Use 0.0.0.0 for servers in benchmarks and tests (denoland#3010)

cli/tests/echo_server.ts

@@ -1,5 +1,5 @@
 const { args, listen, copy } = Deno;
-const addr = args[1] || "127.0.0.1:4544";
+const addr = args[1] || "0.0.0.0:4544";
 const [hostname, port] = addr.split(":");
 const listener = listen({ hostname, port: Number(port) });
 console.log("listening on", addr);

tools/http_benchmark.py

@@ -17,128 +17,133 @@
 LAST_PORT = 4544
 
 
-def get_addr(port=None):
+def server_addr(port):
+ return "0.0.0.0:%s" % port
+
+
+def get_port(port=None):
  global LAST_PORT
  if port is None:
  port = LAST_PORT
  LAST_PORT = LAST_PORT + 1
- return ("127.0.0.1:%d" % (port))
+ # Return port as str because all usages below are as a str and having it an
+ # integer just adds complexity.
+ return str(port)
 
 
 def deno_tcp(deno_exe):
- addr = get_addr()
- deno_cmd = [deno_exe, "run", "--allow-net", "tools/deno_tcp.ts", addr]
+ port = get_port()
+ deno_cmd = [
+ deno_exe, "run", "--allow-net", "tools/deno_tcp.ts",
+ server_addr(port)
+ ]
  print "http_benchmark testing DENO tcp."
- return run(deno_cmd, addr)
+ return run(deno_cmd, port)
 
 
 def deno_tcp_current_thread(deno_exe):
- addr = get_addr()
+ port = get_port()
  deno_cmd = [
  deno_exe, "run", "--current-thread", "--allow-net",
- "tools/deno_tcp.ts", addr
+ "tools/deno_tcp.ts",
+ server_addr(port)
  ]
  print "http_benchmark testing DENO tcp (single-thread)."
- return run(deno_cmd, addr)
+ return run(deno_cmd, port)
 
 
 def deno_http(deno_exe):
- addr = get_addr()
+ port = get_port()
  deno_cmd = [
  deno_exe, "run", "--allow-net",
- "js/deps/https/deno.land/std/http/http_bench.ts", addr
+ "js/deps/https/deno.land/std/http/http_bench.ts",
+ server_addr(port)
  ]
  print "http_benchmark testing DENO using net/http."
- return run(deno_cmd, addr)
+ return run(deno_cmd, port)
 
 
 def deno_tcp_proxy(deno_exe, hyper_hello_exe):
- addr = get_addr()
- origin_addr = get_addr()
+ port = get_port()
+ origin_port = get_port()
  deno_cmd = [
- deno_exe, "run", "--allow-net", "tools/deno_tcp_proxy.ts", addr,
- origin_addr
+ deno_exe, "run", "--allow-net", "tools/deno_tcp_proxy.ts",
+ server_addr(port),
+ server_addr(origin_port)
  ]
  print "http_proxy_benchmark testing DENO using net/tcp."
  return run(
  deno_cmd,
- addr,
- origin_cmd=http_proxy_origin(hyper_hello_exe, origin_addr))
+ port,
+ origin_cmd=http_proxy_origin(hyper_hello_exe, origin_port))
 
 
 def deno_http_proxy(deno_exe, hyper_hello_exe):
- addr = get_addr()
- origin_addr = get_addr()
+ port = get_port()
+ origin_port = get_port()
  deno_cmd = [
- deno_exe, "run", "--allow-net", "tools/deno_http_proxy.ts", addr,
- origin_addr
+ deno_exe, "run", "--allow-net", "tools/deno_http_proxy.ts",
+ server_addr(port),
+ server_addr(origin_port)
  ]
  print "http_proxy_benchmark testing DENO using net/http."
  return run(
  deno_cmd,
- addr,
- origin_cmd=http_proxy_origin(hyper_hello_exe, origin_addr))
+ port,
+ origin_cmd=http_proxy_origin(hyper_hello_exe, origin_port))
 
 
 def deno_core_single(exe):
  print "http_benchmark testing deno_core_single"
- return run([exe, "--single-thread"], "127.0.0.1:4544")
+ return run([exe, "--single-thread"], 4544)
 
 
 def deno_core_multi(exe):
  print "http_benchmark testing deno_core_multi"
- return run([exe, "--multi-thread"], "127.0.0.1:4544")
+ return run([exe, "--multi-thread"], 4544)
 
 
 def node_http():
- addr = get_addr()
- node_cmd = ["node", "tools/node_http.js", addr.split(":")[1]]
+ port = get_port()
+ node_cmd = ["node", "tools/node_http.js", port]
  print "http_benchmark testing NODE."
- return run(node_cmd, addr)
+ return run(node_cmd, port)
 
