> At first blush it doesn't provide an elegant dsl as capistrano does
Hmm, I love DSL, but in this I'm not sure that there's need for it in
this case. There are only couple of methods, it meant to be simple,
DSL will introduce additional layer of abstraction & complexity.

> and I didn't see any direct way that you handle roles.
It does support 'tags', I believe that this is more universal and
explicit term for this purpose, but basically it's the same as
Capistrano roles.
More details - the Deployment/Administration stuff is actually mix of
3 distinct activities: working with remote file system, working with
remote console & processes, managing lot's of settings & configs.
Capistrano does it all in one solution. I split it into 3 distinct
components:
- Vfs - working with remote file system, any system (currently it
support local & ssh, but it can handle any type of FS, like Hadoop or
Amazon S3)
- Vos - working with remote console & processes 
https://github.com/alexeypetrushin/vos
- ClusterManagement (tags are here) - managing configuration and
dependencies https://github.com/alexeypetrushin/cluster_management

And it undestand dependencies, if you run 'rake deploy' on the clean
box - it first install & configres ruby, nginx, database, then run
database and nginx processes and only after it begin to deploy
application.

> And capistrano can be made to do anything, after all it's just a framework
Thanks for link, I would like to compare how both of them are managing
MongoDB:
- using Capistrano: 
https://github.com/donnoman/cap-recipes/tree/master/lib/cap_recipes/tasks/mongodb
- using Vfs: 
https://github.com/alexeypetrushin/my_cluster/blob/master/lib/services/mongodb.rb

Vfs version is a little bigger because there's also backup/restore
logic. Here's how You can use it:
   cluster.services.mongodb.install (start/stop/backup/restore)

> Your example doesn't imply that it runs in parallel on all boxes.
You are right, it doesn't (and by the way there's a notion in
documentations that right now it's supposed to be used for small
cluster only https://github.com/alexeypetrushin/cluster_management).
Parallel execution is very important feature but it's impossible to do
all in one step, it's planned for the next version.

> I think there's a lot you could learn from capistrano and implement into your 
> framework.
Agreed, actually I used it in past, and also other great tool - Rush

> However you should create your own Google group for it so that conversations 
> regarding your framework can be carried on there. The capistrano group is the 
> wrong place for it.
Hmm, agreed, You are right. I thought that maybe someone may be
interested in alternative and beacuse of it I posted it here.

And, thanks for link to cap-recipes.

-- 
* You received this message because you are subscribed to the Google Groups 
"Capistrano" group.
* To post to this group, send email to [email protected]
* To unsubscribe from this group, send email to 
[email protected] For more options, visit this group at 
http://groups.google.com/group/capistrano?hl=en

Reply via email to