Mailing-List: contact dev-help@cloudstack.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cloudstack.apache.org
Date: Wed, 21 Dec 2016 12:31:56 +0100 (CET)
From: Wido den Hollander <wido@widodh.nl>
To: Rohit Yadav <rohit.yadav@shapeblue.com>, dev@cloudstack.apache.org
Message-ID: <598487586.5073.1482319916767@ox.pcextreme.nl>
In-Reply-To: <DB5PR07MB168580C9AF74220AD68D6861E9930@DB5PR07MB1685.eurprd07.prod.outlook.com>
References: <DB5PR07MB168580C9AF74220AD68D6861E9930@DB5PR07MB1685.eurprd07.prod.outlook.com>
Subject: Re: [DISCUSS][KVM][BUG] Detaching of volume fails on KVM
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Importance: Medium
archived-at: Wed, 21 Dec 2016 11:32:25 -0000


> Op 21 december 2016 om 12:00 schreef Rohit Yadav <rohit.yadav@shapeblue.com>:
> 
> 
> All,
> 
> 
> Based on results from recent Trillian test runs [1], I've discovered that on KVM (CentOS7) based detaching a volume fails to update the virt/domain xml and fails to remove the xml. So, while the agent and cloudstack-mgmt server succeeds, the entry in the xml is not removed. When the volume is attached again, we can an error like:
> 
> 
> Failed to attach volume xxx to VM VM-yyyy; org.libvirt.LibvirtException: XML error: target 'vdb' duplicated for disk sources '/mnt/8a70be4e-4c3c-38e5-aea2-4b38fef83fd5/af85ff7e-a452-43de-8c6b-948dc44aae21' and '/mnt/8a70be4e-4c3c-38e5-aea2-4b38fef83fd5/af85ff7e-a452-43de-8c6b-948dc44aae21'This is seen in agent logs:
> 
> Dec 21 10:46:35 pr1837-t692-kvm-centos7-kvm2 sh[27400]: DEBUG [kvm.storage.KVMStorageProcessor] (agentRequest-Handler-2:) (logid:0648ae70) Detaching device: <disk  device='disk' type='file'>
> Dec 21 10:46:35 pr1837-t692-kvm-centos7-kvm2 sh[27400]: <driver name='qemu' type='qcow2' cache='none' />
> Dec 21 10:46:35 pr1837-t692-kvm-centos7-kvm2 sh[27400]: <source file='/mnt/8a70be4e-4c3c-38e5-aea2-4b38fef83fd5/af85ff7e-a452-43de-8c6b-948dc44aae21'/>
> Dec 21 10:46:35 pr1837-t692-kvm-centos7-kvm2 sh[27400]: <target dev='vdb' bus='virtio'/>
> Dec 21 10:46:35 pr1837-t692-kvm-centos7-kvm2 sh[27400]: </disk>
> 
> While, after above completes. This is still seen in the VM's dumped xml:
>     <disk type='file' device='disk'>
>       <driver name='qemu' type='qcow2' cache='none'/>
>       <source file='/mnt/8a70be4e-4c3c-38e5-aea2-4b38fef83fd5/af85ff7e-a452-43de-8c6b-948dc44aae21'/>
>       <backingStore/>
>       <target dev='vdb' bus='virtio'/>
>       <serial>af85ff7ea45243de8c6b</serial>
>       <alias name='virtio-disk1'/>
>       <address type='pci' domain='0x0000' bus='0x00' slot='0x05' function='0x0'/>
>     </disk>
> Steps to reproduce:
> 1. Deploy a VM, create a data volume disk and attach to the VM.
> 2. Detach the volume.
> 3. Attach the volume to the same VM again, exception is caught.Thoughts, comments?[1] https://github.com/apache/cloudstack/pull/1837
> Regards.
> 

Isn't this a Qemu bug? That it thinks it detached it, but it doesn't. Or a libvirt thing?

> 
> rohit.yadav@shapeblue.com 
> www.shapeblue.com
> 53 Chandos Place, Covent Garden, London  WC2N 4HSUK
> @shapeblue
>   
>  
>