 
 def node_http_proxy(hyper_hello_exe):
- addr = get_addr()
- origin_addr = get_addr()
- node_cmd = [
- "node", "tools/node_http_proxy.js",
- addr.split(":")[1],
- origin_addr.split(":")[1]
- ]
+ port = get_port()
+ origin_port = get_port()
+ node_cmd = ["node", "tools/node_http_proxy.js", port, origin_port]
  print "http_proxy_benchmark testing NODE."
- return run(node_cmd, addr, None,
- http_proxy_origin(hyper_hello_exe, origin_addr))
+ return run(node_cmd, port, None,
+ http_proxy_origin(hyper_hello_exe, origin_port))
 
 
 def node_tcp_proxy(hyper_hello_exe):
- addr = get_addr()
- origin_addr = get_addr()
- node_cmd = [
- "node", "tools/node_tcp_proxy.js",
- addr.split(":")[1],
- origin_addr.split(":")[1]
- ]
+ port = get_port()
+ origin_port = get_port()
+ node_cmd = ["node", "tools/node_tcp_proxy.js", port, origin_port]
  print "http_proxy_benchmark testing NODE tcp."
- return run(node_cmd, addr, None,
- http_proxy_origin(hyper_hello_exe, origin_addr))
+ return run(node_cmd, port, None,
+ http_proxy_origin(hyper_hello_exe, origin_port))
 
 
 def node_tcp():
- addr = get_addr()
- node_cmd = ["node", "tools/node_tcp.js", addr.split(":")[1]]
+ port = get_port()
+ node_cmd = ["node", "tools/node_tcp.js", port]
  print "http_benchmark testing node_tcp.js"
- return run(node_cmd, addr)
+ return run(node_cmd, port)
 
 
-def http_proxy_origin(hyper_hello_exe, addr):
- return [hyper_hello_exe, addr.split(":")[1]]
+def http_proxy_origin(hyper_hello_exe, port):
+ return [hyper_hello_exe, port]
 
 
 def hyper_http(hyper_hello_exe):
- addr = get_addr()
- hyper_cmd = [hyper_hello_exe, addr.split(":")[1]]
+ port = get_port()
+ hyper_cmd = [hyper_hello_exe, port]
  print "http_benchmark testing RUST hyper."
- return run(hyper_cmd, addr)
+ return run(hyper_cmd, port)
 
 
 def http_benchmark(build_dir):
@@ -165,7 +170,7 @@ def http_benchmark(build_dir):
  }
 
 
-def run(server_cmd, addr, merge_env=None, origin_cmd=None):
+def run(server_cmd, port, merge_env=None, origin_cmd=None):
 
  # Run deno echo server in the background.
  if merge_env is None:
@@ -183,13 +188,14 @@ def run(server_cmd, addr, merge_env=None, origin_cmd=None):
  if origin_cmd is not None:
  origin = subprocess.Popen(origin_cmd, env=env)
 
+ print server_cmd
  server = subprocess.Popen(server_cmd, env=env)
 
- time.sleep(10) # wait for server to wake up. TODO racy.
+ time.sleep(5) # wait for server to wake up. TODO racy.
 
  try:
- cmd = "third_party/wrk/%s/wrk -d %s --latency https://%s/" % (
- util.platform(), DURATION, addr)
+ cmd = "third_party/wrk/%s/wrk -d %s --latency https://127.0.0.1:%s/" % (
+ util.platform(), DURATION, port)
  print cmd
  output = subprocess.check_output(cmd, shell=True)
  stats = util.parse_wrk_output(output)

tools/throughput_benchmark.py

@@ -13,7 +13,8 @@
 import util
 
 MB = 1024 * 1024
-ADDR = "127.0.0.1:4544"
+SERVER_ADDR = "0.0.0.0:4544"
+CLIENT_ADDR = "127.0.0.1 4544"
 
 
 def cat(deno_exe, megs):
@@ -30,14 +31,17 @@ def cat(deno_exe, megs):
 def tcp(deno_exe, megs):
  size = megs * MB
  # Run deno echo server in the background.
- echo_server = subprocess.Popen(
- [deno_exe, "run", "--allow-net", "tests/echo_server.ts", ADDR])
+ args = [
+ deno_exe, "run", "--allow-net", "tests/echo_server.ts", SERVER_ADDR
+ ]
+ print args
+ echo_server = subprocess.Popen(args)
 
  time.sleep(5) # wait for deno to wake up. TODO racy.
  try:
  start = time.time()
- cmd = ("head -c %s /dev/zero " % size) + "| nc " + ADDR.replace(
-  ":", " ")
+ nc_cmd = "nc " + CLIENT_ADDR
+ cmd = ("head -c %s /dev/zero " % size) + " | " + nc_cmd
  print cmd
  subprocess.check_output(cmd, shell=True)
  end = time.time()