-------------------------------------------------------------------------------------------------正文---------------------------------------------------------------------------------------------------------------
场景 :
Crash发生时的数据库版本: MySQL-5.7.12, 官方标注在5.7.17进行了fix;
开启半同步的主从架构中, 从库开启半同步, 启动/重启slave线程导致Master实例Crash;
结论 :
mysql bug, 附上bug单链接: https://bugs.mysql.com/bug.php?id=79865
问题描述(摘抄):
主要问题就出在tcp连接的select方法, 通常, 操作系统通过宏FD_SET_SIZE来声明一个进程中select能操作的文件描述符的最大数据, 然而通常情况下, 这个FD_SET_SIZE的值仅为1024;
实际上, 用epoll或者poll会比较少, select貌似是用的很少的;
问题复现:
准备一套MySQL-5.7.12的主从架构, 开启半同步:
为了能尽量简单的启用大量的文件描述符, 这里利用MyISAM分区表的"特性";
这时候在主库上连续执行select语句多次(>5);
这时候看一下主库的文件描述符数量;
那么现在在开启半同步的从库上重启一下slave, 同时tail一下主库的日志;
在重启线程几秒钟之后, 主库就发生了Crash;
PS: 在测试的过程中, 多次执行了select语句, 然后确认主库的半同步状态也是ON的情况下迅速在从库上重启slave, 基本是必现的;
PPS: MyISAM表在open的时候会同时打开所有的分区文件, 所以能比较方便的模拟占用大量文件描述符的情景;
(MyISAM分区表: http://blog.itpub.net/29510932/viewspace-2134679/)
PPPPPPPS: _(:з」∠)_
附上测试用的脚本与Crash的信息
点击(此处)折叠或打开
CREATE TABLE `myisam_t` (
`id` int(11) DEFAULT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4
/*!50100 PARTITION BY HASH (id)
PARTITIONS 2000 */
2017-04-28T22:10:00.731611+08:00 5092 [Note] Start binlog_dump to master_thread_id(5092) slave_server(13043), pos(, 4)
2017-04-28T22:10:01.648365+08:00 5092 [Note] Start semi-sync binlog_dump to slave (server_id: 13043), pos(, 4)
*** buffer overflow detected ***: /usr/sbin/mysqld terminated
======= Backtrace: =========
/lib/x86_64-linux-gnu/libc.so.6(+0x731af)[0x7fcdfc7981af]
/lib/x86_64-linux-gnu/libc.so.6(__fortify_fail+0x37)[0x7fcdfc81dcf7]
/lib/x86_64-linux-gnu/libc.so.6(+0xf6f10)[0x7fcdfc81bf10]
/lib/x86_64-linux-gnu/libc.so.6(+0xf8c67)[0x7fcdfc81dc67]
/usr/lib/mysql/plugin/semisync_master.so(_ZN12Ack_receiver17get_slave_socketsEP6fd_set+0x83)[0x7fcc73d4a493]
/usr/lib/mysql/plugin/semisync_master.so(_ZN12Ack_receiver3runEv+0x603)[0x7fcc73d4aaf3]
/usr/lib/mysql/plugin/semisync_master.so(ack_receive_handler+0x19)[0x7fcc73d4aba9]
/usr/sbin/mysqld(pfs_spawn_thread+0x1b4)[0xe90784]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x80a4)[0x7fcdfdf650a4]
/lib/x86_64-linux-gnu/libc.so.6(clone+0x6d)[0x7fcdfc80d87d]
14:10:01 UTC - mysqld got signal 6 ;
This could be because you hit a bug. It is also possible that this binary
or one of the libraries it was linked against is corrupt, improperly built,
or misconfigured. This error can also be caused by malfunctioning hardware.
Attempting to collect some information that could help diagnose the problem.
As this is a crash and something is definitely wrong, the information
collection process might fail.
key_buffer_size=8388608
read_buffer_size=131072
max_used_connections=5
max_threads=9999
thread_count=8
connection_count=2
It is possible that mysqld could use up to
key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 21899362 K bytes of memory
Hope that's ok; if not, decrease some variables in the equation.
Thread pointer: 0x0
Attempting backtrace. You can use the following information to find out
where mysqld died. If you see no messages after this, something went
terribly wrong...
stack_bottom = 0 thread_stack 0x40000
/usr/sbin/mysqld(my_print_stacktrace+0x2c)[0xe77fec]
/usr/sbin/mysqld(handle_fatal_signal+0x459)[0x7a7019]
/lib/x86_64-linux-gnu/libpthread.so.0(+0xf8d0)[0x7fcdfdf6c8d0]
/lib/x86_64-linux-gnu/libc.so.6(gsignal+0x37)[0x7fcdfc75a067]
/lib/x86_64-linux-gnu/libc.so.6(abort+0x148)[0x7fcdfc75b448]
/lib/x86_64-linux-gnu/libc.so.6(+0x731b4)[0x7fcdfc7981b4]
The manual page at http://dev.mysql.com/doc/mysql/en/crashing.html contains
information that should help you find out what is causing the crash.