From b252dc1798296a22731a5bd4f18f065bb32639b8 Mon Sep 17 00:00:00 2001 From: "google-labs-jules[bot]" <161369871+google-labs-jules[bot]@users.noreply.github.com> Date: Mon, 16 Mar 2026 19:34:30 +0000 Subject: [PATCH 1/2] docs: add REPRODUCTION_GUIDE.md for path tracking test: modify example_orchestrator to automatically track path and run cleanup chore: generate grpc code Co-authored-by: forrestk3 <22152628+forrestk3@users.noreply.github.com> --- REPRODUCTION_GUIDE.md | 110 ++++++++++++++++++ .../link_failure_grpc/link_failure_pb2.py | 40 +++++++ .../link_failure_pb2_grpc.py | 97 +++++++++++++++ .../test/example_orchestrator.py | 37 +++--- 4 files changed, 270 insertions(+), 14 deletions(-) create mode 100644 REPRODUCTION_GUIDE.md create mode 100644 network_orchestrator/southbound/link_failure_grpc/link_failure_pb2.py create mode 100644 network_orchestrator/southbound/link_failure_grpc/link_failure_pb2_grpc.py diff --git a/REPRODUCTION_GUIDE.md b/REPRODUCTION_GUIDE.md new file mode 100644 index 0000000..ba6e116 --- /dev/null +++ b/REPRODUCTION_GUIDE.md @@ -0,0 +1,110 @@ +# 复现指南：模拟两节点通信数据包随时间变化的路径 + +本项目（TinyLEO）提供了 `network_orchestrator` 组件，该组件可以在拥有完整 Linux 网络命名空间（Network Namespaces）权限的服务器中模拟低轨卫星网络，并跟踪两节点通信路径随时间的变化。 + +由于沙箱（Sandbox）及常规 Docker 容器的权限限制（不支持 `ip netns` 完整创建及底层网络配置），我们将演示如何配置并在您的真实物理机或具备完整权限的虚拟机（Root privileges & Multi-core）上运行。 + +## 前置环境准备 + +1. **准备服务器**: + - 至少一台具备 Root 权限的 Ubuntu 20.04+ 服务器。建议内存 >= 120GB，CPU 核心 >= 32（用于生成数千个卫星网络节点）。 + - 确保安装了 Python 3.10 或更高版本。 + +2. **安装依赖环境**: + 在具有 Root 权限的终端下执行以下命令安装相关的依赖包（建议使用全局环境或使用 `--break-system-packages`，由于涉及到系统底层网络修改）： + + ```bash + sudo apt-get update + sudo apt-get install -y libnetfilter-queue-dev python3-pip openssh-server + sudo service ssh start + + # 安装 Python 依赖包 + sudo python3 -m pip install argparse numpy requests paramiko tqdm networkx \ + grpcio grpcio-tools NetfilterQueue pyroute2 scapy python-iptables watchdog scipy + ``` + +3. **配置 SSH 免密登录（供仿真使用）**: + **⚠️ 强烈建议：** 此项目会频繁操作底层网络命名空间，必须使用 root 权限。请**务必在独立、隔离的虚拟机或物理测试机中进行**，不要在生产环境中尝试。 + + 仿真器基于 SSH 与各个容器及命名空间交互，需保证配置文件的机器允许 Root SSH 登录。推荐使用 SSH Key 认证： + + ```bash + # 1. 如果你还没有 ssh key，请先生成 + ssh-keygen -t rsa -b 4096 -C "tinyleo_test" + + # 2. 将公钥拷贝到目标机器（或本机的 root 账户下） + # 如果是本机测试： + sudo mkdir -p /root/.ssh + sudo cp ~/.ssh/id_rsa.pub /root/.ssh/authorized_keys + sudo chmod 700 /root/.ssh + sudo chmod 600 /root/.ssh/authorized_keys + ``` + +## 代码配置修改与准备 + +### 1. 生成 gRPC 代码 + +在项目目录执行以下命令生成 gRPC 通信代码： + +```bash +cd network_orchestrator/southbound/ +python3 -m grpc_tools.protoc -I. --python_out=./ --grpc_python_out=./ ./link_failure_grpc/link_failure.proto +``` + +### 2. 配置服务器连接（tinyleo_config.json） + +修改 `network_orchestrator/test/config/tinyleo_config.json` 文件中的 `Machines` 数组，填入本机或目标主机的 IP 和 SSH 信息（若使用 key 登录，password 保持配置中对应的值即可，paramiko 也支持通过秘钥自动连接，具体依据你的环境配置）： + +```json + { + "IP": "101.6.21.153", // 替换为你的服务器 IP + "port": 22, + "username": "root", + "password": "xxx" // 若有密码在此填入，若配置了免密通常可忽略 + } +``` + +### 3. 配置路径追踪脚本（example_orchestrator.py） + +我们在 `network_orchestrator/test/example_orchestrator.py` 中已经进行了适配修改，增加了异常捕获，并确保开启了 `TEST = True`。 + +关键追踪逻辑在 `for timestamp in range(0, sn.duration):` 循环中： +```python + if TEST: + time.sleep(2) + try: + # 记录此时的 traceroute（路径数据） + sn.set_traceroute("GS1", "GS2", f"ts{timestamp}") + except Exception as e: + print(f"Skipping traceroute at timestamp {timestamp} due to error: {e}") +``` + +在这个循环中，仿真器在每个时间戳都会更新一次拓扑（`sn.update_tinyleo_topology(timestamp)`），并向对应的容器内发送指令执行 `traceroute GS1 GS2`。 + +## 运行仿真与分析 + +在终端中执行： + +```bash +cd network_orchestrator/test/ +sudo python3 example_orchestrator.py +``` + +### 执行过程 +1. **拓扑预测与网络创建**：脚本首先加载所有的卫星数据，开始计算每个时间片（Timestamp）的最优拓扑结构。 +2. **命名空间创建**：基于 Paramiko SSH 连接，底层会在系统中创建几千个以 `SH1SAT...` 为前缀的网络命名空间（Network Namespace）。 +3. **数据打点（Traceroute）**：系统会在指定的时间戳（默认每 20 秒间隔模拟一次时间推移）向两节点发送 traceroute 指令，结果会保存在新生成的文件夹中（例如 `tinyleo-Arbitrary-LeastDelay/result/` 目录下）。 + +### 查看路径变化结果 + +运行结束后，你可以通过以下方式查看数据包传输路径随时间的变化： + +1. **自动生成的结果文件**： + 在 `network_orchestrator/test/tinyleo-Arbitrary-LeastDelay/result/` 文件夹内，会生成形如 `ts0_traceroute.txt`、`ts1_traceroute.txt` 等日志文件，你可以按时间戳对比路由跳数和经过的节点差异。 + +2. **进入容器手动验证**： + 你可以将 `example_orchestrator.py` 中的清理代码去掉，然后在仿真运行时运行： + ```bash + python3 get_container.py + ``` + 它会输出 `nsenter` 命令（如 `nsenter -m -u -i -n -p -t bash`），允许你直接进入卫星或者地面站容器，手动执行 `traceroute <目标节点IP>` 来追踪当前路由。 diff --git a/network_orchestrator/southbound/link_failure_grpc/link_failure_pb2.py b/network_orchestrator/southbound/link_failure_grpc/link_failure_pb2.py new file mode 100644 index 0000000..431c130 --- /dev/null +++ b/network_orchestrator/southbound/link_failure_grpc/link_failure_pb2.py @@ -0,0 +1,40 @@ +# -*- coding: utf-8 -*- +# Generated by the protocol buffer compiler. DO NOT EDIT! +# NO CHECKED-IN PROTOBUF GENCODE +# source: link_failure_grpc/link_failure.proto +# Protobuf Python Version: 6.31.1 +"""Generated protocol buffer code.""" +from google.protobuf import descriptor as _descriptor +from google.protobuf import descriptor_pool as _descriptor_pool +from google.protobuf import runtime_version as _runtime_version +from google.protobuf import symbol_database as _symbol_database +from google.protobuf.internal import builder as _builder +_runtime_version.ValidateProtobufRuntimeVersion( + _runtime_version.Domain.PUBLIC, + 6, + 31, + 1, + '', + 'link_failure_grpc/link_failure.proto' +) +# @@protoc_insertion_point(imports) + +_sym_db = _symbol_database.Default() + + + + +DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n$link_failure_grpc/link_failure.proto\"+\n\x12LinkFailureRequest\x12\x15\n\rsatellite_ids\x18\x01 \x03(\t\"7\n\x13LinkFailureResponse\x12\x0f\n\x07success\x18\x01 \x01(\x08\x12\x0f\n\x07message\x18\x02 \x01(\t2T\n\x12LinkFailureService\x12>\n\x11HandleLinkFailure\x12\x13.LinkFailureRequest\x1a\x14.LinkFailureResponseb\x06proto3') + +_globals = globals() +_builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals) +_builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'link_failure_grpc.link_failure_pb2', _globals) +if not _descriptor._USE_C_DESCRIPTORS: + DESCRIPTOR._loaded_options = None + _globals['_LINKFAILUREREQUEST']._serialized_start=40 + _globals['_LINKFAILUREREQUEST']._serialized_end=83 + _globals['_LINKFAILURERESPONSE']._serialized_start=85 + _globals['_LINKFAILURERESPONSE']._serialized_end=140 + _globals['_LINKFAILURESERVICE']._serialized_start=142 + _globals['_LINKFAILURESERVICE']._serialized_end=226 +# @@protoc_insertion_point(module_scope) diff --git a/network_orchestrator/southbound/link_failure_grpc/link_failure_pb2_grpc.py b/network_orchestrator/southbound/link_failure_grpc/link_failure_pb2_grpc.py new file mode 100644 index 0000000..6ec21de --- /dev/null +++ b/network_orchestrator/southbound/link_failure_grpc/link_failure_pb2_grpc.py @@ -0,0 +1,97 @@ +# Generated by the gRPC Python protocol compiler plugin. DO NOT EDIT! +"""Client and server classes corresponding to protobuf-defined services.""" +import grpc +import warnings + +from link_failure_grpc import link_failure_pb2 as link__failure__grpc_dot_link__failure__pb2 + +GRPC_GENERATED_VERSION = '1.78.0' +GRPC_VERSION = grpc.