发表新帖

swoole 1.7.5压测到一定量之后服务器端主进程状态变为Rsl，请求无法再响应

首先，我的客户端压测脚本是@Range 的bench.php 修改的，基本上保证了原汁原味，没有大的改动
压测命令：
{{{
./bench.php -c 1000 -n 10000 -s tcp://ip.14:9001 -f short_tcp
}}}

服务器端使用strace命令追踪程序，输出

{{{
27649 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27648 msgsnd(1572912, {18, "L\0\0\0?\0\0\0\0\0\0\0\0\0\0;{\"op\":\"role.crea"...}, 75, IPC_NOWAIT <unfinished ...>
27577 <... sched_yield resumed> )       = 0
27651 sched_yield( <unfinished ...>
27650 msgsnd(1572912, {16, "J\0\0\0?\0\2\0\0\0\0\0\0\0\0;{\"op\":\"role.crea"...}, 75, IPC_NOWAIT <unfinished ...>
27649 sched_yield( <unfinished ...>
27577 msgsnd(1572912, {5526, "P\0\0\0\0\0\0\0\6E`\377"}, 12, IPC_NOWAIT <unfinished ...>
27651 <... sched_yield resumed> )       = 0
27650 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27649 <... sched_yield resumed> )       = 0
27577 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27651 msgsnd(1572912, {17, "K\0\0\0?\0\3\0\0\0\0\0\0\0\0;{\"op\":\"role.crea"...}, 75, IPC_NOWAIT <unfinished ...>
27650 sched_yield( <unfinished ...>
27649 msgsnd(1572912, {19, "M\0\0\0?\0\1\0\0\0\0\0\0\0\0;{\"op\":\"role.crea"...}, 75, IPC_NOWAIT <unfinished ...>
27648 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27577 sched_yield( <unfinished ...>
27651 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27650 <... sched_yield resumed> )       = 0
27649 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27648 sched_yield( <unfinished ...>
27577 <... sched_yield resumed> )       = 0
27651 sched_yield( <unfinished ...>
27650 msgsnd(1572912, {16, "J\0\0\0?\0\2\0\0\0\0\0\0\0\0;{\"op\":\"role.crea"...}, 75, IPC_NOWAIT <unfinished ...>
27649 sched_yield( <unfinished ...>
27648 <... sched_yield resumed> )       = 0
27577 msgsnd(1572912, {5526, "P\0\0\0\0\0\0\0\6E`\377"}, 12, IPC_NOWAIT <unfinished ...>
27651 <... sched_yield resumed> )       = 0
27650 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27649 <... sched_yield resumed> )       = 0
27648 msgsnd(1572912, {18, "L\0\0\0?\0\0\0\0\0\0\0\0\0\0;{\"op\":\"role.crea"...}, 75, IPC_NOWAIT <unfinished ...>
27577 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27651 msgsnd(1572912, {17, "K\0\0\0?\0\3\0\0\0\0\0\0\0\0;{\"op\":\"role.crea"...}, 75, IPC_NOWAIT <unfinished ...>
27650 sched_yield( <unfinished ...>
27649 msgsnd(1572912, {19, "M\0\0\0?\0\1\0\0\0\0\0\0\0\0;{\"op\":\"role.crea"...}, 75, IPC_NOWAIT <unfinished ...>
27648 <... msgsnd resumed> )            = -1 EAGAIN (Resource temporarily unavailable)
27577 sched_yield( <unfinished ...>
}}}

大概到5k左右的请求时，主进程状态就是Rsl了
切换回1.7.4就没有问题了

我不懂这个是因为什么原因

发布于10年前 · 1 次浏览 · 来自提问

泰

泰迪熊～

服务器端使用strace命令追踪程序，输出

大概到5k左右的请求时，主进程状态就是Rsl了
切换回1.7.4就没有问题了

我不懂这个是因为什么原因