storage_do_download_file_ex 下载文件失败

laoying11112

[2012-02-10 17:04:26] ERROR - file: ../tracker/tracker_proto.c, line: 125, tracker server: 192.168.10.210:23000, recv data fail, errno: 110, error info: Connection timed out


下载第n个文件时出错！storage只发送了部分数据就不发了，接收超时！
请版主看看！谢谢

laoying11112

3.06版本

laoying11112

pstack 了一下 storage


pstack 10811
Thread 9 (Thread 0x4089a940 (LWP 10814)):
#0  0x0000003ed34d3368 in epoll_wait () from /lib64/libc.so.6
#1  0x00002b2e9f2252c0 in ?? () from /usr/lib64/libevent-1.4.so.2
#2  0x00002b2e9f2184d1 in event_base_loop () from /usr/lib64/libevent-1.4.so.2
#3  0x000000000042789d in work_thread_entrance ()
#4  0x0000003ed4006367 in start_thread () from /lib64/libpthread.so.0
#5  0x0000003ed34d2f7d in clone () from /lib64/libc.so.6
Thread 8 (Thread 0x4091b940 (LWP 10815)):
#0  0x0000003ed34d3368 in epoll_wait () from /lib64/libc.so.6
#1  0x00002b2e9f2252c0 in ?? () from /usr/lib64/libevent-1.4.so.2
#2  0x00002b2e9f2184d1 in event_base_loop () from /usr/lib64/libevent-1.4.so.2
#3  0x000000000042789d in work_thread_entrance ()
#4  0x0000003ed4006367 in start_thread () from /lib64/libpthread.so.0
#5  0x0000003ed34d2f7d in clone () from /lib64/libc.so.6
Thread 7 (Thread 0x4099c940 (LWP 10816)):
#0  0x0000003ed34d3368 in epoll_wait () from /lib64/libc.so.6
#1  0x00002b2e9f2252c0 in ?? () from /usr/lib64/libevent-1.4.so.2
#2  0x00002b2e9f2184d1 in event_base_loop () from /usr/lib64/libevent-1.4.so.2
#3  0x000000000042789d in work_thread_entrance ()
#4  0x0000003ed4006367 in start_thread () from /lib64/libpthread.so.0
#5  0x0000003ed34d2f7d in clone () from /lib64/libc.so.6
Thread 6 (Thread 0x40a1d940 (LWP 10817)):
#0  0x0000003ed34d3368 in epoll_wait () from /lib64/libc.so.6
#1  0x00002b2e9f2252c0 in ?? () from /usr/lib64/libevent-1.4.so.2
#2  0x00002b2e9f2184d1 in event_base_loop () from /usr/lib64/libevent-1.4.so.2
#3  0x000000000042789d in work_thread_entrance ()
#4  0x0000003ed4006367 in start_thread () from /lib64/libpthread.so.0
#5  0x0000003ed34d2f7d in clone () from /lib64/libc.so.6
Thread 5 (Thread 0x40a9e940 (LWP 1081):
#0  0x0000003ed3499761 in nanosleep () from /lib64/libc.so.6
#1  0x0000003ed3499584 in sleep () from /lib64/libc.so.6
#2  0x000000000041b342 in tracker_report_thread_entrance ()
#3  0x0000003ed4006367 in start_thread () from /lib64/libpthread.so.0
#4  0x0000003ed34d2f7d in clone () from /lib64/libc.so.6
Thread 4 (Thread 0x40b1f940 (LWP 10819)):
#0  0x0000003ed3499761 in nanosleep () from /lib64/libc.so.6
#1  0x0000003ed3499584 in sleep () from /lib64/libc.so.6
#2  0x00000000004142e9 in sched_thread_entrance ()
#3  0x0000003ed4006367 in start_thread () from /lib64/libpthread.so.0
#4  0x0000003ed34d2f7d in clone () from /lib64/libc.so.6
Thread 3 (Thread 0x40ba0940 (LWP 10820)):
#0  0x0000003ed400a899 in pthread_cond_wait@@GLIBC_2.3.2 ()
#1  0x000000000042dbbc in dio_thread_entrance ()
#2  0x0000003ed4006367 in start_thread () from /lib64/libpthread.so.0
#3  0x0000003ed34d2f7d in clone () from /lib64/libc.so.6
Thread 2 (Thread 0x40c21940 (LWP 10821)):
#0  0x0000003ed400a899 in pthread_cond_wait@@GLIBC_2.3.2 ()
#1  0x000000000042dbbc in dio_thread_entrance ()
#2  0x0000003ed4006367 in start_thread () from /lib64/libpthread.so.0
#3  0x0000003ed34d2f7d in clone () from /lib64/libc.so.6
Thread 1 (Thread 0x2b2e9f42e7e0 (LWP 10811)):
#0  0x0000003ed400d4eb in accept () from /lib64/libpthread.so.0
#1  0x00000000004272ac in storage_accept_loop ()
#2  0x000000000040a63c in main ()

happy_fish100

回复 3# laoying11112

libevent版本是多少？请确认安装的是stable版本。

laoying11112

现在用了libevent-2.0.16-stable.tar.gz，还是出错！
[2012-02-13 10:28:43] ERROR - file: ../tracker/tracker_proto.c, line: 125, tracker server: 192.168.10.210:23000, recv data fail, errno: 110, error info: Connection timed out

laoying11112

多个线程读取反而没出错，单线程读取出错，大概有3万个文件，每次读到几千个文件就停止了

laoying11112

看了下日志：

[2012-02-14 09:46:10] ERROR - file: storage_nio.c, line: 404, client ip: 192.168.10.211, recv failed, errno: 32, error info: Broken pipe

是这个错误！

laoying11112

也许是我g_fdfs_network_timeout这个值设置小了，2M的文件，高并发传输超时导致的！

laoying11112

设置network_timeout很大也没用，不知道为什么会所有的线程都停了，
我再启动读的时候，把线程给激活了，storage继续写对端已经关闭的socket，出现Broken pipe错误了

happy_fish100

回复 9# laoying11112

出现这种诡异的问题，先检查一下你的环境。
确认一下用的libevent确实是你所看到的stable版本。
实在不行，可以将libevent降级到v1.4.14b试试