提交 8a2667cd 编写于 作者: L Liu Yiqun

Merge branch 'develop' into core_inference_prepare

# FileManager设计文档
## 目标
在本文档中,我们设计说明了名为FileManager系统,方便用户上传自己的训练数据以进行分布式训练
主要功能包括:
- 提供常用的命令行管理命令管理文件和目录
- 支持大文件的断点上传、下载
## 名词解释
- PFS:是`Paddlepaddle cloud File System`的缩写,是对用户文件存储空间的抽象,与之相对的是local filesystem。目前我们用CephFS来搭建。
- [CephFS](http://docs.ceph.com/docs/master/cephfs/):一个POSIX兼容的文件系统。
- Chunk:逻辑划上文件分块的单位。
## 模块
### 架构图
<image src=./src/filemanager.png width=900>
### PFSClient
- 功能: 详细设计[link](./pfs/pfsclient.md)
- 提供用户管理文件的命令
- 需要可以跨平台执行
- 双向验证
PFSClient需要和Ingress之间做双向验证<sup>[tls](#tls)</sup>,所以用户需要首先在`cloud.paddlepaddle.org`上注册一下,申请用户空间,并且把系统生成的CA(certificate authority)、Key、CRT(CA signed certificate)下载到本地,然后才能使用PFSClient。
### [Ingress](https://kubernetes.io/docs/concepts/services-networking/ingress/)
- 功能:
提供七层协议的反向代理、基于粘性会话的负载均衡功能。
- 透传用户身份的办法
Ingress需要把PFSClient的身份信息传给PFSServer,配置的方法参考[link](http://www.integralist.co.uk/posts/clientcertauth.html#3)
### PFSServer
PFSServer提供RESTful API接口,接收处理PFSClient端的文件管理请求,并且把结果返回PFSClient端。
RESTful API
- /api/v1/files
- `GET /api/v1/files`: Get metadata of files or directories.
- `POST /api/v1/files`: Create files or directories.
- `PATCH /api/v1/files`: Update files or directories.
- `DELETE /api/v1/files`: Delete files or directories.
- /api/v1/file/chunks
- `GET /api/v1/storage/file/chunks`: Get chunks's metadata of a file.
- /api/v1/storage/files
- `GET /api/v1/storage/files`: Download files or directories.
- `POST /api/v1/storage/files`: Upload files or directories.
- /api/v1/storage/file/chunks
- `GET /api/v1/storage/file/chunks`: Download chunks's data.
- `POST /api/v1/storage/file/chunks`: Upload chunks's data.
## 文件传输优化
### 分块文件传输
用户文件可能是比较大的,上传到Cloud或者下载到本地的时间可能比较长,而且在传输的过程中也可能出现网络不稳定的情况。为了应对以上的问题,我们提出了Chunk的概念,一个Chunk由所在的文件偏移、数据、数据长度及校验值组成。文件的上传和下载都是通过对Chunk的操作来实现的。由于Chunk比较小(默认256K),完成一个传输动作完成的时间也比较短,不容易出错。PFSClient需要在传输完毕最后一个Chunk的时候检查destination文件的MD5值是否和source文件一致。
一个典型的Chunk如下所示:
```
type Chunk struct {
fileOffset int64
checksum uint32
len uint32
data []byte
}
```
### 生成sparse文件
当destination文件不存在或者大小和source文件不一致时,可以用[Fallocate](https://Go.org/pkg/syscall/#Fallocate)生成sparse文件,然后就可以并发写入多个Chunk。
### 覆盖不一致的部分
文件传输的的关键在于需要PFSClient端对比source和destination的文件Chunks的checksum是否保持一致,不一致的由PFSClient下载或者传输Chunk完成。这样已经传输成功的部分就不用重新传输了。
## 用户使用流程
参考[link](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/cluster_train/data_dispatch.md)
## 框架生成
[swagger](https://github.com/swagger-api/swagger-codegen)生成PFSClient和PFSServer的框架部分,以便我们可以把更多的精力放到逻辑本身上。
## 参考文档
- <a name=tls></a>[TLS complete guide](https://github.com/k8sp/tls/blob/master/tls.md)
- [aws.s3](http://docs.aws.amazon.com/cli/latest/reference/s3/)
- [linux man document](https://linux.die.net/man/)
# PFSClient
## Description
The `pfs` command is a Command Line Interface to manage your files on PaddlePaddle Cloud
## Synopsis
```
paddle [options] pfs <subcommand> [parameters]
```
## Options
```
--profile (string)
Use a specific profile from your credential file.
