Mysql管理必备工具Maatkit详解之十（mk-parallel-restore）

2009年07月27日　作者： 大头刚　

mk-parallel-restore - Load files into MySQL in parallel. 安装方法可以参考。

在介绍了mk-parallel-dump并行备份，会把mysql的按照database模式，每个表都生成一个单独的备份文件,并统一在你指定的目录。
mk-parallel-restore其实就是mk-parallel-dump的反向操作，也会同时启用4个mysql进程去进行恢复，默认优先恢复大表。
使用起来很简单。例如：

mk-parallel-restore -u -p -h /u01/backup/bk_mysql_20090724

如果dump的时候指定了–tab，这里也加上–tab参数

mk-parallel-restore -u -p -h –tab /u01/backup/bk_mysql_20090724

这里需要注意一下，恢复的时候，如果恢复库没有备份库的database需要提前手工创建，mk-parallel-restore不会自动创建。否则会报错

gzip: stdout: Broken pipe
Issuing rollback() for database handle being DESTROY’d without explicit disconnect().
DBD::mysql::db do failed: Unknown database ‘xxxxxx’ at /usr/bin/mk-parallel-restore line 1778.

ok，在看下其他的参数。

mk-parallel-restore --help
mk-parallel-restore loads files into MySQL in parallel.  For more details,
please use the --help option, or try 'perldoc mk-parallel-restore' for complete
documentation.
 
Usage: mk-parallel-restore <options> PATH [PATH...]
 
Options:
  --askpass             Prompt for a password when connecting to MySQL
  --[no]atomicresume    Treat chunks as atomic when resuming restore (default
                        yes)
  --basedir             Directory where FIFO files will be created
  --[no]biggestfirst    Restore the biggest tables first for highest concurrency
                        (default yes)
  --binlog              Set SQL_LOG_BIN to 1 or 0 to enable or disable binary
                        logging (default 1)
  --bulkinsbufsize      Set bulk_insert_buffer_size before each LOAD DATA INFILE
  --charset         -A  Sets the connection, database, and LOAD DATA INFILE
                        character set (default BINARY)
  --commit              Commit after each load via LOAD DATA INFILE
  --createdb            Create databases if they don't exist
  --csv                 Files are in CSV format (implies --tab)
  --database        -D  Load all files into this database
  --databases       -d  Restore only this comma-separated list of databases
  --dbregex             Restore only databases whose names match this regex
  --decompress          Command used to decompress and print .gz files to STDOUT
                        (like zcat). (default gzip -d -c)
  --defaults-file   -F  Only read mysql options from the given file. You must
                        give an absolute pathname.
  --[no]disablekeys     Execute ALTER TABLE DISABLE KEYS before each table
  --[no]fifo            Stream files into a FIFO for --tab (default yes)
  --help                Show this help message
  --host            -h  Connect to host
  --ignore          -i  Adds the IGNORE modifier to LOAD DATA INFILE
  --ignoredb        -g  Ignore this comma-separated list of databases
  --ignoretbl       -n  Ignore this comma-separated list of table names
  --local           -L  Uses the LOCAL option to LOAD DATA INFILE
  --[no]locktables      Lock tables before LOAD DATA INFILE
  --[no]noautovalon0    Set SQL NO_AUTO_VALUE_ON_ZERO before LOAD DATA INFILE
  --[no]noforeignkeys   Set FOREIGN_KEY_CHECKS=0 before LOAD DATA INFILE
  --noresume            Do not resume restore
  --[no]nouniquechecks  Set UNIQUE_CHECKS=0 before LOAD DATA INFILE
  --numthread       -m  Specifies the number of parallel processes to run
  --password        -p  Password to use when connecting
  --port            -P  Port number to use for connection
  --progress            Display progress messages
  --quiet           -q  Sets --verbose to 0
  --replace         -r  Adds the REPLACE modifier to LOAD DATA INFILE
  --setvars             Set these MySQL variables (default wait_timeout=10000)
  --socket          -S  Socket file to use for connection
  --tab             -T  Load tab-separated files with LOAD DATA INFILE
  --tables          -t  Restore only this comma-separated list of table names
  --tblregex            Restore only tables whose names match this regex
  --test                Print commands instead of executing them
  --truncate            Run TRUNCATE TABLE before LOAD DATA INFILE
  --umask               Set the program's umask to this octal value
  --user            -u  User for login if not current user
  --verbose         -v  Verbosity; can specify multiple times (default 1)
  --version             Output version information and exit
  --wait            -w  Wait limit when server is down (default 5m).  Optional
                        suffix s=seconds, m=minutes, h=hours, d=days; if no
                        suffix, s is used.
 
Options and values after processing arguments:
  --askpass             FALSE
  --atomicresume        TRUE
  --basedir             /root/cron
  --biggestfirst        TRUE
  --binlog              1
  --bulkinsbufsize      (No value)
  --charset             BINARY
  --commit              FALSE
  --createdb            FALSE
  --csv                 FALSE
  --database            (No value)
  --databases           (No value)
  --dbregex             (No value)
  --decompress          gzip -d -c
  --defaults-file       (No value)
  --disablekeys         FALSE
  --fifo                TRUE
  --help                TRUE
  --host                (No value)
  --ignore              FALSE
  --ignoredb            
  --ignoretbl           
  --local               FALSE
  --locktables          FALSE
  --noautovalon0        FALSE
  --noforeignkeys       FALSE
  --noresume            FALSE
  --nouniquechecks      FALSE
  --numthread           4
  --password            (No value)
  --port                (No value)
  --progress            FALSE
  --quiet               FALSE
  --replace             FALSE
  --setvars             wait_timeout=10000
  --socket              (No value)
  --tab                 FALSE
  --tables              (No value)
  --tblregex            (No value)
  --test                FALSE
  --truncate            FALSE
  --umask               0
  --user                (No value)
  --verbose             1
  --version             FALSE
  --wait                300