On 07/27/11 - 04:56:14PM, mtaylor@redhat.com wrote:
From: Martyn Taylor mtaylor@redhat.com
src/app/util/condormatic.rb | 2 ++ src/config/condor.yml | 1 + 2 files changed, 3 insertions(+), 0 deletions(-) create mode 100644 src/config/condor.yml
diff --git a/src/app/util/condormatic.rb b/src/app/util/condormatic.rb index b52ff46..45aa4bd 100644 --- a/src/app/util/condormatic.rb +++ b/src/app/util/condormatic.rb @@ -29,6 +29,7 @@ class Possible @hwp = hwp @provider_image = provider_image @realm = realm
- @condor_config = YAML.load_file("#{RAILS_ROOT}/config/condor.yml") end
end
@@ -93,6 +94,7 @@ def condormatic_instance_create(task) pipe_and_log(pipe, "DeltacloudUsername = #{found.account.credentials_hash['username']}\n") pipe_and_log(pipe, "DeltacloudPasswordFile = #{pwfilename}") pipe_and_log(pipe, "DeltacloudImageId = #{found.provider_image.target_identifier}\n")
- pipe_and_log(pipe, "DeltacloudRetryTimeout = #{@condor_config[:request_timeout]}\n") pipe_and_log(pipe, "DeltacloudHardwareProfile = #{found.hwp.external_key}\n") pipe_and_log(pipe,
Hm, I'm not sure I love these kinds of patches. In particular, I want us to start getting away from using configuration files, because the administrator has no idea on how to set them and they are not dynamic enough. For instance, in this case, what value would you want the administrator to set? In what circumstances would he change it? If you add a new cloud provider type, is the timeout correct?
In this case, I think we can do a lot better by modifying the timeout ourselves. That is, by this point of the code, we know which provider backend we are going to be using. So we should use that information and do a table lookup.
EC2? -> default timeout VMware -> 180 seconds RHEV -> 90 seconds etc.
(note that I just made the above numbers up, they can be whatever we need to be correct)