【分析】Ceph文件系统修复机制cephfs-data-scan(2)

时间:2021-11-22 16:20:26

前述章节已经介绍过,ceph-data-scan是通过函数data_scan.main(args)解析并执行用户命令的。本章节主要介绍data_scan

=========================
作者:Younger Liu,
本作品采用知识共享署名-非商业性使用-相同方式共享 3.0 未本地化版本许可协议进行许可。
本文链接地址为:http://blog.csdn.net/younger_china/article/details/76718726
=========================

1. ceph-scan-data的用法

通过代码来看,如下:

  std::cout << "Usage: \n"    << " cephfs-data-scan init [--force-init]\n"    << " cephfs-data-scan scan_extents [--force-pool] [--worker_n N --worker_m M] <data pool name>\n"    << " cephfs-data-scan scan_inodes [--force-pool] [--force-corrupt] [--worker_n N --worker_m M] <data pool name>\n"    << " cephfs-data-scan pg_files <path> <pg id> [<pg id>...]\n"    << " cephfs-data-scan scan_links\n"    << "\n"    << "   --force-corrupt: overrite apparently corrupt structures\n"    << "   --force-init: write root inodes even if they exist\n"    << "   --force-pool: use data pool even if it is not in FSMap\n"    << "   --worker_m: Maximum number of workers\n"    << "   --worker_n: Worker number, range 0-(worker_m-1)\n"    << "\n"    << " cephfs-data-scan scan_frags [--force-corrupt]\n"    << " cephfs-data-scan cleanup <data pool name>\n"    << std::endl;  generic_client_usage();}

2. 参数解析

(1) 判断参数个数,参数个数至少为1,否则datascan是无法理解用户想做什么的.

  if (args.size() < 1) {    usage();    return -EINVAL;  }

(2)通用RADOS初始化,及初始化全局变量g_ceph_context

g_ceph_context是CephContext *的类型,

  // Common RADOS init: open metadata pool  // =====================================  librados::Rados rados;  int r = rados.init_with_context(g_ceph_context);  if (r < 0) {    derr << "RADOS unavailable" << dendl;    return r;  }

(3) 通过for循环,依次解析每一个参数,在解析过程中会验证各个参数,然后以全局变量进行标识

  for (std::vector<const char *>::const_iterator i = args.begin() + 1;       i != args.end(); ++i) {       ......

(4) 检查namespace,如果没有指定并且本地只有一个,则选择这一个,否则报错

  // If caller didn't specify a namespace, try to pick  // one if only one exists  if (fscid == FS_CLUSTER_ID_NONE) {    if (fsmap->filesystem_count() == 1) {      fscid = fsmap->get_filesystem()->fscid;    } else {      std::cerr << "Specify a filesystem with --filesystem" << std::endl;      return -EINVAL;    }  }  auto fs =  fsmap->get_filesystem(fscid);  assert(fs != nullptr);

(5) 如果没有初始化output目录,则初始化为MetadataDriver(),否则初始化为LocalFileDriver()

  // Default to output to metadata pool  if (driver == NULL) {    driver = new MetadataDriver();    driver->set_force_corrupt(force_corrupt);    driver->set_force_init(force_init);    dout(4) << "Using metadata pool output" << dendl;  }

3. 相关初始化

(1) 与Rados建立通信联系

  dout(4) << "connecting to RADOS..." << dendl;  r = rados.connect();  if (r < 0) {    std::cerr << "couldn't connect to cluster: " << cpp_strerror(r)              << std::endl;    return r;  }

(2) 初始化driver

  r = driver->init(rados, metadata_pool_name, fsmap, fscid);  if (r < 0) {    return r;  } 

接下来是对用户的各种操作分析

4. init操作分析

当用户输入命令如下命令

# cephfs-data-scan init [--force-init]

1.解析参数

当进入data_scan.main中时,首先会解析参数,当解析到--force-init时,会设置全局变量force_init=true

  if (arg == "--force-pool") {    ...  } else if (arg == "--force-init") {    force_init = true;    return true;  } else {    return false;  }

2. 初始化

然后会初始化fs、 driver,与rados建立通信联系

3. 执行init操作

然后调用init_rootfs来初始化

  // Finally, dispatch command  if (command == "scan_inodes") {    ....  } else if (command == "init") {    return driver->init_roots(fs->mds_map.get_first_data_pool());  } else {    std::cerr << "Unknown command '" << command << "'" << std::endl;    return -EINVAL;  }

MetadataDriver->init_rootfs

【分析】Ceph文件系统修复机制cephfs-data-scan(2)

LocalFileDriver->init_rootfs

【分析】Ceph文件系统修复机制cephfs-data-scan(2)

=========================

作者:Younger Liu,

本作品采用知识共享署名-非商业性使用-相同方式共享 3.0 未本地化版本许可协议进行许可。

本文链接地址为:http://blog.csdn.net/younger_china/article/details/76718726


=========================