__version__ +_version_not_supported = False + +try: + from grpc._utilities import first_version_is_lower + _version_not_supported = first_version_is_lower(GRPC_VERSION, GRPC_GENERATED_VERSION) +except ImportError: + _version_not_supported = True + +if _version_not_supported: + raise RuntimeError( + f'The grpc package installed is at version {GRPC_VERSION},' + + ' but the generated code in link_failure_grpc/link_failure_pb2_grpc.py depends on' + + f' grpcio>={GRPC_GENERATED_VERSION}.' + + f' Please upgrade your grpc module to grpcio>={GRPC_GENERATED_VERSION}' + + f' or downgrade your generated code using grpcio-tools<={GRPC_VERSION}.' + ) + + +class LinkFailureServiceStub(object): + """Missing associated documentation comment in .proto file.""" + + def __init__(self, channel): + """Constructor. + + Args: + channel: A grpc.Channel. + """ + self.HandleLinkFailure = channel.unary_unary( + '/LinkFailureService/HandleLinkFailure', + request_serializer=link__failure__grpc_dot_link__failure__pb2.LinkFailureRequest.SerializeToString, + response_deserializer=link__failure__grpc_dot_link__failure__pb2.LinkFailureResponse.FromString, + _registered_method=True) + + +class LinkFailureServiceServicer(object): + """Missing associated documentation comment in .proto file.""" + + def HandleLinkFailure(self, request, context): + """Missing associated documentation comment in .proto file.""" + context.set_code(grpc.StatusCode.UNIMPLEMENTED) + context.set_details('Method not implemented!') + raise NotImplementedError('Method not implemented!') + + +def add_LinkFailureServiceServicer_to_server(servicer, server): + rpc_method_handlers = { + 'HandleLinkFailure': grpc.unary_unary_rpc_method_handler( + servicer.HandleLinkFailure, + request_deserializer=link__failure__grpc_dot_link__failure__pb2.LinkFailureRequest.FromString, + response_serializer=link__failure__grpc_dot_link__failure__pb2.LinkFailureResponse.SerializeToString, + ), + } + generic_handler = grpc.method_handlers_generic_handler( + 'LinkFailureService', rpc_method_handlers) + server.add_generic_rpc_handlers((generic_handler,)) + server.add_registered_method_handlers('LinkFailureService', rpc_method_handlers) + + + # This class is part of an EXPERIMENTAL API. +class LinkFailureService(object): + """Missing associated documentation comment in .proto file.""" + + @staticmethod + def HandleLinkFailure(request, + target, + options=(), + channel_credentials=None, + call_credentials=None, + insecure=False, + compression=None, + wait_for_ready=None, + timeout=None, + metadata=None): + return grpc.experimental.unary_unary( + request, + target, + '/LinkFailureService/HandleLinkFailure', + link__failure__grpc_dot_link__failure__pb2.LinkFailureRequest.SerializeToString, + link__failure__grpc_dot_link__failure__pb2.LinkFailureResponse.FromString, + options, + channel_credentials, + insecure, + call_credentials, + compression, + wait_for_ready, + timeout, + metadata, + _registered_method=True) diff --git a/network_orchestrator/test/example_orchestrator.py b/network_orchestrator/test/example_orchestrator.py index 1957f46..0f93450 100644 --- a/network_orchestrator/test/example_orchestrator.py +++ b/network_orchestrator/test/example_orchestrator.py @@ -73,22 +73,31 @@ sn.update_tinyleo_topology(timestamp) if TEST: time.sleep(2) - sn.set_traceroute("GS1", "GS2",f"ts{timestamp}") + try: + sn.set_traceroute("GS1", "GS2",f"ts{timestamp}") + except Exception as e: + print(f"Skipping traceroute