我看到在proxy中，文件上传的逻辑，是通过directory获取stores，然后遍历stores，将文件上传到volume对应的store中，如果部分上传到失败，会直接返回，这样是不是会有文件残留，使得各个volume大小不同，会导致有些volume已经写满，有些未写满的状况出现？

我看了volume中的compact逻辑，貌似里面对于删除了的文件，直接跳过，不写入新的volume中，没对多个volume中，文件差异进行处理。

可能调查不是很详细，先假定是这样的流程，我想问一下，有很多时候，网络稍微不大稳定，很容易出现写入的失败的情况，这样是不是导致文件残留现象比较频繁出现？我可以怎样处理这个问题呢？

params = url.Values{}
for _, host = range res.Stores {
	params.Set("key", strconv.FormatInt(res.Key, 10))
	params.Set("cookie", strconv.FormatInt(int64(res.Cookie), 10))
	params.Set("vid", strconv.FormatInt(int64(res.Vid), 10))
	uri = fmt.Sprintf(_storeUploadApi, host)
	if err = Http("POST", uri, params, buf, &sRet); err != nil {
		return
	}
	if sRet.Ret != 1 {
		log.Errorf("http.Post store sRet.Ret: %d  %s %d %d %d", sRet.Ret, uri, res.Key, res.Cookie, res.Vid)
		err = errors.ErrInternal
		return
	}
}
if res.Ret == errors.RetNeedleExist {
	err = errors.ErrNeedleExist
}
log.Infof("bfs.upload bucket:%s filename:%s key:%d cookie:%d vid:%d", bucket, filename, res.Key, res.Cookie, res.Vid)
return

2017/07/04 18:10:42 Connected to 172.20.2.92:2181 2017/07/04 18:10:42 Authenticated: id=242280265568797060, timeout=6000 2017/07/04 18:10:42 Re-submitting 0` credentials after reconnect
panic: runtime error: invalid memory address or nil pointer dereference

[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x768225]
goroutine 1 [running]:
main.NewServer(0xc420020cc0, 0xc42001e120, 0x0, 0x0, 0x0)
/home/xxx/go/src/bfs/store/http.go:36 +0x45
main.main()
/home/xxx/go/src/bfs/store/main.go:35 +0x2f3`

如上，使用https://github.com/Terry-Mao/bfs/blob/master/doc/store.md#config的默认配置，出现上面的错误，尝试下面的配置依然报上面的错误

[limit] [read] rate = 150.0 brust = 200 [write] rate = 150.0 brust = 200 [delete] rate = 150.0 brust = 200

"bfs/libs/errors" 应该 “github.com/Terry-Mao/bfs/libs/errors” 类似的问题，使得项目无法正常编译

辛苦处理一下，谢谢～～

在directory 中的STORE GROUP VOLUME内存信息 在syncZookeeper go route 中定时更新，和在应用操作api 的go route 中使用这些公共内存资源 不需要加锁么？

@Terry-Mao

想了解在实际操作的过程中，BFS是如何初始化机器的。多谢！

问一下 和B站 的 bfs 是同一个吗？？

add empty file process (errors.RetNeedleIsEmpty)

    if size > int64(h.c.NeedleMaxSize) {
        res["ret"] = errors.RetNeedleTooLarge
        return
    }
    if size == 0 {
        res["ret"] = errors.RetNeedleIsEmpty
        return
    }

I got error when building bfs/store

github.com/Terry-Mao/bfs/store/os

./fadvise_linux.go:26: cannot use C.__off_t(off) (type C.__off_t) as type C.off_t in argument to _Cfunc_posix_fadvise
./fadvise_linux.go:26: cannot use C.__off_t(size) (type C.__off_t) as type C.off_t in argument to _Cfunc_posix_fadvise

 if errno = int(C.posix_fadvise(C.int(fd), C.__off_t(off), C.__off_t(size), C.int(advise))); errno != 0 {
                err = syscall.Errno(errno)
        }

https://github.com/Terry-Mao/bfs/blob/master/proxy/bfs/bfs.go#L99

这里是遍历store取文件的操作是处于什么考虑呢？

I got error when building bfs/store

github.com/Terry-Mao/bfs/store/os

./fadvise_linux.go:26: cannot use C.__off_t(off) (type C.__off_t) as type C.off_t in argument to _Cfunc_posix_fadvise
./fadvise_linux.go:26: cannot use C.__off_t(size) (type C.__off_t) as type C.off_t in argument to _Cfunc_posix_fadvise

// if change C.__off_t to C.off_t, it works
 if errno = int(C.posix_fadvise(C.int(fd), C.__off_t(off), C.__off_t(size), C.int(advise))); errno != 0 {
                err = syscall.Errno(errno)
        }

distributed file system(small file storage) writen in golang by Terry.

