search

ArchipelProject / Archipel

XMPP Based Orchestrator
http://archipelproject.org
GNU Affero General Public License v3.0
803 stars 127 forks source link

Find a way to coexist whith HA #605

Open franx opened 11 years ago

franx commented 11 years ago

If you configure a VM to be managed by corosync it's state will be shown uncorrect under archipel.

Step to reproduce:

Create a VM in archipel. (start , stop reconfigure, everything work fine)

(stop the VM in archipel)

Join the vm to the cluster:

crm configure primitive My_HA_WM ocf:heartbeat:VirtualDomain \ params \ config="/etc/libvirt/qemu/my_archipel_creted_one.xml" \ hypervisor="qemu:///system" \ meta allow-migrate="true" \ op start timeout="120s" \ op stop timeout="120s" \ op monitor depth="0" timeout="30" interval="10"

execute: crm status

Stack: openais Current DC: FS - partition with quorum Version: 1.1.6-3.el6-a02c0f19a00c1eb2527ad38f146ebc0834814558 2 Nodes configured, 2 expected votes

9 Resources configured.

Online: .......

My_HA_WM (ocf::heartbeat:VirtualDomain): Started

Archipel Will show the VM as off , red ballon , under the Controls tab of the VM will be shown red ballon and running.

Play Button will not be show as accessibile, and the configuration should be changed at any time.

Schermata 2012-12-13 alle 00 23 26

CyrilPeponnet commented 11 years ago

Need logs from agent.

bio-erasmus commented 10 years ago

Agent hardly poduces logs for this problem, to me it looks like the archipel agent loses track of the virual machine when its migrated via the cluster managment software until the agent is restarted. in my case 2 node RHEL6 ha cluster using clusvcadm to manage services defined in cluster config.

CyrilPeponnet commented 10 years ago

Well, I'm pretty sure it's related to #1020. When migrating the VM through libvirt I'm not sure the agent is aware of the migration. So libvirt event loop are going wrong.