On 05/07/2013 06:28 AM, Alex Jurkiewicz wrote:
Hi all,
I'm working to migrate our ~500 DNS records from bind flat files to
individual route53 resources. Each route53 resource definition is very
slow -- checking a record exists requires 2 API lookups and adding
requires 3, which means a resource can take >5seconds to run. This
implies a massive slowdown for Puppet -- our current catalogs finish in
90seconds and this migration will add 40 minutes(!) to that.
I can probably cache more to speed up these resources, but a significant
increase in Puppet run-time seems inevitable. I'm worried that longer
run-times are going to be a problem as I move more and more system
config from file resources to custom Puppet types.
How are people dealing with this sort of slowdown? Do I live with it, or
is there a solution?
I had the same issue with my own 'cobblersystem' type. Each parameter
had at least 1 API or cli call.
After that I changed the provider to use prefetching. That means 1 API
call to get all the systems with all the parameters at once... Thing got
~9 times faster.
I suggest you to rewrite your provider also...
--
Jakov Sosic
www.srce.unizg.hr
--
You received this message because you are subscribed to the Google Groups "Puppet
Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/puppet-users.
For more options, visit https://groups.google.com/groups/opt_out.