--help (string)
Display more information about command
--version
Output version information and exit
--debug
Show detailed debugging log
--only-show-errors (boolean)
Only errors and warnings are displayed. All other output is suppressed.
```
## Path Arguments
When using a command, we need to specify path arguments. There are two path argument type: `localpath` and `pfspath`.
A `pfspath` begin with `/pfs`, eg: `/pfs/$DATACENTER/home/$USER/folder`.
[Here](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/design/cluster_train/data_dispatch.md#上传训练文件) is how to config datacenters.
## order of Path Arguments
Commonly, if there are two path arguments, the first is the source, and the second is the destination.
## Subcommonds
- rm - remove files or directories
```
Synopsis:
rm [-r] [-v] <PFSPath> ...
Options:
-r
Remove directories and their contents recursively
-v
Cause rm to be verbose, showing files after they are removed.
Examples:
paddle pfs rm /pfs/$DATACENTER/home/$USER/file
paddle pfs rm -r /pfs/$DATACENTER/home/$USER/folder
```
- mv - move (rename) files
```
Synopsis:
mv [-f | -n] [-v] <LocalPath> <PFSPath>
mv [-f | -n] [-v] <LocalPath> ... <PFSPath>
mv [-f | -n] [-v] <PFSPath> <LocalPath>
mv [-f | -n] [-v] <PFSPath> ... <LocalPath>
mv [-f | -n] [-v] <PFSPath> <PFSPath>
mv [-f | -n] [-v] <PFSPath> ... <PFSPath>
Options:
-f
Do not prompt for confirmation before overwriting the destination path. (The -f option overrides previous -n options.)
-n
Do not overwrite an existing file. (The -n option overrides previous -f options.)
-v
Cause mv to be verbose, showing files after they are moved.
Examples:
paddle pfs mv ./text1.txt /pfs/$DATACENTER/home/$USER/text1.txt
```
- cp - copy files or directories
```
Synopsis:
cp [-r] [-f | -n] [-v] [--preserve--links] <LocalPath> <PFSPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <LocalPath> ... <PFSPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <PFSPath> <LocalPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <PFSPath> ... <LocalPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <PFSPath> <PFSPath>
cp [-r] [-f | -n] [-v] [--preserve--links] <PFSPath> ... <PFSPath>
Options:
-r
Copy directories recursively
-f
Do not prompt for confirmation before overwriting the destination path. (The -f option overrides previous -n options.)
-n
Do not overwrite an existing file. (The -n option overrides previous -f options.)
-v
Cause cp to be verbose, showing files after they are copied.
--preserve--links
Reserve links when copy links
Examples:
paddle pfs cp ./file /pfs/$DATACENTER/home/$USER/file
paddle pfs cp /pfs/$DATACENTER/home/$USER/file ./file
```
- ls- list files
```
Synopsis:
ls [-r] <PFSPath> ...
Options:
-R
List directory(ies) recursively
Examples:
paddle pfs ls /pfs/$DATACENTER/home/$USER/file
paddle pfs ls /pfs/$DATACENTER/home/$USER/folder
```
- mkdir - mkdir directory(ies)
Create intermediate directory(ies) as required.
```
Synopsis:
mkdir <PFSPath> ...
Examples:
paddle pfs mkdir /pfs/$DATACENTER/home/$USER/folder
```
......@@ -39,10 +39,13 @@ class CreateBatchReaderOp : public framework::OperatorBase {
private:
void RunImpl(const framework::Scope& scope,
const platform::Place& dev_place) const override {
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto* out = scope.FindVar(Output("Out"))
->template GetMutable<framework::ReaderHolder>();
if (out->Get() != nullptr) {
return;
}
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
out->Reset(
new BatchReader(underlying_reader.Get(), Attr<int>("batch_size")));
}
......
......@@ -99,10 +99,13 @@ class CreateDoubleBufferReaderOp : public framework::OperatorBase {
private:
void RunImpl(const framework::Scope& scope,
const platform::Place& dev_place) const override {
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto* out = scope.FindVar(Output("Out"))
->template GetMutable<framework::ReaderHolder>();
if (out->Get() != nullptr) {
return;
}
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto place_str = Attr<std::string>("place");
platform::Place place;
......
......@@ -62,12 +62,15 @@ class CreateMultiPassReaderOp : public framework::OperatorBase {
private:
void RunImpl(const framework::Scope& scope,
const platform::Place& dev_place) const override {
auto* out = detail::Ref(scope.FindVar(Output("Out")))
.GetMutable<framework::ReaderHolder>();
if (out->Get() != nullptr) {
return;
}
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto& out = detail::Ref(scope.FindVar(Output("Out")));
int pass_num = Attr<int>("pass_num");
out.GetMutable<framework::ReaderHolder>()->Reset(
new MultiPassReader(underlying_reader.Get(), pass_num));
out->Reset(new MultiPassReader(underlying_reader.Get(), pass_num));
}
};
......
