122801== LLi miss rate:          0.00%
==122801==
==122801== D   refs:        525,891,193  (278,135,882 rd + 247,755,311 wr)
==122801== D1  misses:       28,966,180  ( 12,650,490 rd +  16,315,690 wr)
==122801== LLd misses:        7,313,769  (  2,481,624 rd +   4,832,145 wr)
==122801== D1  miss rate:           5.5% (        4.5%   +         6.6%  )
==122801== LLd miss rate:           1.4% (        0.9%   +         2.0%  )
==122801==
==122801== LL refs:          53,419,832  ( 37,104,142 rd +  16,315,690 wr)
==122801== LL misses:         7,324,701  (  2,492,556 rd +   4,832,145 wr)
==122801== LL miss rate:            0.4% (        0.2%   +         2.0%  )

NOTICE: We are using an specific tracker configuration for profiling that removes all features except the UDP tracker and sets the logging level to error.

Build the aquatic UDP load test command:

cd /tmp
git clone git@github.com:greatest-ape/aquatic.git
cd aquatic
cargo build --profile=release-debug -p aquatic_udp_load_test
./target/release-debug/aquatic_udp_load_test -p > "load-test-config.toml"

Modify the “load-test-config.toml” file to change the UDP tracker port from 3000 to 6969.

Running the aquatic UDP load test command:

./target/release-debug/aquatic_udp_load_test -c "load-test-config.toml"

The output should be something like this:

Starting client with config: Config {
    server_address: 127.0.0.1:6969,
    log_level: Error,
    workers: 1,
    duration: 0,
    summarize_last: 0,
    extra_statistics: true,
    network: NetworkConfig {
        multiple_client_ipv4s: true,
        sockets_per_worker: 4,
        recv_buffer: 8000000,
    },
    requests: RequestConfig {
        number_of_torrents: 1000000,
        number_of_peers: 2000000,
        scrape_max_torrents: 10,
        announce_peers_wanted: 30,
        weight_connect: 50,
        weight_announce: 50,
        weight_scrape: 1,
        peer_seeder_probability: 0.75,
    },
}

Requests out: 45097.51/second
Responses in: 4212.70/second
  - Connect responses:  2098.15
  - Announce responses: 2074.95
  - Scrape responses:   39.59
  - Error responses:    0.00
Peers per announce response: 0.00
Announce responses per info hash:
  - p10: 1
  - p25: 1
  - p50: 1
  - p75: 2
  - p90: 3
  - p95: 4
  - p99: 6
  - p99.9: 8
  - p100: 10

After running the tracker for some seconds the tracker will automatically stop and valgrindwill write the file callgrind.out with the data.

You can now analyze the collected data with:

kcachegrind callgrind.out

Functions

run