b18391d5ce
* Register a sqlalchemy event listener on all models that are subclasses of HasId to make sure id is set on object creation so the id can be referenced prior to commit. * Add TODO entry to think about refactoring out sqlalchemy from the plugin
32 lines
1.5 KiB
Plaintext
32 lines
1.5 KiB
Plaintext
Import the tags and meta objects from newtonian
|
|
* Clean up and subset aicq/aiclib for quark
|
|
* Implement All port methods
|
|
Figure out a better way to patch the allowed attributes in the API (bulk create subnets in the network create etc)
|
|
Import and implement zope transaction all the way down
|
|
Implement API extensions for basically everything
|
|
* CRUD routes
|
|
CRUD IP Allocations
|
|
* CRUD MacAddressRanges
|
|
* AIC doesn't support creating multiple switches on the same network. i.e. if we expand past the port count per switch, undefined
|
|
|
|
|
|
Later:
|
|
Maybe implement a straight passthrough driver for testing and debugging purposes
|
|
Separate all SQL alchemy out of the plugin
|
|
|
|
|
|
Allocations Controller Extension:
|
|
|
|
Instead of assigning ports through IP address, we need a separate way of handling that. One
|
|
suggestion was to provide a generic "allocations" controller with a means of defining what
|
|
resources you want to allocate for a particular instance.
|
|
|
|
Examples:
|
|
A traditional instance allocation, requesting a MAC, IP and Port.
|
|
If I were a baremetal server, I may have a MAC but want an IP address and port on a given network for bridging.
|
|
I might simply want to allocate MAC addresses for devices serving a cloud, but don't need any connections on a particular network.
|
|
As a customer, I might want to build an unmanaged isolated network, wherein I assign addresses via a DHCP server I set up myself
|
|
I need an IP address for a device that sits on the same network
|
|
I need to assign additional IPs to a device I already have, on a port I've already attached to
|
|
|