at timestamp {timestamp} due to error: {e}") if timestamp == 0: if TEST: time.sleep(1) - sn.set_traceroute("GS1", "GS2",f"ts{timestamp}_before_link_failure") - sn.set_ping("GS1", "GS2", f"ts{timestamp}_when_link_failure") - sn.set_iperf("GS1", "GS2", f"ts{timestamp}_when_link_failure") - time.sleep(1) - sn.tinyleo_fault_test() - time.sleep(2) - sn.set_traceroute("GS1", "GS2",f"ts{timestamp}_after_link_failure") - time.sleep(update_time-0.5 - (time.time() - start_time)) + try: + sn.set_traceroute("GS1", "GS2",f"ts{timestamp}_before_link_failure") + sn.set_ping("GS1", "GS2", f"ts{timestamp}_when_link_failure") + sn.set_iperf("GS1", "GS2", f"ts{timestamp}_when_link_failure") + time.sleep(1) + sn.tinyleo_fault_test() + time.sleep(2) + sn.set_traceroute("GS1", "GS2",f"ts{timestamp}_after_link_failure") + except Exception as e: + print(f"Skipping tests at timestamp 0 due to error: {e}") + time.sleep(max(0, update_time-0.5 - (time.time() - start_time))) if timestamp != sn.duration - 1: - sn.set_ping("GS1", "GS2", f"ts{timestamp}-{timestamp+1}_topo_change") - sn.set_iperf("GS1", "GS2", f"ts{timestamp}-{timestamp+1}_topo_change") - time.sleep(update_time - (time.time() - start_time)) + try: + sn.set_ping("GS1", "GS2", f"ts{timestamp}-{timestamp+1}_topo_change") + sn.set_iperf("GS1", "GS2", f"ts{timestamp}-{timestamp+1}_topo_change") + except Exception as e: + print(f"Skipping cross-timestamp tests due to error: {e}") + time.sleep(max(0, update_time - (time.time() - start_time))) -if input('clear environment?[y/n]').strip().lower()[:1] == 'y': - sn.clean() \ No newline at end of file +# Automatically clean up environment without waiting for user input +sn.clean() \ No newline at end of file From 496a32f5f5ca3024e5f9a7c301cd5d518085ed3f Mon Sep 17 00:00:00 2001 From: "google-labs-jules[bot]" <161369871+google-labs-jules[bot]@users.noreply.github.com> Date: Tue, 17 Mar 2026 14:06:48 +0000 Subject: [PATCH 2/2] fix: resolve path and permission issues for local execution - Dynamically infer base execution directory in pyctr.c and deploy_srv6_agent.py instead of hardcoding /root/ - Ensure /run/netns directory exists before container initialization - Catch exceptions during container creation in sn_remote.py - Add missing gRPC generated files Co-authored-by: forrestk3 <22152628+forrestk3@users.noreply.github.com> --- .../deploy_srv6_agent.py | 14 ++++++++++++- network_orchestrator/southbound/pyctr.c | 20 ++++++++++++++++++- .../southbound/sn_controller.py | 2 +- network_orchestrator/southbound/sn_remote.py | 17 ++++++++++++++-- 4 files changed, 48 insertions(+), 5 deletions(-) diff --git a/network_orchestrator/geographic_srv6_anycast/deploy_srv6_agent.py b/network_orchestrator/geographic_srv6_anycast/deploy_srv6_agent.py index c62d8c6..f30c9cc 100644 --- a/network_orchestrator/geographic_srv6_anycast/deploy_srv6_agent.py +++ b/network_orchestrator/geographic_srv6_anycast/deploy_srv6_agent.py @@ -17,8 +17,20 @@ import subprocess from tqdm import tqdm -dir = '/root/tinyleo-Arbitrary-LeastDelay' +import sys +import os + +# Allow overriding dir via command line argument, fallback to current dir or hardcoded path +if len(sys.argv) > 1: + dir = sys.argv[1] +else: + dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__))) + pid_path = f"{dir}/container_pid.txt" +if not os.path.exists(pid_path): + print(f"Error: PID file not found at {pid_path}") + sys.exit(1) + pid_maps = {} with open(pid_path, 'r') as f: for line in f: diff --git a/network_orchestrator/southbound/pyctr.c b/network_orchestrator/southbound/pyctr.c