WIP: Storage Benchmarking

yardstick_benchmark/__init__.py

@@ -1,6 +1,40 @@
 from yardstick_benchmark.model import Node, RemoteAction
 from pathlib import Path
+import os
+import tarfile
+from pathlib import Path
+
+def unarchive(src: Path, dest: Path = None):
+    """
+    Extract all .tar.gz archives in the specified source directory.
 
+    Parameters:
+    - src (Path): The path to the directory containing .tar.gz files.
+    - dest (Path, optional): The destination directory for extraction.
+      Defaults to the current working directory.
+    """
+    src = Path(src)
+    dest = Path(dest) if dest else Path.cwd()
+
+    if not src.is_dir():
+        raise FileNotFoundError(f"Source directory {src} does not exist.")
+
+    dest.mkdir(parents=True, exist_ok=True)
+    tar_files = src.glob("*.tar.gz")
+
+    for tar_file in tar_files:
+        try:
+            with tarfile.open(tar_file, "r:gz") as tar:
+                if tar_file.suffixes[-2:] == [".tar", ".gz"]:
+                    extract_name = tar_file.name.removesuffix(".tar.gz")
+                else:
+                    extract_name = tar_file.stem
+                extract_path = dest / extract_name
+                extract_path.mkdir(parents=True, exist_ok=True)
+                tar.extractall(path=extract_path)
+                print(f"Extracted {tar_file} to {extract_path}")
+        except tarfile.TarError as e:
+            print(f"Failed to extract {tar_file}: {e}")
 
 def fetch(dest: Path, nodes: list[Node]):
     dest.mkdir(parents=True, exist_ok=True)
@@ -11,7 +45,6 @@ def fetch(dest: Path, nodes: list[Node]):
         extravars={"dest": str(dest)},
     ).run()
 
-
 def clean(nodes: list[Node]):
     return RemoteAction(
         "clean",

yardstick_benchmark/clean.yml

@@ -2,6 +2,9 @@
 - name: Clean data from nodes
   gather_facts: true
   hosts: all
+  become: "{{ lookup('env', 'ANSIBLE_USE_SUDO') | default('false') == 'true' }}"
+  become_method: sudo
+  become_user: root
   tasks:
     - name: Remove data from nodes
       file:

yardstick_benchmark/fetch.yml

@@ -3,10 +3,20 @@
   gather_facts: true
   hosts: all
   tasks:
-    - name: Get data from nodes
-      ansible.posix.synchronize:
-        src: "{{node_wd}}"
-        dest: "{{dest}}"
-        mode: pull
-    - debug:
-        msg: "Output files have been written to {{dest}}"
+    - name: Archive the directory
+      ansible.builtin.archive:
+        path: "{{node_wd}}"
+        dest: "{{node_wd}}.tar.gz"
+        format: gz
+
+    - name: Fetch the archive
+      ansible.builtin.fetch:
+        src: "{{node_wd}}.tar.gz"
+        dest: "{{dest}}/"
+        flat: yes
+
+    - name: Remove the archive file
+      ansible.builtin.file:
+        path: "{{node_wd}}.tar.gz"
+        state: absent
+

yardstick_benchmark/games/minecraft/server/__init__.py

@@ -4,7 +4,7 @@
 
 
 class PaperMC(RemoteApplication):
-    def __init__(self, nodes: list[Node]):
+    def __init__(self, nodes: list[Node], world_dir_path='', copy_player_data=False, use_strace=False):
         super().__init__(
             "papermc",
             nodes,
@@ -15,5 +15,8 @@ def __init__(self, nodes: list[Node]):
             extravars={
                 "hostnames": [n.host for n in nodes],
                 "papermc_template": str(Path(__file__).parent / "server.properties.j2"),
+                "use_strace": use_strace,
+                "world_dir_path": world_dir_path,
+                "copy_player_data": copy_player_data
             },
         )

