Help with setting facts for MySQL replication

[Matthew Macdonald-Wallace]

Hi all,

I'm trying to work on a solution to setting up mysql in a semi-automated
fashion using facts to populate a puppet template.

I'm using Cobbler as my build system and I was hoping to pass the values
needed for replicate_do_db and server_id as ksmeta information, however
it's looking increasingly unlikely that will work.

Can anyone share how they configure MySQL for this kind of thing using
puppet given that multiple servers may be replicating different
databases and all servers require a unique ID?

Thanks in advance,

Matt
--
Matthew Macdonald-Wallace
li...@truthisfreedom.org.uk
http://www.threedrunkensysadsonthe.net/

[Tore]

I use cobbler and I push certain information to the kickstarting
system via ksmeta, e.g.:

cobbler system add --name=rhel-32bit --mac=XX:XX:XX:XX:XX:XX [...] --
ksmeta="swap=256 puppet=true"

This example allows us to define the swap size and if puppet should be
installed and configured. We use the last part alot to allow people to
deploy a test node which have a lifespan of <7 days.

Is this helpful?

On 27 Jul, 23:27, Matthew Macdonald-Wallace

[Disconnect]

I use a custom fact in modules/mysql/lib/facter/server_id.rb:
# Converts ip address to long for mysql id
def get_mysql_id
    mysql_id = nil;
    mysql_id = Facter.ipaddress.split('.').inject(0) {|total,value| (total << 8 ) + value.to_i}
end

Facter.add("mysql_server_id") do
    setcode do
        get_mysql_id
    end
end

Then the mysql_id is used by the template as the node id.

--
You received this message because you are subscribed to the Google Groups "Puppet Users" group.
To post to this group, send email to puppet...@googlegroups.com.
To unsubscribe from this group, send email to puppet-users...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en.

[Marc Fournier]

Hello,

> Can anyone share how they configure MySQL for this kind of thing using
> puppet given that multiple servers may be replicating different
> databases and all servers require a unique ID?

This all boils down to each host having different options in it's
my.cnf file.

There is an official augeas lens for editing mysql's my.cnf file since a
couple of days (not in any release yet).

If this can be of any help, I used this code for a basic "1 master-1
slave" setup on a few projects:

class mysql::master inherits mysql::server {

augeas { "my.cnf/replication":
context => "/files/etc/my.cnf/mysqld/",
changes => [
"set log-bin mysql-bin",
"set server-id ${mysql_serverid}",
"set expire_logs_days 7",
"set max_binlog_size 100M"
],
}

}

class mysql::slave inherits mysql::master {

augeas { "my.cnf/slave-replication":
context => "/files/etc/my.cnf/mysqld/",
changes => [
"set relay-log /var/lib/mysql/mysql-relay-bin",
"set relay-log-index /var/lib/mysql/mysql-relay-bin.index",
"set relay-log-info-file /var/lib/mysql/relay-log.info",
"set relay_log_space_limit 2048M",
"set max_relay_log_size 100M",
"set master-host ${mysql_masterhost}",
"set master-user ${mysql_masteruser}",
"set master-password ${mysql_masterpw}",
"set report-host ${hostname}"
],
}
}

I used a home-brew augeas lens though. I'm not sure the paths are
exactly the same with the official one.

Cheers,
Marc

[Charlotte McLaughlin]

I put server_id.rb at
/etc/puppetlabs/puppet/environments/qa/ecm-puppet/ecm/lib/facter/server_id.rb

how do i build on this to create a master-slave replication db01.xxx.xxx(master) db02.xxx.xxx(slave).

Thanks

Mac