......@@ -80,10 +80,14 @@ class CreateShuffleReaderOp : public framework::OperatorBase {
private:
void RunImpl(const framework::Scope& scope,
const platform::Place& dev_place) const override {
auto* out = detail::Ref(scope.FindVar(Output("Out")))
.GetMutable<framework::ReaderHolder>();
if (out->Get() != nullptr) {
return;
}
const auto& underlying_reader = scope.FindVar(Input("UnderlyingReader"))
->Get<framework::ReaderHolder>();
auto& var = detail::Ref(scope.FindVar(Output("Out")));
var.GetMutable<framework::ReaderHolder>()->Reset(
out->Reset(
new ShuffleReader(underlying_reader.Get(),
static_cast<size_t>(Attr<int>("buffer_size"))));
}
......
......@@ -640,6 +640,20 @@ class Operator(object):
"""
return self.desc.block_attr(name)
def all_attrs(self):
"""
Get the attribute dict
Returns(dict): The Operator's attribute dict
"""
attr_names = self.attr_names
attr_map = {}
for n in attr_names:
if n == 'sub_block':
attr_map[n] = self.block_attr(n)
else:
attr_map[n] = self.attr(n)
return attr_map
class Block(object):
def __init__(self, program, idx):
......
......@@ -255,7 +255,32 @@ def _copy_reader_var_(block, var):
new_var.desc.set_shapes(var.desc.shapes())
new_var.desc.set_dtypes(var.desc.dtypes())
new_var.persistable = True
return monkey_patch_reader_methods(new_var)
return new_var
def _copy_reader_create_op_(block, op):
input_param_names = op.input_names
new_input_map = {}
for param_name in input_param_names:
new_input_map[param_name] = []
arg_names = op.input(param_name)
for arg_name in arg_names:
new_input_map[param_name].append(block.var(arg_name))
output_param_names = op.output_names
new_output_map = {}
for param_name in output_param_names:
new_output_map[param_name] = []
arg_names = op.output(param_name)
for arg_name in arg_names:
new_output_map[param_name].append(block.var(arg_name))
new_op = block.append_op(
type=op.type,
inputs=new_input_map,
outputs=new_output_map,
attrs=op.all_attrs())
return new_op
def open_recordio_file(filename, shapes, lod_levels, dtypes):
......@@ -283,8 +308,9 @@ def open_recordio_file(filename, shapes, lod_levels, dtypes):
startup_var.desc.set_dtypes(dtypes)
startup_var.persistable = True
return _copy_reader_var_(default_main_program().current_block(),
startup_var)
main_prog_var = _copy_reader_var_(default_main_program().current_block(),
startup_var)
return monkey_patch_reader_methods(main_prog_var)
def open_files(filenames, thread_num, shapes, lod_levels, dtypes):
......@@ -313,22 +339,25 @@ def open_files(filenames, thread_num, shapes, lod_levels, dtypes):
startup_var.desc.set_dtypes(dtypes)
startup_var.persistable = True
return _copy_reader_var_(default_main_program().current_block(),
startup_var)
main_prog_var = _copy_reader_var_(default_main_program().current_block(),
startup_var)
return monkey_patch_reader_methods(main_prog_var)
def __create_decorated_reader__(op_type, reader, attrs):
var_name = unique_name(op_type)
startup_blk = default_startup_program().current_block()
startup_var = startup_blk.create_var(name=var_name)
startup_blk.append_op(
startop_op = startup_blk.append_op(
type=op_type,
inputs={'UnderlyingReader': reader},
outputs={'Out': [startup_var]},
attrs=attrs)
startup_var.persistable = True
return _copy_reader_var_(default_main_program().current_block(),
startup_var)
main_prog_block = default_main_program().current_block()
main_prog_var = _copy_reader_var_(main_prog_block, startup_var)
_copy_reader_create_op_(main_prog_block, startop_op)
return monkey_patch_reader_methods(main_prog_var)
def create_shuffle_reader(reader, buffer_size):
......
......@@ -15,8 +15,8 @@
import unittest
import paddle.fluid as fluid
import paddle
import paddle.dataset.mnist as mnist
import paddle.v2 as paddle
import paddle.v2.dataset.mnist as mnist
class TestRecordIO(unittest.TestCase):
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册