misc: add image-size for benchmark

Signed-off-by: Desiki-high <ding_yadong@foxmail.com>

misc/benchmark/bench.py

@@ -184,13 +184,16 @@ def run_cmd_url_wait(self, repo, runargs):
         run_elapsed = datetime.timestamp(end_run) - datetime.timestamp(start_run)
 
         print("Run time: %f s" % run_elapsed)
+
         read_amount, read_count = "-", "-"
         if self.snapshotter == "nydus":
             read_amount, read_count = metrics.collect_backend()
+        image_size = metrics.collect_size(image_repo(repo), image_tag(repo))
+
         if self.cleanup:
             self.clean_up(image_ref, container_id)
 
-        return pull_elapsed, create_elapsed, run_elapsed, read_amount, read_count
+        return pull_elapsed, create_elapsed, run_elapsed, image_size, read_amount, read_count
 
     def pull_cmd(self, image_ref):
         insecure_flag = "--insecure-registry" if self.insecure_registry else ""
@@ -253,11 +256,11 @@ def bench_image(local_registry, insecure_local_registry, image, f: TextIOWrapper
         cleanup=True,
         insecure_registry=insecure_local_registry,
     )
-    pull_elapsed, create_elapsed, run_elapsed, read_amount, read_count = runner.run(bench)
+    pull_elapsed, create_elapsed, run_elapsed, image_size, read_amount, read_count = runner.run(bench)
     total_elapsed = f"{pull_elapsed + create_elapsed + run_elapsed: .6f}"
     pull_elapsed = f"{pull_elapsed: .6f}"
     create_elapsed = f"{create_elapsed: .6f}"
     run_elapsed = f"{run_elapsed: .6f}"
-    line = f"{bench.name},{pull_elapsed},{create_elapsed},{run_elapsed},{total_elapsed},{read_amount},{read_count}"
+    line = f"{pull_elapsed},{create_elapsed},{run_elapsed},{total_elapsed},{image_size},{read_amount},{read_count}"
     f.writelines(line + "\n")
     f.flush()

misc/benchmark/benchmark.py

@@ -41,6 +41,7 @@ def main():
     # bench
     start_bench(cfg, cfg["image"], mode)
 
+
 def collect_metrics(cfg: dict, image: str) -> str:
     """
     collect container access metrics
@@ -53,9 +54,6 @@ def start_bench(cfg: dict, image: str, mode: str):
     bench oci, nydus without prefetch, nydus with all prefetch, nydus witch prefetch file list
     """
     f = open(util.image_repo(image) + ".csv", "w")
-    csv_headers = "repo,pull_elapsed(s),create_elapsed(s),run_elapsed(s),total_elapsed(s),read_amount(MB),read_count"
-    f.writelines(csv_headers + "\n")
-    f.flush()
     if mode == "oci":
         util.enable_wondersphaper(cfg["bandwith"])
         bench.bench_image(cfg["local_registry"], cfg["insecure_local_registry"], image, f)

misc/benchmark/benchmark_summary.py

@@ -37,11 +37,11 @@
     "zran-all-prefetch-master.csv",
     "zran-all-prefetch.csv",
     "nydus-no-prefetch-master.csv",
-    "nydus-no-prefetch-master.csv",
-    "zran-no-prefetch.csv",
+    "nydus-no-prefetch.csv",
     "zran-no-prefetch-master.csv",
-    "nydus-filelist-prefetch.csv",
-    "nydus-filelist-prefetch-master.csv"
+    "zran-no-prefetch.csv",
+    "nydus-filelist-prefetch-master.csv",
+    "nydus-filelist-prefetch.csv"
 ]
 
 
@@ -51,8 +51,8 @@ def __init__(self, mode):
 
     def summary(self):
         self.prepare_csv()
-        print("| bench-result | pull-elapsed(s) | create-elapsed(s) | run-elapsed(s) | total-elapsed(s) |read-amount(MB) |read-count |")
-        print("|:-------------|:---------------:|:-----------------:|:--------------:|:----------------:|:--------------:|:---------:|")
+        print("| bench-result | pull-elapsed(s) | create-elapsed(s) | run-elapsed(s) | total-elapsed(s) | image-size(MB) | read-amount(MB) | read-count |")
+        print("|:-------------|:---------------:|:-----------------:|:--------------:|:----------------:|:--------------:|:---------------:|:----------:|")
         if self.mode == "benchmark-result":
             self.print_csv_result()
         else:
@@ -81,10 +81,9 @@ def print_csv(file: str):
     with open(file, 'r', newline='') as f:
         filename = file.rstrip(".csv")
         rows = csv.reader(f)
-        next(rows)  # we don't need header
         for row in rows:
-            _, pull_elapsed, create_elapsed, run_elapsed, total_elapsed, read_amount, read_count = row
-            print(f"|{filename}|{pull_elapsed}|{create_elapsed}|{run_elapsed}|{total_elapsed}|{read_amount}|{read_count}|")
+            pull_elapsed, create_elapsed, run_elapsed, total_elapsed, image_size, read_amount, read_count = row
+            print(f"|{filename}|{pull_elapsed}|{create_elapsed}|{run_elapsed}|{total_elapsed}|{image_size}|{read_amount}|{read_count}|")
 
 
 def main():

misc/benchmark/metrics.py

@@ -227,7 +227,6 @@ def send_request(sock_path, url):
     file_path = random_string()
     cmd = ["sudo", "curl", "--unix-socket", sock_path]
     cmd.extend(["-X", "GET", URL_PREFIX + url])
-    # caruon
     with open(TEMP_DIR + "/" + file_path, 'w') as f:
         _ = run_cmd(
             cmd,
@@ -285,3 +284,43 @@ def collect_backend() -> Tuple[str, str]:
     backend = get_backend_metrics(socket)
     shutil.rmtree(TEMP_DIR)
     return round(backend["read_amount_total"] / 1024 / 1024, 2), backend["read_count"]
+
+
+def collect_size(repo: str, tag: str):
+    """
+    collect image size for benchmark
+
+    return the image size
+    """
+    init()
+    file_path = random_string()
+    cmd = ["sudo", "curl", "-H", "'Accept: application/vnd.docker.distribution.manifest.v2+json'"]
+    cmd.extend([f"localhost:5000/v2/{repo}/manifests/{tag}"])
+    with open(TEMP_DIR + "/" + file_path, 'w') as f:
+        _ = run_cmd(
+            cmd,
+            shell=True,
+            stdout=f,
+            stderr=subprocess.PIPE
+        )
+    with open(TEMP_DIR + "/" + file_path, 'r') as file:
+        content = file.read()
+        if "errors" in content:
+            file_path_zran = random_string()
+            cmd = ["sudo", "curl", "-H", "'Accept: application/vnd.oci.image.manifest.v1+json'"]
+            cmd.extend([f"localhost:5000/v2/{repo}/manifests/{tag}"])
+            with open(TEMP_DIR + "/" + file_path_zran, 'w') as f:
+                _ = run_cmd(
+                    cmd,
+                    shell=True,
+                    stdout=f,
+                    stderr=subprocess.PIPE
+                )
+            with open(TEMP_DIR + "/" + file_path_zran, 'r') as file_zran:
+                content = file_zran.read()
+    manifest = json.loads(content)
+    size = 0
+    for item in manifest["layers"]:
+        size += item["size"]
+    shutil.rmtree(TEMP_DIR)
+    return round(size / 1024 / 1024, 2)