yardstick_benchmark/games/minecraft/server/papermc_deploy.yml

@@ -2,6 +2,7 @@
 - name: Deploy PaperMC
   gather_facts: true
   hosts: all
+  become: "{{ use_sudo }}"
   tasks:
     - file:
         path: "{{ wd }}"
@@ -22,3 +23,15 @@
       get_url:
         url: https://search.maven.org/remotecontent?filepath=org/jolokia/jolokia-agent-jvm/2.0.3/jolokia-agent-jvm-2.0.3-javaagent.jar
         dest: "{{wd}}/jolokia-agent-jvm-2.0.3-javaagent.jar"
+
+    - name: Copy world dir to destination
+      copy:
+        src: "{{ world_dir_path }}"
+        dest: "{{ wd }}"
+      when: world_dir_path is defined and world_dir_path | length > 0
+
+    - name: Delete playerdata subdirectory if necessary
+      file:
+        path: "{{ wd }}/world/playerdata"
+        state: absent
+      when: copy_player_data != true

yardstick_benchmark/games/minecraft/server/papermc_start.yml

@@ -2,15 +2,40 @@
 - name: Deploy PaperMC
   gather_facts: true
   hosts: all
+  become: "{{ use_sudo }}"
   tasks:
     - name: Run PaperMC
       shell:
         cmd: |
           module load java/jdk-17 || true # just in case we are on DAS
-          nohup java -javaagent:jolokia-agent-jvm-2.0.3-javaagent.jar -jar paper-1.20.1-58.jar &> /dev/null &
-          echo $! > papermc.pid
+          nohup sh -c "{{ 'strace -f -ttt -T -e trace=pread64,pwrite64,read,write,openat,close -o fsysstrace.log' if use_strace else '' }} java -Xms16G -Xmx16G -javaagent:jolokia-agent-jvm-2.0.3-javaagent.jar -jar paper-1.20.1-58.jar" {{ '' if use_strace else '> /dev/null' }} & 
+          NOHUP_PID=$!
+          echo $NOHUP_PID > nohup.pid 
+          sleep 2
+          {% if use_strace %}
+          STRACE_PID=$(pgrep -P $NOHUP_PID -x 'strace.*') 
+          JAVA_PID=$(pgrep -P $STRACE_PID -x 'java.*')
+          echo $JAVA_PID > papermc.pid
+          {% else %}
+          JAVA_PID=$(pgrep -P $NOHUP_PID -x 'java.*')
+          echo $JAVA_PID > papermc.pid
+          {% endif %}
         chdir: "{{ wd }}"
+    - name: Wait for the latest.log file to appear
+      wait_for:
+        path: "{{ wd }}/logs/latest.log"
+        state: present
+        timeout: 60
     - name: Waiting for server to become ready
       wait_for:
+        timeout: 60
         path: "{{wd}}/logs/latest.log"
         search_regex: 'For help, type "help"'
+      when: use_sudo != true
+    - name: Waiting for server to become ready using cat (used due to compatibility issue with nullblk)
+      shell: |
+        while ! cat "{{ wd }}/logs/latest.log" | grep -q 'For help, type "help"'; do
+          sleep 0.05
+        done
+      timeout: 60
+      when: use_sudo == true

yardstick_benchmark/games/minecraft/server/papermc_stop.yml

@@ -2,10 +2,12 @@
 - name: Stop PaperMC
   gather_facts: true
   hosts: all
+  become: "{{ use_sudo }}"
   tasks:
     - name: Find PaperMC PID
-      shell: |
-        cat {{wd}}/papermc.pid
+      shell: 
+        cmd: cat papermc.pid
+        chdir: "{{ wd }}"
       register: shell_papermc_pid
     - name: Obtain PaperMC pid
       set_fact:

yardstick_benchmark/games/minecraft/server/server.properties.j2

@@ -2,8 +2,8 @@
 #Mon Apr 29 15:15:15 CEST 2024
 enable-jmx-monitoring=true
 rcon.port=25575
-level-seed=
-gamemode=survival
+level-seed=-3691007458655063350
+gamemode=creative
 enable-command-block=false
 enable-query=false
 generator-settings={}
@@ -25,7 +25,7 @@ enable-status=true
 allow-flight=true
 initial-disabled-packs=
 broadcast-rcon-to-ops=true
-view-distance=10
+view-distance=32
 server-ip=
 resource-pack-prompt=
 allow-nether=true
@@ -56,4 +56,6 @@ spawn-monsters=false
 enforce-whitelist=false
 spawn-protection=16
 resource-pack-sha1=
+sync-chunk-writes=true
+region-file-compression=none
 max-world-size=29999984

