Hi, We are using puppet to deploy our application. On each server, we can have several instances of the application. For example: An application is a tomcat webapp. We can deploy the webapp on several tomcat instances, and we also can deploy the webapp several times on each tomcat (to have one instance by country for example).
So, each application is composed by: - a rpm package, - some configuration files that are the same for all instances - some crontabs. - and the definition of the instances. So we have write a puppet class for each application. and the class call the installation of an instance using a "define". This works but if something goes wrong we are enable to know what! if puppet have to install 5 applications, the only information we can retrieve is that something fails. And so, we cannot know the state of the machine. Which instances have been installed and so. So it's impossible to try to recover the machine automatically. The other problem with this approach is that the puppet classes are now very huge and each resource need to have a require on all the resources that have to be performed before (order is random...). This make the puppet classe unreadable, and the debugging unreadable. So my question is: Should we write a custom type to install an application ? I doesn't have read a lot about this, but it seems that we will be able to make the installation atomically (and so we will be able to know the state of the machine) using a custom type ? Is this a good approach of our problem ? Benoit --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Puppet Users" group. To post to this group, send email to puppet-users@googlegroups.com To unsubscribe from this group, send email to puppet-users+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en -~----------~----~----~----~------~----~------~--~---