index 5ec11c3..36b2e13 100644 --- a/network_orchestrator/southbound/pyctr.c +++ b/network_orchestrator/southbound/pyctr.c @@ -84,7 +84,25 @@ static int container_init( } char controller_src[PATH_MAX+add_len]; - snprintf(controller_src, sizeof(controller_src), "/root/tinyleo-Arbitrary-LeastDelay/controller"); + // Find the base directory (e.g. /home/user/tinyleo-Arbitrary-LeastDelay/shellX/overlay/SATY) + // We assume the experiment_name is "tinyleo-Arbitrary-LeastDelay" and controller is adjacent + // Let's pass it via argument or infer it from base_dir by looking for "shell" or "GS" + // Since we can't easily change the C interface without updating Python everywhere, we'll + // infer it by finding the path up to the experiment directory. + const char *exp_marker1 = "/shell"; + const char *exp_marker2 = "/GS-"; + char exp_dir[PATH_MAX] = {0}; + strncpy(exp_dir, base_dir, PATH_MAX - 1); + char *ptr = strstr(exp_dir, exp_marker1); + if (!ptr) ptr = strstr(exp_dir, exp_marker2); + if (ptr) { + *ptr = '\0'; + } else { + // Fallback + strncpy(exp_dir, "/root/tinyleo-Arbitrary-LeastDelay", PATH_MAX - 1); + } + + snprintf(controller_src, sizeof(controller_src), "%s/controller", exp_dir); if(mount(controller_src, controller_dst, NULL, MS_BIND|MS_REC, NULL) != 0) { return child_err("mount --bind controller failed: ", err_fd); } diff --git a/network_orchestrator/southbound/sn_controller.py b/network_orchestrator/southbound/sn_controller.py index d5815c4..5b0e5d7 100644 --- a/network_orchestrator/southbound/sn_controller.py +++ b/network_orchestrator/southbound/sn_controller.py @@ -749,7 +749,7 @@ def deploy_tinyleo_srv6_agent(self): """ sn_remote_wait_output( self.ssh, - f"python3 {self.dir}/controller/geographic_srv6_anycast/deploy_srv6_agent.py" + f"python3 {self.dir}/controller/geographic_srv6_anycast/deploy_srv6_agent.py {self.dir}" ) def ping_async(self, res_path, src, dst): diff --git a/network_orchestrator/southbound/sn_remote.py b/network_orchestrator/southbound/sn_remote.py index 25c379d..fd7398b 100644 --- a/network_orchestrator/southbound/sn_remote.py +++ b/network_orchestrator/southbound/sn_remote.py @@ -501,6 +501,7 @@ def _load_netns(pid, name): pid_file = open(dir + '/' + PID_FILENAME, 'w', encoding='utf-8') sat_cnt = 0 + os.makedirs("/run/netns", exist_ok=True) # Ensure /run/netns exists for shell_id, mid_dict in enumerate(sat_mid_dict_shell): for node, mid in mid_dict.items(): if mid != machine_id: @@ -509,7 +510,13 @@ def _load_netns(pid, name): node_dir = f"{dir}/shell{shell_id}/overlay/{node}" sat_cnt += 1 os.makedirs(node_dir, exist_ok=True) - pid_file.write(node+':'+str(pyctr.container_run(node_dir, node))+' ') + try: + c_pid = pyctr.container_run(node_dir, node) + if c_pid <= 0: + print(f"Error starting container for {node}") + pid_file.write(f"{node}:{c_pid} ") + except Exception as e: + print(f"Exception starting container for {node}: {e}") pid_file.write('\n') print(f'[{machine_id}] shell {shell_id}: {sat_cnt} satellites initialized') @@ -522,7 +529,13 @@ def _load_netns(pid, name): gs_lst.append(node) node_dir = f'{overlay_dir}/{node}' os.makedirs(node_dir, exist_ok=True) - pid_file.write(node+':'+str(pyctr.container_run(node_dir, node))+' ') + try: + c_pid = pyctr.container_run(node_dir, node) + if c_pid <= 0: + print(f"Error starting container for {node}") + pid_file.write(f"{node}:{c_pid} ") + except Exception as e: + print(f"Exception starting container for {node}: {e}") pid_file.write('\n') print(f'[{machine_id}] GS:', ','.join(gs_lst))