Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
PaddlePaddle
PaddleDetection
提交
0bebaa05
P
PaddleDetection
项目概览
PaddlePaddle
/
PaddleDetection
大约 1 年 前同步成功
通知
694
Star
11112
Fork
2696
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
184
列表
看板
标记
里程碑
合并请求
40
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
PaddleDetection
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
184
Issue
184
列表
看板
标记
里程碑
合并请求
40
合并请求
40
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
提交
0bebaa05
编写于
6月 13, 2017
作者:
H
Helin Wang
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix according to comments
上级
f6148eb2
变更
4
隐藏空白更改
内联
并排
Showing
4 changed file
with
101 addition
and
59 deletion
+101
-59
go/master/client.go
go/master/client.go
+2
-0
go/master/client_test.go
go/master/client_test.go
+15
-0
go/master/service.go
go/master/service.go
+66
-44
go/pserver/client.go
go/pserver/client.go
+18
-15
未找到文件。
go/master/client.go
浏览文件 @
0bebaa05
...
...
@@ -28,6 +28,8 @@ func NewClient(addr Addresser) *Client {
func
(
c
*
Client
)
monitorMaster
(
addr
Addresser
)
{
lastMaster
:=
""
monitor
:=
func
()
{
// get the lastest address of the master server,
// connect to the new address once address changed.
curMaster
:=
addr
.
Address
()
if
curMaster
!=
lastMaster
{
if
curMaster
==
""
{
...
...
go/master/client_test.go
浏览文件 @
0bebaa05
...
...
@@ -11,6 +11,8 @@ import (
"testing"
"time"
log
"github.com/sirupsen/logrus"
"github.com/PaddlePaddle/Paddle/go/master"
"github.com/PaddlePaddle/recordio"
)
...
...
@@ -23,6 +25,8 @@ const (
var
port
int
func
init
()
{
log
.
SetLevel
(
log
.
ErrorLevel
)
l
,
err
:=
net
.
Listen
(
"tcp"
,
":0"
)
if
err
!=
nil
{
panic
(
err
)
...
...
@@ -91,6 +95,17 @@ func TestClientFull(t *testing.T) {
t
.
Fatal
(
i
,
"should get error."
)
}
err
=
c
.
TaskFinished
(
tasks
[
0
]
.
ID
)
if
err
!=
nil
{
t
.
Fatal
(
err
)
}
tasks
=
tasks
[
1
:
]
task
,
err
:=
c
.
GetTask
()
if
err
!=
nil
{
t
.
Fatal
(
err
)
}
tasks
=
append
(
tasks
,
task
)
for
_
,
task
:=
range
tasks
{
err
=
c
.
TaskFinished
(
task
.
ID
)
if
err
!=
nil
{
...
...
go/master/service.go
浏览文件 @
0bebaa05
...
...
@@ -2,12 +2,13 @@ package master
import
(
"errors"
"log"
"os"
"path/filepath"
"sync"
"time"
log
"github.com/sirupsen/logrus"
"github.com/PaddlePaddle/recordio"
)
...
...
@@ -112,7 +113,7 @@ func readChunks(globPaths []string) ([]Chunk, error) {
}
if
len
(
paths
)
==
0
{
return
nil
,
errors
.
New
(
"no valid datset specified"
)
return
nil
,
errors
.
New
(
"no valid dat
a
set specified"
)
}
for
_
,
path
:=
range
paths
{
...
...
@@ -170,6 +171,7 @@ func (s *Service) SetDataset(globPaths []string, dummy *int) error {
err
=
s
.
snapshot
()
if
err
!=
nil
{
log
.
Errorln
(
err
)
return
err
}
...
...
@@ -178,6 +180,43 @@ func (s *Service) SetDataset(globPaths []string, dummy *int) error {
return
nil
}
func
(
s
*
Service
)
checkTimeoutFunc
(
taskID
int
,
epoch
int
)
func
()
{
return
func
()
{
s
.
mu
.
Lock
()
defer
s
.
mu
.
Unlock
()
t
,
ok
:=
s
.
taskQueues
.
Pending
[
taskID
]
if
!
ok
{
return
}
if
t
.
Epoch
!=
epoch
{
// new epoch, task launched after the
// schedule of this timeout check.
return
}
defer
func
()
{
err
:=
s
.
snapshot
()
if
err
!=
nil
{
log
.
Errorln
(
err
)
}
}()
delete
(
s
.
taskQueues
.
Pending
,
t
.
Task
.
ID
)
t
.
NumTimeout
++
if
t
.
NumTimeout
>
s
.
timeoutMax
{
log
.
Warningf
(
"Task %v failed %d times, discard.
\n
"
,
t
.
Task
,
t
.
NumTimeout
)
s
.
taskQueues
.
Failed
=
append
(
s
.
taskQueues
.
Failed
,
t
.
Task
)
return
}
log
.
Warningf
(
"Task %v failed %d times, retry.
\n
"
,
t
.
Task
,
t
.
NumTimeout
)
s
.
taskQueues
.
Todo
=
append
(
s
.
taskQueues
.
Todo
,
t
)
}
}
// GetTask gets a new task from the service.
func
(
s
*
Service
)
GetTask
(
dummy
int
,
task
*
Task
)
error
{
select
{
...
...
@@ -190,19 +229,25 @@ func (s *Service) GetTask(dummy int, task *Task) error {
if
len
(
s
.
taskQueues
.
Todo
)
==
0
{
if
len
(
s
.
taskQueues
.
Done
)
==
0
{
if
len
(
s
.
taskQueues
.
Pending
)
==
0
{
return
errors
.
