We are in the process of deploying puppet for ~2500 servers. The main reason why we even went with puppet was due to the fact that it supports external node lookups. As the previous poster mentioned, defining node configs for even 100 machines is already painful, not to mention several thousand. By utilizing external node lookups, we were able to avoid having any node definitions in our manifests and directly integrate everything with our internal server database. This includes adding specific fields to our existing database which specifies which classes a node is supposed to use.
The possibilities for customization are truly endless at that point - version control is easily implementable as well, as you can modify your front end to keep historical data on what classes were removed/added for a specific node. Implementing something like this for sub 20 servers is certainly overkill, however for large deployments it is one of the most attractive points about puppet. Thanks, alasla...@gmail.com wrote: > I'm not necessarily selling it. Use whatever works best for you. But > consider: > > - scale: if you have 20 nodes, I don't think its a big deal to > maintain all the nodes in a .pp file, but what if you have 200, or > 2000 ( or more ). Having the ability to group nodes in some external > tool ( ldap, sql, etc ) allows you to deal with large numbers of nodes > more easily > - node roles - for the most part, don't assign more than one class to > a host using external nodes. This mostly negates the version control > issue, since any changes are usually made to the role classes > themselves, rather than to the host. > - integration - easier to integrate information about hosts and host > roles with other tools if it is in a generically accessible location > > If I didn't need the integration and the scale, then perhaps I > wouldn't have gone down the external nodes path. > > > On Sun, Oct 4, 2009 at 3:36 AM, Douglas Garstang > <doug.garst...@gmail.com> wrote: >> >> So... >> >> I've got a question post puppet camp. I'm not sure I understand the >> benefit of external nodes and storeconfigs. >> >> If you put node data or other puppet configuration in external >> sources, it would seem to me that: >> >> 1. Your complicating your configuration by essentially putting your >> puppet config in two different locations, and >> 2. Most importantly, your losing the ability to version control your >> configuration. It would seem to me that of utmost importance in any >> prod environment is the ability to diff incremental config pushes, and >> maybe roll them back where necessary. >> >> Am I missing a benefit that over rides disadvantages 2 and 3? >> >> Doug. >> >> >> >> > > > --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---
