perf(l1,l2): use the new load test for the CI scripts (lambdaclass#2467)

Changes:
- Flamegraph Watcher srcript now:
  - Uses the new load test.
  - Fails if any line fails (e.g. the load test binary panics).
- CI:
  - The flamegraphs are now updated on push to main again.
- Compilation and running is separated to delete the "while not
compiled" polling.
- Reth version is pinned so it does not rely on 2024 features and can be
compiled again.
- Load test:
  - `make` targets now run in release mode.
- now waits until all transactions are included before exciting. There's
a flag to set a timeout.
- All ethrex_l2 references are deleted from CI and the watcher.


Closes lambdaclass#2466

Author: Arkenan

.github/scripts/flamegraph_watcher.sh

@@ -1,31 +1,28 @@
 #!/bin/bash
+set -e
 
-# This script sends 171 * <iterations> transactions to a test account, per defined private key
-# then polls the account balance until the expected balance has been reached
-# and then kills the process. It also measures the elapsed time of the test and
-# outputs it to Github Action's outputs.
-iterations=3500
-value=1
-account=0x33c6b73432B3aeA0C1725E415CC40D04908B85fd
-end_val=$((171 * $iterations * $value))
+# This script runs a load test and then kills the node under test. The load test sends a 
+# transaction from each rich account to a random one, so we can check their nonce to
+# determine that the load test finished.
+#
+# Usage:
+# ./flamegraph_watcher.sh
+# Requires a PROGRAM variable to be set (e.g. ethrex). This $PROGRAM will be killed when the
+# load test finishes. Must be run from the context of the repo root.
 
-start_time=$(date +%s)
-ethrex_l2 test load --path /home/runner/work/ethrex/ethrex/test_data/private_keys.txt -i $iterations -v --value $value --to $account >/dev/null
+# TODO(#2486): Move this to a cached build outside.
+echo "Building load test"
+cargo build --release --manifest-path ./cmd/load_test/Cargo.toml
 
-output=$(ethrex_l2 info -b -a $account --wei 2>&1)
-echo "balance: $output"
-while [[ $output -lt $end_val ]]; do
-    sleep 2
-    output=$(ethrex_l2 info -b -a $account --wei 2>&1)
-    echo "balance: $output"
-done
+echo "Starting load test"
+start_time=$(date +%s)
+RUST_BACKTRACE=1 ./target/release/load_test -k ./test_data/private_keys.txt -t eth-transfers -N 1000 -n http://localhost:1729 -w 5 >/dev/null
 end_time=$(date +%s)
-elapsed=$((end_time - start_time))
 
+elapsed=$((end_time - start_time))
 minutes=$((elapsed / 60))
 seconds=$((elapsed % 60))
-output=$(ethrex_l2 info -b -a $account --wei 2>&1)
-echo "Balance of $output reached in $minutes min $seconds s, killing process"
+echo "All load test transactions included in $minutes min $seconds s, killing node process."
 
 echo killing "$PROGRAM"
 sudo pkill "$PROGRAM"

.github/workflows/main_flamegraph_report.yaml

@@ -6,9 +6,8 @@ permissions:
   id-token: write
 
 on:
-  # re-enable this when fixing the workflow
-  # push:
-  #   branches: ["main"]
+  push:
+    branches: ["main"]
   workflow_dispatch:
 
 env:
@@ -183,13 +182,6 @@ jobs:
             ${{ env.HOME }}/.cargo/bin/inferno-*
           key: ${{ runner.os }}-extra-binaries
 
-      - name: Cache ethrex_l2
-        id: cache-ethrex-l2
-        uses: actions/cache@v4
-        with:
-          path: ${{ env.HOME }}/.cargo/bin/ethrex_l2
-          key: ${{ runner.os }}-ethrex-l2-${{ hashFiles('cmd/ethrex_l2/Cargo.lock') }}
-
       - name: Change perf settings
         run: |
           sudo sysctl kernel.perf_event_paranoid=-1
@@ -236,30 +228,18 @@ jobs:
             echo "$HOME/.cargo/bin/inferno-collapse-perf" already found
           fi
 
-      - name: Install ethrex_l2 cli
-        run: |
-          if [ -f "$HOME/.cargo/bin/ethrex_l2" ]; then
-            echo "$HOME/.cargo/bin/ethrex_l2" already found
-          else
-            cargo install --force --path cmd/ethrex_l2
-          fi
-          ethrex_l2 config create default --default
-          ethrex_l2 config set default
-
       # By default ethrex uses revm as evm backend.
       - id: generate-flamegraph-ethrex
         name: Generate Flamegraph data for Ethrex
         shell: bash
         run: |
           rm -rf target/release/ethrex
+          cargo build --release --bin ethrex --features dev
           CARGO_PROFILE_RELEASE_DEBUG=true cargo flamegraph -c "record -o perf.data -F997 --call-graph dwarf,16384 -g" \
-          --bin ethrex --features dev -- --dev --network /home/runner/work/ethrex/ethrex/test_data/genesis-l2.json --http.port 1729 >/dev/null &
-          while [ ! -x "./target/release/ethrex" ]; do
-            echo "Waiting for ethrex binary to be ready..."
-            sleep 2
-          done
+          --bin ethrex --release --features dev -- \
+          --dev --network /home/runner/work/ethrex/ethrex/test_data/genesis-l2-ci.json --http.port 1729 >/dev/null &
           sleep 10
-          echo "Compilation finished. Executing load test..."
+          echo "Executing load test..."
           bash /home/runner/work/ethrex/ethrex/.github/scripts/flamegraph_watcher.sh &&
           echo "Load test finished"
 