New
(
"all task failed"
)
err
:=
errors
.
New
(
"all task failed"
)
log
.
Warningln
(
err
)
return
err
}
// TODO(helin): client need to retry in this
// error case. Gotcha: RPC client can't
// compare returned error with predefined
// errors like io.EOF
. B
ecause interface don't
// errors like io.EOF
, b
ecause interface don't
// have same dynamic value when in different
// process.
return
errors
.
New
(
"no more available task"
)
// process. So we need to figure out a way for
// client to check this error correctly.
err
:=
errors
.
New
(
"no more available task"
)
log
.
Warningln
(
err
)
return
err
}
s
.
taskQueues
.
Todo
=
s
.
taskQueues
.
Done
s
.
taskQueues
.
Todo
=
nil
s
.
taskQueues
.
Done
=
nil
log
.
Infoln
(
"No more todo task, but trainer is requesting task to do. Move all done task to todo."
)
}
t
:=
s
.
taskQueues
.
Todo
[
0
]
...
...
@@ -215,41 +260,9 @@ func (s *Service) GetTask(dummy int, task *Task) error {
}
*
task
=
t
.
Task
log
.
Infof
(
"Task #%d dispatched
\n
"
,
task
.
ID
)
time
.
AfterFunc
(
s
.
timeoutDur
,
func
(
taskID
int
,
epoch
int
)
func
()
{
return
func
()
{
s
.
mu
.
Lock
()
defer
s
.
mu
.
Unlock
()
t
,
ok
:=
s
.
taskQueues
.
Pending
[
taskID
]
if
!
ok
{
return
}
if
t
.
Epoch
!=
epoch
{
// new epoch, task launched after the
// schedule of this timeout check.
return
}
defer
func
()
{
err
:=
s
.
snapshot
()
if
err
!=
nil
{
log
.
Println
(
err
)
}
}()
delete
(
s
.
taskQueues
.
Pending
,
t
.
Task
.
ID
)
t
.
NumTimeout
++
if
t
.
NumTimeout
>
s
.
timeoutMax
{
s
.
taskQueues
.
Failed
=
append
(
s
.
taskQueues
.
Failed
,
t
.
Task
)
return
}
s
.
taskQueues
.
Todo
=
append
(
s
.
taskQueues
.
Todo
,
t
)
}
}(
t
.
Task
.
ID
,
t
.
Epoch
))
time
.
AfterFunc
(
s
.
timeoutDur
,
s
.
checkTimeoutFunc
(
t
.
Task
.
ID
,
t
.
Epoch
))
return
nil
}
...
...
@@ -262,9 +275,13 @@ func (s *Service) TaskFinished(taskID int, dummy *int) error {
s
.
mu
.
Lock
()
defer
s
.
mu
.
Unlock
()
log
.
Infof
(
"Task %d finished
\n
"
,
taskID
)
t
,
ok
:=
s
.
taskQueues
.
Pending
[
taskID
]
if
!
ok
{
return
errors
.
New
(
"pending task not found"
)
err
:=
errors
.
New
(
"pending task not found"
)
log
.
Warningln
(
err
)
return
err
}
// task finished, reset timeout
...
...
@@ -272,10 +289,15 @@ func (s *Service) TaskFinished(taskID int, dummy *int) error {
s
.
taskQueues
.
Done
=
append
(
s
.
taskQueues
.
Done
,
t
)
delete
(
s
.
taskQueues
.
Pending
,
taskID
)
if
len
(
s
.
taskQueues
.
Pending
)
==
0
{
if
len
(
s
.
taskQueues
.
Pending
)
==
0
&&
len
(
s
.
taskQueues
.
Todo
)
==
0
{
log
.
Infoln
(
"No more todo and pending task, start a new pass."
)
s
.
taskQueues
.
Todo
=
append
(
s
.
taskQueues
.
Todo
,
s
.
taskQueues
.
Done
...
)
s
.
taskQueues
.
Done
=
nil
}
return
s
.
snapshot
()
err
:=
s
.
snapshot
()
if
err
!=
nil
{
log
.
Errorln
(
err
)
}
return
err
}
go/pserver/client.go
浏览文件 @
0bebaa05
...
...
@@ -57,26 +57,29 @@ func (c *Client) monitorPservers(l Lister, pserverNum int) {
}
for
i
:=
range
lastServers
{
if
lastServers
[
i
]
.
Addr
!=
curServers
[
i
]
.
Addr
{
if
curServers
[
i
]
.
Addr
==
""
{
err
:=
c
.
pservers
[
i
]
.
Close
()
if
err
!=
nil
{
log
.
Println
(
err
)
}
continue
}
if
lastServers
[
i
]
.
Addr
==
curServers
[
i
]
.
Addr
{
continue
}
err
:=
c
.
pservers
[
i
]
.
Connect
(
curServers
[
i
]
.
Addr
)
if
curServers
[
i
]
.
Addr
==
""
{
err
:=
c
.
pservers
[
i
]
.
Close
()
if
err
!=
nil
{
log
.
Println
(
err
)
// connect to addr failed, set
// to last known addr in order
// to retry next time.
curServers
[
i
]
.
Addr
=
lastServers
[
i
]
.
Addr
}
continue
}
err
:=
c
.
pservers
[
i
]
.
Connect
(
curServers
[
i
]
.
Addr
)
if
err
!=
nil
{
log
.
Println
(
err
)
// connect to addr failed, set
// to last known addr in order
// to retry next time.
curServers
[
i
]
.
Addr
=
lastServers
[
i
]
.
Addr
}
}
lastServers
=
curServers
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录