John Burwell created CLOUDSTACK-8565:
----------------------------------------
Summary: KVM agent fails to rejoin when agent is restarted after
custom offering vm is stopped
Key: CLOUDSTACK-8565
URL: https://issues.apache.org/jira/browse/CLOUDSTACK-8565
Project: CloudStack
Issue Type: Bug
Security Level: Public (Anyone can view this level - this is the default.)
Components: KVM
Affects Versions: 4.5.1
Reporter: John Burwell
Priority: Critical
We've found a case where we have a local storage KVM cluster, running 4.5.1,
and it seems there's a reproducible bug where the management server refuses to
allow agent connection if a VM with dynamic offering was deployed to it:
1. Deploy/start local storage VMs that have a dynamic service offering.
2. Get a VM into a state where it is stopped, but has a last_host_id (stop a vm
that you just created).
3. Restart the agent.
The management server fails to allow agent connect, until the last_host_id
field is cleared in the db for any stopped vm that ran on this host. This may
actually not be specific to local storage VMs, but we think it might, because
the vms may start elsewhere and auto-reset this field without local storage. It
seems to be related to the vm sync process that occurs during agent join.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