@@ -293,12 +273,13 @@ jobs:
         with:
           toolchain: ${{ env.RUST_RETH_VERSION }}
 
+      # We need a reth version that requires a rustc version <= 1.82.0
       - name: Checkout reth
         uses: actions/checkout@v4
         with:
           repository: paradigmxyz/reth
           path: "reth"
-          ref: main
+          ref: b2ead06d1d0804101de0d1eb3a070e08d8eab857
 
       - name: Caching
         uses: Swatinem/rust-cache@v2
@@ -316,13 +297,6 @@ jobs:
             ${{ env.HOME }}/.cargo/bin/inferno-*
           key: ${{ runner.os }}-extra-binaries
 
-      - name: Cache ethrex_l2
-        id: cache-ethrex-l2
-        uses: actions/cache@v4
-        with:
-          path: ${{ env.HOME }}/.cargo/bin/ethrex_l2
-          key: ${{ runner.os }}-ethrex-l2-${{ hashFiles('cmd/ethrex_l2/Cargo.lock') }}
-
       - name: Change perf settings
         run: |
           sudo sysctl kernel.perf_event_paranoid=-1
@@ -369,35 +343,22 @@ jobs:
             echo "$HOME/.cargo/bin/inferno-collapse-perf" already found
           fi
 
-      - name: Install ethrex_l2 cli
-        run: |
-          if [ -f "$HOME/.cargo/bin/ethrex_l2" ]; then
-            echo "$HOME/.cargo/bin/ethrex_l2" already found
-          else
-            cargo install --force --path cmd/ethrex_l2
-          fi
-          ethrex_l2 config create default --default
-          ethrex_l2 config set default
-
       - id: generate-flamegraph-reth
         name: Build and test reth
         shell: bash
         # --dev.block-time 1000ms set to 1000ms to match ethrex block generation time
         run: |
           cd ./reth
           rm -rf target/profiling/reth
-          CARGO_PROFILE_RELEASE_DEBUG=true cargo flamegraph -c "record -o perf.data -F997 --call-graph dwarf,16384 -g" \
-          --bin reth --profile profiling -- node --chain /home/runner/work/ethrex/ethrex/test_data/genesis-load-test.json --dev \
+          cargo build --bin reth --profile profiling
+          CARGO_PROFILE_RELEASE_DEBUG=true cargo flamegraph -c "record -o perf.data -F997 --call-graph dwarf,16384 -g" --bin reth --profile profiling -- \
+          node --chain /home/runner/work/ethrex/ethrex/test_data/genesis-l2-ci.json --dev \
           --dev.block-time 1000ms --http.port 1729  --txpool.max-pending-txns 100000000 --txpool.max-new-txns 1000000000 \
           --txpool.pending-max-count 100000000 --txpool.pending-max-size 10000000000 --txpool.basefee-max-count 100000000000 \
           --txpool.basefee-max-size 1000000000000 --txpool.queued-max-count 1000000000 >/dev/null &
-          while [ ! -x "./target/profiling/reth" ]; do
-            echo "Waiting for reth binary to be ready..."
-            sleep 10
-          done
           sleep 30
           echo "Executing load test..."
-          bash /home/runner/work/ethrex/ethrex/.github/scripts/flamegraph_watcher.sh &&
+          (cd /home/runner/work/ethrex/ethrex; ./.github/scripts/flamegraph_watcher.sh)
           echo "Load test finished"
 
       - name: Generate SVG

Makefile

@@ -161,16 +161,16 @@ start-node-with-flamegraph: rm-test-db ## 🚀🔥 Starts an ethrex client used
 	--datadir test_ethrex
 
 load-test: ## 🚧 Runs a load-test. Run make start-node-with-flamegraph and in a new terminal make load-node
-	cargo run --manifest-path ./cmd/load_test/Cargo.toml -- -k ./test_data/private_keys.txt -t eth-transfers
+	cargo run --release --manifest-path ./cmd/load_test/Cargo.toml -- -k ./test_data/private_keys.txt -t eth-transfers
 
 load-test-erc20:
-	cargo run --manifest-path ./cmd/load_test/Cargo.toml -- -k ./test_data/private_keys.txt -t erc20
+	cargo run --release --manifest-path ./cmd/load_test/Cargo.toml -- -k ./test_data/private_keys.txt -t erc20
 
