[
https://issues.apache.org/jira/browse/JCLOUDS-393?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13842318#comment-13842318
]
Andrew Bayer commented on JCLOUDS-393:
--------------------------------------
Ok, woo. I seem to have actually finished it. Maybe. Pending much review. All
live tests appear to be working now, I rigged up new default behavior to delete
boot disks on jclouds-created instances unless specified otherwise (note that
this'll just be ignored completely with non-jclouds-created instances, which
felt the safest to me), and the default image for tests is Debian 7, not CentOS
6, due to some weird auth issues I'd like to poke at more. I think osFamiliy
detection may be a bit borked right now, but I'm not 100% sure so I'll let it
sit for the moment.
> Support GCE API v1
> ------------------
>
> Key: JCLOUDS-393
> URL: https://issues.apache.org/jira/browse/JCLOUDS-393
> Project: jclouds
> Issue Type: Improvement
> Components: jclouds-labs-google
> Reporter: Andrew Bayer
> Assignee: Andrew Bayer
> Fix For: 1.7.0
>
>
> So this is a bit of a stretch goal for 1.7.0, but GCE has gone GA and moved
> to API v1, which...well,
> https://developers.google.com/compute/docs/transition-v1 - it's got a number
> of breaking changes from v1beta16 and earlier. Most notably, instance
> creation and images are now more than a little different. The images from
> v1beta16 and earlier with Google-provided kernels don't work with v1, the
> kernel resource is just gone completely, and most annoyingly, you no longer
> can create instances from images directly. Instead, you have to first create
> a root persistent disk from an image, and then specify that disk when
> creating the instance. When you get an instance's information, it no longer
> has anything about the image it was created from - instead you've got to go
> through the disk to see what its source image is, but even that's a bit odd,
> since the current API reference doesn't actually include the sourceImage in
> the list of fields for disk resources, though it's pretty strongly implied
> that it should be there...bah.
> Anyway, I'm going to try to get this done for 1.7.0, but given the timeframe,
> it may not be completed in time. v1beta16 is supposed to stick around until
> March, so we've got some cushion to get out 1.7.1 with v1 support if needed.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
