[
https://issues.apache.org/jira/browse/CLOUDSTACK-9877?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15969108#comment-15969108
]
ASF GitHub Bot commented on CLOUDSTACK-9877:
--------------------------------------------
Github user DaanHoogland commented on the issue:
https://github.com/apache/cloudstack/pull/2044
@serg38
ad 1. yes they can, if you look in the ticket there is a part of
description that describes this possibility. But that said I don't see how
restore is going to be hindered if the vm was created using a full clone. The
image is not deleted.
ad 2. No it is only marking templates for gc and not touching any cleanup
process, it relies on the regular process to do the actual cleanup. The benefit
is that short lived templates that will be replaced by cron jobs for instance
will be cleaned without having to wait until the VMs cloned from it are all
gone (on a per primary store basis btw)
ad 3. Yes so far it is I have not studied other hypervisors on this area.
The code is now in the vmware plugin but not tightly coupled to the vmware API
so generalizing it shoud be rather trivial.
ad 4. the host scan task was an empty placeholder with only todo. The
vcente clusters are not scanned for new hosts atm. If you look at the old code
you can find that the task consists of only a comment:
```
private Runnable getHostScanTask() {
return new Runnable() {
@Override
public void run() {
// TODO scan vSphere for newly added hosts.
// we are going to both support adding host from CloudStack
UI and
// adding host via vSphere server
//
// will implement host scanning later
}
};
}
```
> remove fully cloned deleted templates from primary storage
> ----------------------------------------------------------
>
> Key: CLOUDSTACK-9877
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-9877
> Project: CloudStack
> Issue Type: Improvement
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Components: VMware
> Environment: cloudstack with vcentre and the use of "full clones"
> Reporter: Daan Hoogland
> Assignee: Daan Hoogland
>
> After a template has been deleted from cloudstack and it has only been used
> with the 'vmware.create.full.clone' setting, it can be cleared from primary
> storage.
> test notes:
> related settings:
> vmware.full.clone.template.cleanup.period = '1' (minutes, '86400')
> vmware.create.full.clone = 'true' (boolean, 'false')
> storage.template.cleanup.enabled = 'true' (boolean, 'false')
> storage.cleanup.interval = '150' (seconds, default)
> storage.cleanup.enabled = 'true' (boolean, default)
> to test:
> a vm can be created using the default template or any template installed
> for this purpose.
> I'll call it the install-template
> The vm should be created with a full clone of the template.
> This can be checked in the table cloud.user_vm_clone_setting
> The install-template should be copied to the primary store of the host it
> is on
> This can be checked in the table cloud.template_spool_ref
> delete the install-template from the system
> after at most a minute the entries in cloud.template_spool_ref for the
> install-template should have the field marked_for_gc set to '1'
> At most a minute and a half after that the field state in the same record
> should be set to 'Destroyed'
> to double check one might go to the mount for this primary storage and
> seek the installpath to make sure the disk space is actually freed.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
