percona toolkit Replication filters error

Question

I put the percona toolkit onto my DB hosts so I could try and deal with a problem with mysql going silently out of sync. That is replication seems fine on all nodes. Slave IO running / Slave SQL running and 0 seconds behind master.

I have 4 dbs setup in master/master on the first two, and two slaves, I'm using MariaDB-server-10.0.21 for the MySQL database on each node.

Yet the content of the wiki I run on them seems to go out of sync even with those positive indicators. For instance, you'll create a page, save it, get the thumbs up from the wiki. Then reload the page and the content will be gone! Then you point the wiki config to look at each db one at a time, reload the page. Until you find the db that saved the changes you made.

Then dump that db, stop the slaves on each host one at time and then import that version of the database. It's a real pain!

So I installed the percona toolkit after reading an article on how to solve this problem.

And when I run the pt-table-checksum command I get this error, saying Replication filters are set on these hosts:

[root@db1:~] #pt-table-checksum --replicate=test.checksum --databases=sean --ignore-tables=semaphore localhost
10-17T00:31:11 Replication filters are set on these hosts:
  db3
    binlog_do_db = jfwiki,jokefire,bacula,mysql
  db2
    binlog_do_db = jfwiki,jokefire,bacula,mysql
  db4
    binlog_do_db = jfwiki,jokefire,bacula,mysql
Please read the --check-replication-filters documentation to learn how to solve this problem. at /bin/pt-table-checksum line 9644.

But that EC2 host it claims that it's having trouble contacting equates to my 4th database host. I found out by ssh'ing in as my user to that DNS address. And I have no trouble at all logging into that host on the command line using mysql:

Can someone please explain what does this error mean, and how can I fix the issue? Is there any general advice you can give for mysql replication falling silently out of sync?

Thanks

Answer 1

Some of the pt tools need to create their own database and have it replicated. Your binlog_do_db prevents the extra db from being replicated, hence preventing that tool from working.

While you have the binlog_do removed, see what db it being built. Then add it.