Does RVM "failover" to another ruby instance on error? - ruby

Question

Have a strange problem in that I have a Rake task that seems to be using multiple versions of Ruby. When one fails, it seems to try another one.
Details
MacBook running 10.6.5
rvm 1.1.0
Rubies: 1.8.7-p302, ree-1.8.7-2010.02, ruby-1.9.2-p0
Rake 0.8.7
Gem 1.3.7
Veewee (provisioning Virtual Machines using Opcode.com, Vagrant and Chef)
I'm not entirely sure the specific details of the error matter, but since it might be an issue with Veewee itself. So, what I'm trying to do is build a new box base on a veewee definition. The command fails with an error about a missing method- but what's interesting is how it fails.
Errors
I managed to figure out that if I only have one Ruby installed with RVM, it just fails. If I have more than one Ruby install, it fails at the same place, but execution seems to continue in another interpreter.
Here are two different clipped console outputs. I've clipped them for size. The full outputs of each error are available as a gist.
One Ruby version installed
Here is the command run when I only have a single version of Ruby (1.8.7) available in RVM
boudica:veewee john$ rvm rake build['mettabox'] --trace
rvm 1.1.0 by Wayne E. Seguin (wayneeseguin#gmail.com) [http://rvm.beginrescueend.com/]
(in /Users/john/Work/veewee)
** Invoke build (first_time)
** Execute build
…
creating new harddrive
rake aborted!
undefined method `max_vdi_size' for #<VirtualBox::SystemProperties:0x102d6af80>
/Users/john/.rvm/gems/ruby-1.8.7-p302/gems/virtualbox-0.8.3/lib/virtualbox/abstract_model/dirty.rb:172:in `method_missing'
<------ stacktraces cut ---------->
/Users/john/.rvm/gems/ruby-1.8.7-p302/gems/rake-0.8.7/bin/rake:31
/Users/john/.rvm/gems/ruby-1.8.7-p302#global/bin/rake:19:in `load'
/Users/john/.rvm/gems/ruby-1.8.7-p302#global/bin/rake:19
Multiple Ruby Versions
Here is the same command run with three versions of Ruby available in RVM. Prior to doing this, I used "rvm use 1.8.7." Again, I don't know how important the details of the specific errors are- what's interesting to me is that there are three separate errors- each with it's own stacktrace- and each in a different Ruby interpreter. Look at the bottom of each stacktrace and you'll see that they are all sourced from different interpreter locations- First ree-1.8.7, then ruby-1.8.7, then ruby-1.9.2:
boudica:veewee john$ rvm rake build['mettabox'] --trace
rvm 1.1.0 by Wayne E. Seguin (wayneeseguin#gmail.com) [http://rvm.beginrescueend.com/]
(in /Users/john/Work/veewee)
** Invoke build (first_time)
** Execute build
…
creating new harddrive
rake aborted!
undefined method `max_vdi_size' for #<VirtualBox::SystemProperties:0x1059dd608>
/Users/john/.rvm/gems/ree-1.8.7-2010.02/gems/virtualbox-0.8.3/lib/virtualbox/abstract_model/dirty.rb:172:in `method_missing'
…
/Users/john/.rvm/gems/ree-1.8.7-2010.02/gems/rake-0.8.7/bin/rake:31
/Users/john/.rvm/gems/ree-1.8.7-2010.02#global/bin/rake:19:in `load'
/Users/john/.rvm/gems/ree-1.8.7-2010.02#global/bin/rake:19
(in /Users/john/Work/veewee)
** Invoke build (first_time)
** Execute build
isofile ubuntu-10.04.1-server-amd64.iso is available
["a1b857f92eecaf9f0a31ecfc39dee906", "30b5c6fdddbfe7b397fe506400be698d"]
[]
Last good state: -1
Current step: 0
last good state -1
destroying machine+disks
(re-)executing step 0-initial-a1b857f92eecaf9f0a31ecfc39dee906
VBoxManage: error: Machine settings file '/Users/john/VirtualBox VMs/mettabox/mettabox.vbox' already exists
VBoxManage: error: Details: code VBOX_E_FILE_ERROR (0x80bb0004), component Machine, interface IMachine, callee nsISupports
Context: "CreateMachine(bstrSettingsFile.raw(), name.raw(), osTypeId.raw(), Guid(id).toUtf16().raw(), FALSE , machine.asOutParam())" at line 247 of file VBoxManageMisc.cpp
rake aborted!
undefined method `memory_size=' for nil:NilClass
/Users/john/Work/veewee/lib/veewee/session.rb:303:in `create_vm'
