2

Some background: There are 3 database servers - two in a master/master replication config with only one server accepting writes. And one other slave replica replicating from the write-master. The slave-master had low disk space recently (to the point where it ran entirely out of space), so I added an additional disk and moved only the datadir to the new disk. Prior to doing this, the slave master was accepting writes. But I moved the writes over to the other master before moving data.

The strange thing is that the datadir of the slave-master is approximately 175GB larger than the write-master. That is until I run the query:

select table_schema,table_name,data_length from tables order by data_length desc limit 10;

As this query is running the OS reports that the amount of free space is increasing. And ends up around 120GB extra. This happens in the space of a minute. Then immediately after the query finishes the free disk space begins to decrease again.

This is with MySQL v5.6. The majority of tables are MyISAM with a minority of InnoDB tables. A show slave status on the slave-master reports replication continues as normal. The logs dir, which remains on the original disk, does not fluctuate free space wildly like this.

Has any seen this behaviour before? This seems somewhat sinister to me, but I'm by no means an expert. Running the same query on the current write-master doesn't produce this behaviour.

  • MyISAM what? Didn't that guy die? – oNare Jul 15 '15 at 13:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.