Likitha Shetty created CLOUDSTACK-5096:
------------------------------------------
Summary: [VMware] Corrupt template is left behind after the copy
of a template from secondary to primary fails.
Key: CLOUDSTACK-5096
URL: https://issues.apache.org/jira/browse/CLOUDSTACK-5096
Project: CloudStack
Issue Type: Bug
Security Level: Public (Anyone can view this level - this is the default.)
Components: VMware
Affects Versions: 4.2.0
Reporter: Likitha Shetty
Assignee: Likitha Shetty
Fix For: 4.2.1
When template copy from secondary to primary doesn't go through, VM does not
bootup. Also, the subsequent VM creations from the same template fails because
it continues to use the corrupt template that was left behind.
1. Monitor the state of HttpNfcLease (object returned by the import task) and
check if it goes into ERROR state. This will handle vCenter operation timeout
issue
2. Monitor the progress of VMDKs upload and catch exceptions if any (Already
handled)
In case of failure perform cleanup by,
- Throw an error with the message containing the task fault
- Abort the HttpNfcLease which will in turn handle the removal of all objects
including VMDKs that were created during import
- Retry in case of upload failure (Already handled)
- Remove the CloudStack DB entry for the template in the primary store. Since
the entry is removed during the next VM launch from the same template fresh
COPY command will be issued (Already handled)
- Today, if a corrupt template is left behind then during the next VM
deployment even though the COPY command has been issued, vCenter will reuse the
template that has been left behind.
Hence, if management server request for a copy based on no DB entry for the
template in that store then delete the old template if present (vCenter
identifies the template present in the store based on the unique uuid name) and
import the template from secondary to primary.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
