I have a primary DB and two replicates. replicate1 became corrupted when the linux server was rebooted without stopping mysqld. I want to load replicate1 with a dump I've taken from replicate2 with the command
mysqldump --master-data --single-transaction --all-databases --compress > full_dumpfle.sql
dump was started after replication on replicate12 was stopped with
stop slave;
as soon as the dump was initiated, I restarted replication on replicate2 with
start slave;
dump file is 50G in size.
I copied the my.cnf file from replicate4 to replicate1, keeping the server-id from replicate1's previous my.cnf file, along with the master_user and master_password lines.
I load the DB with
mysql < full_dumpfile.sql
I find that the load exhausts the disk, which was sufficient for replicate1 before the unfortunate reboot. I've even added 100G of disk and the load threatens to exhaust that as well.
I truncated the slow_log table which was an enormous 128G and dropped the main database which was another 56G, after one failed load that nearly exhausted the available disk, about 245G is free when I begin the second load.
All I want to do is return replicate1 to service as a replicate of the master. Obviously the method I'm using is not working. Can anyone give me directions on how to load the dump correctly? Not wedded to the dump, can take another. Can even initiate replication on replicate1 from scratch, if need be, but would rather not load the master, if I can help it.
All of this takes place on our private cloud, so keeping the load down is advisable.
All of this is on ubuntu10.04LTS and MaridDB.
Server version: 5.2.12-MariaDB-mariadb115~lucid-log
Any constructive advice would be appreciated.
Raul