| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
cloudmasterd
Changes include:
* everest-bootstrap's default mode is now "cloud".
- It is assumed that all systems provisioned by everest-bootstrap will be sent to a cloudmaster
for provisioning. However, a cloudmaster does not necessarily have to control several minions
to qualify. A cloudmaster can be configured to simply control its own disk space.
- The "advanced" mode also expects that systems are getting provisioned in a cloud, and requires
a "cloudmaster" parameter.
- The previous "--config-only" command line parameter has been changed to a mode. Now a user can
select between "standard", "advanced", and "config-only" modes.
- Hooks for cobbler dns have been added. A great deal more work needs to be done here before it
is ready, though.
- everest-bootstrap can now ask everestd if it knows the IP address and hostname of newly provisioned
machines
* everestd can now examine the /var/log/cobbler/install.log file to determine the IP address and hostname
of new systems.
- I suspect that this functionality will get moved from everestd to the cloudmaster. Users of
everest-bootstrap will be directed to the cloud's status page to access the systems' IP addy and
hostnames.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
I don't know why this was changed.
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
* everest-dsl
* everest-bootstrap
Remaining problems:
* everest-sync still missing a spec file
* everest-repo cannog build RPM without a local copy of "docs" git repo
|
| | |
|
| |
|
|
| |
Spec file / version cleanup
|
| |
|
|
|
| |
Updated everestd Rakefile to clean the gem package data before RPM build
Updated cloudmasterd Rakefile to correctly clean the gem package data before RPM build
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
virtual machines within the cloud.
Updated everest-bootstrap to call the cloud controller when provisioning machines
Removed "koan" operations from everestd. Now everestd only knows how to register machines with everest and cobbler. Basically a rollback to its previous version.
|
| | |
|
| |
|
|
| |
Ignoring built files
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
------------------------------------------------------
The "cloud" mode acts like the "wizard" in the standard mode of everest-bootstrap in that it asks relevant questions for provisioning a new machine.
After collecting the standard information, it registers the new system with cobbler and with everestd. It then asks everestd to koan the new machine.
Adding support to everestd to call func to koan a requested machine on the most suitable everest host
-----------------------------------------------------------------------------------------------------
Now everestd asks "cloud-inventory" which host/vol_group combination is the best fit for a particular machine. Then, everestd will delegate to
func to koan the new machine on the host/vol_group combination.
|
|
|
everestd is a simple service for handling our everest machine types. Currently
is serves the following purposes:
1) Backend service for everest-bootstrap
2) Creates the yaml files used for Puppet's external nodes.
Usage:
Once installed simply browse to http://[hostname]:8601/nodes.html. From there
it should be easy to navigate to the rest of the functionality. Most resources
have several representations. Try changing 'html' to 'xml' or 'yaml' in the
urls.
Configuration:
The config file gets laid down in /etc/everestd/
In the future it could be switched to use a db for storing the external node
data. This would allow us to do some interesting things but it would make
bootstrapping a repo more complicated. Currently Puppet installs everestd once
it has alread been bootstrapped. Since our external node script simply looks
to the yaml files on disk Puppet doesn't even have to know everestd exists.
This is also closer to how Prod Ops manages there external nodes.
Implementation details:
everestd uses the everest gem to get at the machine type information for a
particular repo. This is similar to how our old external nodes script worked.
For the web portion everestd makes use of the camping microframework. This
includes the reststop library for restful controllers and views as well as the
the picnic gem for the service script. Yay for stupid names.
The gem was created with newgem and the rpm spec file was generated with
gem2rpm.
TODO:
* make the yaml node dir configurable.
* consider allowing machines to be configured/koan from the web ui (as a
possible replacement for everest-bootstrap)
|