/Users/john/Work/veewee/lib/veewee/session.rb:166:in `build'
/Users/john/Work/veewee/lib/veewee/session.rb:560:in `transaction'
/Users/john/Work/veewee/lib/veewee/session.rb:163:in `build'
/Users/john/Work/veewee/Rakefile:87
/Users/john/.rvm/gems/ruby-1.8.7-p302/gems/rake-0.8.7/lib/rake.rb:636:in `call'
/Users/john/.rvm/gems/ruby-1.8.7-p302/gems/rake-0.8.7/lib/rake.rb:636:in `execute'
/Users/john/.rvm/gems/ruby-1.8.7-p302/gems/rake-0.8.7/lib/rake.rb:631:in `each'
…
/Users/john/.rvm/gems/ruby-1.8.7-p302/gems/rake-0.8.7/bin/rake:31
/Users/john/.rvm/gems/ruby-1.8.7-p302#global/bin/rake:19:in `load'
/Users/john/.rvm/gems/ruby-1.8.7-p302#global/bin/rake:19
(in /Users/john/Work/veewee)
** Invoke build (first_time)
** Execute build
isofile ubuntu-10.04.1-server-amd64.iso is available
["a9c4ab3257e1da3479c984eae9905c2a", "30b5c6fdddbfe7b397fe506400be698d"]
[]
Last good state: -1
Current step: 0
last good state -1
(re-)executing step 0-initial-a9c4ab3257e1da3479c984eae9905c2a
VBoxManage: error: Machine settings file '/Users/john/VirtualBox VMs/mettabox/mettabox.vbox' already exists
VBoxManage: error: Details: code VBOX_E_FILE_ERROR (0x80bb0004), component Machine, interface IMachine, callee nsISupports
Context: "CreateMachine(bstrSettingsFile.raw(), name.raw(), osTypeId.raw(), Guid(id).toUtf16().raw(), FALSE , machine.asOutParam())" at line 247 of file VBoxManageMisc.cpp
rake aborted!
undefined method `memory_size=' for nil:NilClass
/Users/john/Work/veewee/lib/veewee/session.rb:303:in `create_vm'
/Users/john/Work/veewee/lib/veewee/session.rb:166:in `block in build'
/Users/john/Work/veewee/lib/veewee/session.rb:560:in `transaction'
/Users/john/Work/veewee/lib/veewee/session.rb:163:in `build'
/Users/john/Work/veewee/Rakefile:87:in `block in <top (required)>'
/Users/john/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/rake.rb:634:in `call'
/Users/john/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/rake.rb:634:in `block in execute'
…
/Users/john/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/rake.rb:2013:in `top_level'
/Users/john/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/rake.rb:1992:in `run'
/Users/john/.rvm/rubies/ruby-1.9.2-p0/bin/rake:35:in `<main>'
It isn't until we reach the last installed version of Ruby that execution halts.
Discussion
Does anyone have any idea what's going on here? Has anyone seen this "failover"-like behavior before? It seems strange to me that the first exception would not halt execution as it did with one interpreter, but I wonder if there are things happening when RVM is installed that we Ruby developers are not considering.

Answer 1

Since rvm rake is SUPPOSED to run the rake task against all versions of Ruby under RVM's control I think it's doing what is expected: rvm help shows:
rake - runs a rake task against specified and/or all rubies
Specifically requesting help for rake shows:
$ rvm help rake
Rake
RVM allows you to run rake tasks across multiple ruby versions, for example:
∴ rvm 1.8.6,1.9.1 rake spec
JSON Summary
Adding a --json flag prior to the word 'rake' and a JSON summary will be printed out at the end of the run.
YAML Summary
Adding a --yaml flag prior to the word 'rake' and a YAML summary will be printed out at the end of the run.
If you are trying to run a rake task, you do not have to involve RVM. Set your desired Ruby version with RVM first, then run the task:
rvm 1.9.2
rake some_task

Answer 2

the error you are mentioning is not a problem of different ruby vms.
Could it be you have upgrade to Virtualbox 4.X because the new vagrant needs it?
Some functions of the virtualbox on which veewee is dependent are (not yet) ported to Virtualbox 4.x
I'm working around this now by execing the virtualbox commands directly. Funny that while implementing the update, I hit the exact same error.
Patrick (veewee inspirator)