 load-test-fibonacci:
-	cargo run --manifest-path ./cmd/load_test/Cargo.toml -- -k ./test_data/private_keys.txt -t fibonacci
+	cargo run --release --manifest-path ./cmd/load_test/Cargo.toml -- -k ./test_data/private_keys.txt -t fibonacci
 
 load-test-io:
-	cargo run --manifest-path ./cmd/load_test/Cargo.toml -- -k ./test_data/private_keys.txt -t io-heavy
+	cargo run --release --manifest-path ./cmd/load_test/Cargo.toml -- -k ./test_data/private_keys.txt -t io-heavy
 
 rm-test-db:  ## 🛑 Removes the DB used by the ethrex client used for testing
 	sudo cargo run --release --bin ethrex -- removedb --force --datadir test_ethrex

cmd/load_test/README.md

@@ -10,8 +10,9 @@ Usage: load_test [OPTIONS] --pkeys <PKEYS>
 Options:
   -n, --node <NODE>            URL of the node being tested. [default: http://localhost:8545]
   -k, --pkeys <PKEYS>          Path to the file containing private keys.
-  -t, --test-type <TEST_TYPE>  Type of test to run. Can be eth_transfers or erc20. [default: erc20] [possible values: eth-transfers, erc20]
+  -t, --test-type <TEST_TYPE>  Type of test to run. Can be eth_transfers or erc20. [default: erc20] [possible values: eth-transfers, erc20, fibonacci, io-heavy]
   -N, --tx-amount <TX_AMOUNT>  Number of transactions to send for each account. [default: 1000]
+  -w, --wait <WAIT>            Timeout to wait for all transactions to be included. If 0 is specified, wait indefinitely. [default: 0]
   -h, --help                   Print help
 ```
 

cmd/load_test/src/main.rs

@@ -57,6 +57,15 @@ struct Cli {
         help = "Number of transactions to send for each account."
     )]
     tx_amount: u64,
+
+    // Amount of minutes to wait before exiting. If the value is 0, the program will wait indefinitely. If not present, the program will not wait for transactions to be included in blocks.
+    #[arg(
+        long,
+        short = 'w',
+        default_value_t = 0,
+        help = "Timeout to wait for all transactions to be included. If 0 is specified, wait indefinitely."
+    )]
+    wait: u64,
 }
 
 #[derive(ValueEnum, Clone, Debug)] // Derive ValueEnum for TestType
@@ -264,6 +273,49 @@ async fn load_test(
     Ok(())
 }
 
+// Waits until the nonce of each account has reached the tx_amount.
+async fn wait_until_all_included(
+    client: EthClient,
+    wait: Option<Duration>,
+    accounts: &[Account],
+    tx_amount: u64,
+) -> Result<(), String> {
+    let start_time = tokio::time::Instant::now();
+
+    for (pk, _) in accounts {
+        let pk = *pk;
+        let client = client.clone();
+        let src = address_from_pub_key(pk);
+        let encoded_src: String = src.encode_hex();
+        loop {
+            let elapsed = start_time.elapsed();
+            let nonce = client.get_nonce(src, BlockByNumber::Latest).await.unwrap();
+            if nonce >= tx_amount {
+                println!(
+                    "All transactions sent from {} have been included in blocks. Nonce: {}",
+                    encoded_src, nonce
+                );
+                break;
+            } else {
+                println!(
+                    "Waiting for transactions to be included from {}. Nonce: {}. Needs: {}. Percentage: {:2}%. Elapsed time: {}s.",
+                    encoded_src, nonce, tx_amount, (nonce as f64 / tx_amount as f64) * 100.0, elapsed.as_secs()
+                );
+            }
+
+            if let Some(wait) = wait {
+                if elapsed > wait {
+                    return Err("Timeout reached for transactions to be included".to_string());
+                }
+            }
+
+            sleep(Duration::from_secs(5)).await;
+        }
+    }
+
+    Ok(())
+}
+
 fn parse_pk_file(path: &Path) -> eyre::Result<Vec<Account>> {
     let pkeys_content = fs::read_to_string(path).expect("Unable to read private keys file");
     let accounts: Vec<Account> = pkeys_content
@@ -335,6 +387,12 @@ async fn main() {
         }
     };
 
+    println!(
+        "Starting load test with {} transactions per account",
+        cli.tx_amount
+    );
+    let time_now = tokio::time::Instant::now();
+
     load_test(
         cli.tx_amount,
         &accounts,
@@ -344,4 +402,22 @@ async fn main() {
     )
     .await
     .expect("Failed to load test");
+
+    let wait_time = if cli.wait > 0 {
+        Some(Duration::from_secs(cli.wait * 60))
+    } else {
+        None
+    };
+
+    println!("Waiting for all transactions to be included in blocks...");
+    wait_until_all_included(client, wait_time, &accounts, cli.tx_amount)
+        .await
+        .unwrap();
+
+    let elapsed_time = time_now.elapsed();
+
+    println!(
+        "Load test finished. Elapsed time: {} seconds",
+        elapsed_time.as_secs()
+    );
 }