#mysqld

mysqld — The MySQL Server

mysqld, also known as MySQL Server, is the main program that does most of the work in a MySQL installation. mysqld, is MySQL server daemon program which runs quietly in background on MySQL Database Server. MySQL Server manages access to the MySQL data directory that contains databases and tables. The data directory is also the default location for other information such as log files and status…

Solution for the error below when connecting to MySQL daemon

ERROR 2002 (HY000): Can't connect to local MySQL server through socket '/tmp/mysql.sock' (2)

Please note that:

This operation will delete all data in MySQL.

Root password will be cahnged and be shown on console

서버의 하드디스크 파일이 깨져 아래와 같은 메시지가 주기적으로 발생 했다.

하드가 문제인지 알았는데 wordpress와 phpbb3 도 동작 하지 않는다.

알고 보니 mysql 파일 중 innoDB 파일 ibdata1 이 깨졌다.

[ 8027.308041] ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x0 [ 8027.308046] ata1.00: BMDMA stat 0x25 [ 8027.308050] ata1.00: failed command: READ DMA EXT [ 8027.308057] ata1.00: cmd 25/00:08:28:93:c8/00:00:46:00:00/e0 tag 0 dma 4096 in […

Recently, I met a valgrind bug. I have struggled for more than one week but got nothing. It obsessed me because the deadline is close. Finally, I got the reason why caused such an error. 

Now, I recorded what I have done so that I can avoid the same error during later work.

The valgrind error is like the follwings:

64 bytes in 1 blocks are definitely lost in loss record 1 of 1 a: malloc (vg_replace_malloc.c:291) b: my_malloc (my_malloc.c:192) b: String::real_alloc(unsigned long) (sql_string.cc:48) b: Item_func::fix_fields (item_func.cc:244) b: Item_str_func::fix_fields (item_strfunc.cc:101) b: unpack_gcol_info_from_frm (table.cc:2626) b: open_table_from_share (table.cc:3126) b: ha_create_table (handler.cc:4759) b: mysql_alter_table (sql_table.cc:9096) b: Sql_cmd_alter_table::execute (sql_alter.cc:324) b: mysql_execute_command (sql_parse.cc:4534) b: mysql_parse (sql_parse.cc:5245) b: dispatch_command (sql_parse.cc:1262) b: do_command (sql_parse.cc:815) b: handle_connection (connection_handler_per_thread.cc:301) b: pfs_spawn_thread (pfs.cc:2213) b: start_thread (pthread_create.c:309) b: clone (clone.S:115) LEAK SUMMARY:   definitely lost: 64 bytes in 1 blocks   indirectly lost: 0 bytes in 0 blocks     possibly lost: 0 bytes in 0 blocks   still reachable: 0 bytes in 0 blocks        suppressed: 0 bytes in 0 blocks ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 2 from 2)

At first, according to the stack, the reason is one Item object is not freed. Most of time was spent to verify when the Item is freed. I used gdb and checked the free_items(This is the function in Mysql which free the Items during sql parse) function to see whether it's freed or not.

During such a process, I found there are many placed where need breakpoints. But I still kept on. I remembered each of the Item address and checked whether it’s freed or not. It’s boring and easy to miss something.

After the above try, I found I should add a deconstruct function where I freed the allocated memory. This try still failed.

I backed to the orignal idea which Item is not freed. So I add some DBUG_PRINT function in the construct and deconstruct function. By this information, I can easily observe which objects are constructed and deconstructed. After this change, the reason caused such an error showed.  I found one Object was not deconstructed by the debug information. (Note, in order to get the debug output, you should add --debug=d:i:o,/tmp/mysqld.log option when start mysqld. By grep, I got the information I want.)

 sudo service mysqld start

Timeout error occurred trying to start MySQL Daemon.

Starting MySQL:                                            [FAILED]

Logs /var/log/mysqld :

mysqld  Error: unable to create temporary file; errno: 28

Can't start server : Bind on unix socket: No space left on device

1. Check /tmp for permissions and space availability

2. df -h

Filesystem            Size  Used Avail Use% Mounted on

/dev/xvd               37G   35G  0M    100% /

...

tmpfs                 2.0G     0  2.0G   0% /dev/shm

Now, /tmp is under / which has 0 M available and is 100% used.

3. ls -lrt /tmp . Its lined up with loads of RackMultipart files

Reason: Paperclip Issue 

Check the no. of such files 

> ls -lrt /tmp/RackMultipart* | wc -l

> 297

4. Remove these files 

rm -f /tmp/RackMultipart*

5. Try again

sudo service mysqld start

Starting MySQL:                                            [  OK  ]

sudo service mysqld status

mysqld (pid 17048) is running...