不好意思，我不是来吹毛求疵来了，就是随手提个issue

😺

which one do you want to choose?

目前所有的卷 和 机器 和组信息都存放在zk 里面，如果这些信息要是足够多的话，zk 是不是会成为瓶颈？我对zk 了解得不够深入，无法完全评估，所有想咨询一下你们的考虑。

api文档等好多文档链接失效了

Any steps to reproduce the behavior?

Install the Go Tools

set GOROOT and GOPATH env in ~/.bash_profile

pitchfork健康检查状态异常

访问地址 http://127.0.0.1:9001，报如下错误
Traceback (most recent call last):
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/app.py", line 1836, in call
return self.wsgi_app(environ, start_response)
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/app.py", line 1820, in wsgi_app
response = self.make_response(self.handle_exception(e))
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/app.py", line 1403, in handle_exception
reraise(exc_type, exc_value, tb)
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/app.py", line 1817, in wsgi_app
response = self.full_dispatch_request()
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/app.py", line 1477, in full_dispatch_request
rv = self.handle_user_exception(e)
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/app.py", line 1381, in handle_user_exception
reraise(exc_type, exc_value, tb)
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/app.py", line 1475, in full_dispatch_request
rv = self.dispatch_request()
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/app.py", line 1461, in dispatch_request
return self.view_functionsrule.endpoint
File "/Users/xxx/GoProjects/src/bfs/ops/bfsOps/views/ops.py", line 17, in home
return render_template('index.html')
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/templating.py", line 127, in render_template
return _render(ctx.app.jinja_env.get_or_select_template(template_name_or_list),
File "/Library/Python/2.7/site-packages/Jinja2-2.8-py2.7.egg/jinja2/environment.py", line 851, in get_or_select_template
return self.get_template(template_name_or_list, parent, globals)
File "/Library/Python/2.7/site-packages/Jinja2-2.8-py2.7.egg/jinja2/environment.py", line 812, in get_template
return self._load_template(name, self.make_globals(globals))
File "/Library/Python/2.7/site-packages/Jinja2-2.8-py2.7.egg/jinja2/environment.py", line 774, in _load_template
cache_key = self.loader.get_source(self, name)[1]
File "/Library/Python/2.7/site-packages/Flask-0.10.1-py2.7.egg/flask/templating.py", line 64, in get_source
raise TemplateNotFound(template)
TemplateNotFound: index.html

这两天测试bfs，发现directory模块连接hbase 出现错误如下：
read tcp 192.168.1.23:60518->192.168.1.193:9090: i/o timeout

为定位问题在hbase_test.go中加了两个TestCase，发现从bfs pool 获取的连接会出现此问题，直接用thrift api 则不会
import (
"bfs/directory/conf"
hbase "bfs/directory/hbase/hbasethrift"
"bfs/libs/meta"
"fmt"
"git.apache.org/thrift.git/lib/go/thrift"
"net"
"os"
"testing"
"time"
）
func TestOriginal(t testing.T) {
logformatstr := "----%s\n"
logformatstr := "----%s 用时:%d-%d=%d毫秒\n\n"
logformattitle := "建立连接"
rowkey := "vim.gif"
temptable := "bucket_test"
protocolFactory := thrift.NewTBinaryProtocolFactoryDefault()
transport, err := thrift.NewTSocketTimeout(net.JoinHostPort(HOST, PORT), 10_time.Second)
if err != nil {
fmt.Fprintln(os.Stderr, "error resolving address:", err)
os.Exit(1)
}
client := hbase.NewTHBaseServiceClientFactory(transport, protocolFactory)
if err := transport.Open(); err != nil {
fmt.Fprintln(os.Stderr, "Error opening socket to "+HOST+":"+PORT, " ", err)
os.Exit(1)
}

//client, err := hbasePool.Get()
//if err != nil {
//  fmt.Println(err)
//}
//--------------Exists
logformattitle = "调用Exists方法"
fmt.Printf(logformatstr_, logformattitle)
tmpstartTime := currentTimeMillis()
//
isexists, err := (client.Exists([]byte(temptable), &hbase.TGet{Row: []byte(rowkey)}))
fmt.Printf("rowkey{%s} in table{%s} Exists:%t\t", rowkey, temptable, isexists)
if err != nil {
    fmt.Printf("Exists err:%s\n", err)
}
fmt.Println("")
tmpendTime := currentTimeMillis()
fmt.Printf(logformatstr, logformattitle, tmpendTime, tmpstartTime, (tmpendTime - tmpstartTime))

