Rebuild Slave-DB & Replication

Since the mysqldump command locks the tables, it is not necessary that there is no traffic on the master database. With --master-data the command mysqldump stores the correct position for inserting the replication on the slave server. 

If the hard disk of the slave is full, please refer to the instructions on this page below - "Slave Disk Full".

1. Login to slave server
   

   • Login to MySQL with credentials USER and PWD, then stop slave and leave MySQL again. Use the following commands for this:

     • mysql -uUSER -pPWD

     • STOP SLAVE;

     • RESET SLAVE;

     • QUIT;

   • Restart MySQL Server with service mysqld restart.

2. Write a MySQL dump. Now perform the following steps on the master server. Create a backup directory and change to the same. The MySQL dumb is now executed with the following command

   • mysqldump -uUSER -pPWD --single-transaction --master-data=2 --databases JTELWeb JTELStats JTELLog --add-drop-database --add-drop-table --events --routines --triggers > filename.sql

     As of release 3.12 the following command is required:

   • mysqldump -uUSER -pPWD --single-transaction --master-data=2 --databases JTELWeb JTELStats JTELStats2 JTELLog --add-drop-database --add-drop-table --events --routines --triggers > filename.sql


   Here is a brief explanation of this:

   • <Name LogFile> and <Position LogFile> are saved through --master-data in the dump
   • use only the JTEL databases '--databases'
   • delete all databases before import  '--add-drop-database'
   • delete all tables of the databases before import '--add-drop-table'
   • Implement all routines and procedures in the dump '--events --routines'
   • Additionally the parameter --default_character_set utf8 can be used
     
     

   In versions 3.12, 3.14 and 3.15

   If someone logs on to the portal while the dump is being pulled, it will go wrong. Enclosed a SQL query. If the time changes after executing the query, a login has taken place.

   If this happens, the dump has to be pulled again and in the meantime it has to be permanently checked if a login has taken place. Only if this is not the case, the dump can be replicated error-free to the slave.

   SELECT Max(dtAcdLoggedIn) FROM Users;

   In versions 3.11 and below and version 3.16 this problem does not exist.

3. Copy MySQL dump from master database server to slave database server
   Use the tool of your choice: command line (scp), WinSCP, ...
   
   

4. We switch back to the slave server and now import the mysqldump.

   • mysql -uUSER -pPWD (we are now in MySql)

   • SET foreign_key_checks = 0;

   • drop database JTELLog;

   • drop database JTELStats;

   • drop database JTELWeb;

   • drop database JTELStats2;

   • SET foreign_key_checks = 1;

   • source <filename>;

5. We are now determining the replication parameters, as we will need them in a moment:
   
   • In the mysqldump file at the very beginning we find a line in the comment block that looks like this: CHANGE MASTER TO MASTER_LOG_FILE='binlog.000872', MASTER_LOG_POS=11940974;Memorize the parameters MASTER_LOG_FILE und MASTER_LOG_POS.
   • We need the name or IP of the Master DB Server
   • We need the password of user repl on the Master DB Server. This is usually always the password of the user root - the easiest way to try it is to log on from the slave to the master with this password: mysql -h <masterIP> -u repl -p<Passwort> - if this works, then you know.
6. Now on the slave server there is already a consistent, fairly new state of the data. It is now only necessary to reconfigure the replication. The following commands do this:

   • mysql -uUSER -pPWD

   • CHANGE MASTER TO MASTER_HOST='<name or ip of master servers>',MASTER_USER='repl',MASTER_PASSWORD='<passwort>',MASTER_LOG_FILE='<Name LogFile>', MASTER_LOG_POS=<Position LogFile>;
     Here all parameters must be replaced with those from step 5.

     • START SLAVE;

   Now we check the slave status (a few times) with the command 

   • SHOW SLAVE STATUS\G
   and expect the following result:
   

   • Slave_IO_Running: Yes

   • Slave_SQL_Running: Yes

   Now the replication is running again. The name of the logfile and the position should promptly correspond pretty much to the show master status on the master database.
   
   
7. Do not forget to clean up data garbage! I.e. delete mysqldumps and directories at least on the slave. A backup can remain on the master database, provided there is enough space. 

Slave Disk Full

There are several reasons why the slave disk can become full.

tmp directory is full

Background

Every time a query creates a tmp table, it is written to the temp directory, usually /tmp. This happens when the maximum size exceeds the maximum "in memory" table size. This is defined with the variables tmp_table_size aswell as max_heap_table_size .

See also https://dev.mysql.com/doc/refman/5.6/en/internal-temporary-tables.html for more information. 

The tables in /tmp are kept until the respective DB connection is closed or a DROP TEMPORARY TABLE is called. If the /tmp directory is full, it is likely that a DROP TEMPORARY TABLE is missing somewhere. This can also happen through customer queries to the DB.

The installation of tmpwatch creates help in a permanent way

Hints:

• On CentOS 6 the /tmp directory is then by default freed from files that have not been accessed for > 10 days
  
  • Circumstantially this may not be sufficient
• On CentOS 7 the /tmp directory is then freed by default from files that have not been accessed for> 1 day

Procedure

1. Space must first be made available. Just mercilessly delete everything from /tmp.
2. If this provides enough space, then restart the mysql service first: service mysql restart
3. If there is enough space (at least about 20%), then proceed with replication as described above.

Lots of "relay logs" available

MySQL first writes the relay logs from the master to a file. Once replication is interrupted, but the slave relay process continues to operate, the disk is filled by relay logs. 

This step should be done in any case, especially before the next one (ibdata too large) to make room.

Procedure

The files for the database are usually located in /var/lib/mysql

If not, the location can be found in /etc/my.cnf The corresponding entry is datadir=(pfad)

1. Alle Relay Logs löschen:

   cd /var/lib/mysql
   rm mysqld-relay-bin*

   Restart MySQL service

   service mysqld restart

   If enough space is available (at least 20%) then proceed with the slave recovery as described above.

ibdata files on the slave "very large"

The files for the database are usually located in /var/lib/mysql

Procedure

1. If no less than 100% disk can be achieved by the steps above:
   

   1. Unsubscribe the MySQL service from the autostart:

      service mysqld disable

   2. Restart the computer with :

      reboot

      Only works if less than 100% disk is reached (possibly after reboot above) then

      1. Log on to the mysql server:

         mysql -u root -p

      2. Drop all JTEL databases:

         SET FOREIGN_KEY_CHECKS=0;
         DROP DATABASE JTELLog;
         DROP DATABASE JTELStats;
         DROP DATABASE JTELWeb;
         SET FOREIGN_KEY_CHECKS=1;

         Press CTRL+C to return to the command line, and then

         service mysqld stop
         rm /var/lib/mysql/ibdata*
         rm /var/lib/mysql/ib_log_*

         Start MySql Server, and enable it again if necessary:

         service mysqld start
         service mysqld enable

         Check disk space, and proceed with Slave Restore as described above.

Still no disk space > 20% free

In this case the slave is simply too small. The hard disk must be expanded (as with extending the STORE role, apply only to the logical volume where the MySQL database data resides). Then perform the steps again, if the disk is too small, then restore the slave as described above.

Or the slave is completely rebuilt with a larger plate.