yardstick_benchmark/games/minecraft/workload/__init__.py

@@ -43,3 +43,33 @@ def __init__(
                 "bots_per_node": bots_per_node,
             },
         )
+
+class TraceBased(RemoteApplication):
+    def __init__(
+            self, 
+            nodes: list[Node], 
+            server_host: str, 
+            trace_file: str,
+            duration: timedelta = timedelta(seconds=60)
+    ):
+        super().__init__(
+            "trace_based",
+            nodes,
+            Path(__file__).parent / "trace_deploy.yml",
+            Path(__file__).parent / "trace_start.yml",
+            Path(__file__).parent / "trace_stop.yml",
+            Path(__file__).parent / "trace_cleanup.yml",
+            extravars={
+                "hostnames": [n.host for n in nodes],
+                "mc_host": server_host,
+                "scripts": [
+                    str(Path(__file__).parent / "trace.js"),
+                    str(Path(__file__).parent / "trace_bot.js"),
+                ],
+                "trace": [
+                    trace_file,
+                ],
+                "trace_file": Path(trace_file).name,
+                "duration": duration.total_seconds()
+            },
+        )

yardstick_benchmark/games/minecraft/workload/trace.js

@@ -0,0 +1,103 @@
+const fs = require('node:fs');
+const readline = require('readline');
+const { Worker, isMainThread } = require("worker_threads");
+const { writeFile } = require('node:fs/promises');
+const Rcon = require('rcon');
+
+const host = process.env.MC_HOST;
+var rcon = new Rcon(host, 25575, 'password');
+const trace = process.env.TRACE;
+
+const trace_start_timestamp = parseInt(process.env.START_TIMESTAMP);
+const timeout_s = parseInt(process.env.DURATION);
+
+const players = {};
+
+function sleep(ms) {
+    return new Promise(resolve => setTimeout(resolve, ms));
+}
+
+async function tp(player, arguments){
+    const [x, z] = arguments;
+    console.log('hello');
+    rcon.send(`tp ${player} ${x} -60 ${z}`);
+}
+
+async function run(){
+    await writeFile('issued.csv', 'timestamp, player_name, action \n');
+    await writeFile('received.csv', 'timestamp, player_name, issuer_name, action \n');
+
+    console.log('script starts');
+
+    rcon.on('auth', async () => {
+        const fileStream = fs.createReadStream(trace);
+        const rl = readline.createInterface({
+            input: fileStream,
+            output: process.stdout
+        });
+
+        const benchmark_starting_time = new Date();
+        console.log('authenticated');
+        let benchmark_starting_timestamp = benchmark_starting_time.valueOf();
+
+        rl.on("line", async (line) => {
+            let [timestamp, player, action, ...arguments] = line.split(' ');
+
+            timestamp = Number(timestamp);
+
+            while(true) {
+                const current_time = new Date();
+                let current_timestamp = current_time.valueOf();
+
+                if((current_timestamp - benchmark_starting_timestamp) > (timestamp - trace_start_timestamp)){
+                    break;
+                }
+                await sleep(1000);
+                console.log('hello???');
+            }
+
+            if(action == 'tp'){
+                tp(player, arguments);
+            } else if(!players[player]){
+                console.log('initalizing a new one!');
+                const worker = new Worker("./trace_bot.js");
+                players[player] = worker;
+
+                worker.on('message', (result) => {
+                    console.log(`Received from worker (player: ${player}): ${result}`);
+                });
+
+                worker.on('error', (error) => {
+                    console.error(`Worker error for player ${player}:`, error);
+                });
+
+                worker.on('exit', (code) => {
+                    if (code !== 0) {
+                        console.error(`Worker stopped with exit code ${code} for player ${player}`);
+                    }
+                });
+
+                players[player].postMessage({ action, arguments: [player] });
+            } else{
+                players[player].postMessage({ action, arguments });
+            }
+        });
+    });
+
+    rcon.on('end', () => {
+        console.log("RCON connection closed.");
+    });
+
+    rcon.on('error', (err) => {
+        console.error("RCON error:", err);
+    });
+
+    rcon.connect();
+
+    await sleep(timeout_s * 1000)
+    process.exit(0);
+}
+
+if (isMainThread) {
+    run();
+}