Question

I have been tasked with templating MySQL's my.cnf in an attempt standardize the configuration amongst the slave databases using Puppet. Right now, I'm targeting the innodb settings. Are there configuration options that can safely be calculated against hardware specifications such as memory, disk and procs?

Was it helpful?

Solution

You need facter.

puppet:/etc/puppet/modules/master/lib/facter$ cat disks.rb
#!/usr/bin/ruby
#require 'facter'

mount = `/bin/mount`
disks=Array.new 
mount.split("\n").each_with_index { | disk,i |
  unless disk.scan(/ext3|simfs|reiserfs|xfs/).empty?
    d=disk.split[2]
    disks.push d
    disks.push ','
 end
}

Facter.add('disks') do
 setcode do
   disks
 end 
end 

` and in puppet.pp i use facts $disks

#add disk check to zabbix
exec { "create_host":
    command => "/bin/echo $fqdn $ipaddress $disks | do_work",
    require => File["/root/ticket"],
    subscribe => File["/root/ticket"],
    refreshonly => true,
}

see "Adding Custom Facts to Facter" on puppet labs.

OTHER TIPS

I'd be tempted to move the calculations into the erb file, for example the key_buffer_size is recommended to be set to 1/4 of the Systems RAM:

set-variable = key_buffer_size=<%= (memorysize.split(' ')[0].to_i * 1024) / 4 -%>M

there is no reason why you couldn't work on other variables available from Facter (number of processors etc) and come up with your own calculations to set other variables as above.

Remember ERB effectively provides a subset of Ruby so almost anything you can do in Ruby can be done in ERB.

puppet have the erb template, erb template can use the facter value ,like hostname or memor. and you can write you self facter shell script.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top