}

func TestPoll(t *testing.T) {
logformatstr_ := "----%s\n"
logformatstr := "----%s 用时:%d-%d=%d毫秒\n\n"
logformattitle := "建立连接"
rowkey := "vim.gif"
temptable := "bucket_test"

client, err := hbasePool.Get()
if err != nil {
    fmt.Println(err)
}
//--------------Exists
logformattitle = "调用Exists方法"
fmt.Printf(logformatstr_, logformattitle)
tmpstartTime := currentTimeMillis()
//
isexists, err := (client.Exists([]byte(temptable), &hbase.TGet{Row: []byte(rowkey)}))
fmt.Printf("rowkey{%s} in table{%s} Exists:%t\t", rowkey, temptable, isexists)
if err != nil {
    fmt.Printf("Exists err:%s\n", err)
}
fmt.Println("")
tmpendTime := currentTimeMillis()
fmt.Printf(logformatstr, logformattitle, tmpendTime, tmpstartTime, (tmpendTime - tmpstartTime))

}

func currentTimeMillis() int64 {
return time.Now().UnixNano() / 1000000
}

运行结果：
=== RUN TestOriginal
----调用Exists方法
rowkey{vim.gif} in table{bucket_test} Exists:false
----调用Exists方法 用时:1461205438652-1461205438636=16毫秒

--- PASS: TestOriginal (0.04s)
=== RUN TestPoll
----调用Exists方法
rowkey{vim.gif} in table{bucket_test} Exists:false Exists err:read tcp 192.168.1.23:60518->192.168.1.193:9090: i/o timeout

----调用Exists方法 用时:1461205439683-1461205438682=1001毫秒

--- PASS: TestPoll (1.03s)

Store 状态为只读状态，因为129这个volume offset错误引起的，不知道怎么恢复这种情况。下边是日志和相关代码。

I0112 12:13:54.480229   95637 pitchfork.go:202] block: /data/bfshome/store/data/129_0, offset: 4294966850

for _, volume = range volumes {
				if volume.Block.LastErr != nil {
					log.Infof("get store block.lastErr:%s host:%s", volume.Block.LastErr, store.Stat)
					store.Status = meta.StoreStatusFail
					break
				} else if volume.Block.Full() {
					log.Infof("block: %s, offset: %d", volume.Block.File, volume.Block.Offset)
					store.Status = meta.StoreStatusRead
				}
				if err = p.zk.SetVolumeState(volume); err != nil {
					log.Errorf("zk.SetVolumeState() error(%v)", err)
				}
			}

NewCofnig("./store.toml") error(Type mismatch for 'conf.Config.Limit' : Type mismatch for 'conf.Limit.Delete' : Type mismatch for 'conf.Rate.Rate' : Expected float but found 'int64'.)

貌似是更新了config.go 之后，原来的 store.toml 未更新，是否需要一并更新？　

// InitBucket
// todo get data from db
func InitBucket() (b map[string]Bucket, err error) {
var (
item Bucket
)
b = make(map[string]Bucket)
// bucket test
item.BucketName = "test"
item.Property = 0
item.KeyId = "121bce6492eba701"
item.KeySecret = "1eb80603e85842542f9736eb13b7e1"
b["test"] = item

return

}

您好，了解下，在打开super block文件，用于写needles时，为什么没有使用o_append，是考虑的store是单进程不存在并发写的原因吗？谢谢。

func NewSuperBlock(file string, c *conf.Config) (b *SuperBlock, err error) {
b = &SuperBlock{}
b.conf = c
b.File = file
b.closed = false
b.write = 0
b.syncOffset = 0
b.Padding = needle.PaddingSize
if b.w, err = os.OpenFile(file, os.O_WRONLY|os.O_CREATE|myos.O_NOATIME, 0664); err != nil {
log.Errorf("os.OpenFile("%s") error(%v)", file, err)
b.Close()
return nil, err
}
if b.r, err = os.OpenFile(file, os.O_RDONLY|myos.O_NOATIME, 0664); err != nil {
log.Errorf("os.OpenFile("%s") error(%v)", file, err)
b.Close()
return nil, err
}
if err = b.init(); err != nil {
log.Errorf("block: %s init() error(%v)", file, err)
b.Close()
return nil, err
}
return
}