Ask Your Question

rmartin's profile - activity

2015-07-24 11:30:00 -0500 received badge  Famous Question (source)
2015-03-05 17:37:00 -0500 answered a question nova volume-detach fails silently, compute log shows libvirtError: invalid argument: no target device vdb

I was able to get the libvirt configuration resynchronized by copying the <disk> XML from the xml dump into its own file, and then updated it back into the current configuration via

virsh attach-device 4 filestore.xml --config

After that, I reran nova volume-detach. It succeeded in cleaning up the compute node, the iscsi session is now idle. However, now the volume was stuck in "detaching" status. No errors anywhere, no idea why it didn't go back to "available." tgt-admin allowed me to take it offline and back to ready, then I got it back to available status with cinder reset-state.

2015-03-05 11:04:19 -0500 received badge  Notable Question (source)
2015-03-05 11:04:19 -0500 received badge  Popular Question (source)
2015-03-04 11:14:25 -0500 commented question nova volume-detach fails silently, compute log shows libvirtError: invalid argument: no target device vdb

Here's the error from the compute node's compute.log:

2015-03-03 19:19:33.125 12332 AUDIT nova.compute.manager [req-c9695029-1529-49d4-be65-ec2a8fe0db49 918670168a67461bb0bc5fbccba26fd5 012c82ce05cd4739994cf7e103300c2f] [instance: 4d05ffe4-d30a-4c93-b710-c9ec80dad1c2] Detach volume 34b0ea26 -f85c-4

2015-03-04 11:14:24 -0500 asked a question nova volume-detach fails silently, compute log shows libvirtError: invalid argument: no target device vdb

I'm trying to detach a cinder volume from a nova instance.

Nova volume-list shows it as in-use:

| 34b0ea26-f85c-4b62-8ebd-884b0e63e2d5 | in-use | filestore | 256  | None| 4d05ffe4-d30a-4c93-b710-c9ec80dad1c2 |

The volume is connected via iscsi:

# iscsiadm -m session
tcp: [5] 10.3.40.10:3260,1 iqn.2010-10.org.openstack:volume-34b0ea26-f85c-4b62-8ebd-884b0e63e2d5 (non-flash)

It's visible as /dev/vdb within the instance, and is mountable and read/writable.

However, after unmounting it in the instance and issuing

# nova volume-detach 4d05ffe4-d30a-4c93-b710-c9ec80dad1c2 34b0ea26-f85c-4b62-8ebd-884b0e63e2d5

it does not detach. The volume stays labeled "in-use". I get this error in the compute.log:

libvirtError: invalid argument: no target device vdb

When I run "virsh edit 4" on the compute node, the disk device for vdb is in fact missing. However, when I run "virsh dumpxml 4" it is still there!

How can I untangle this?

Virsh dumpxml vs. edit diff: http://pastebin.com/3CrGbcwu

Compute.log of the error: http://pastebin.com/1UCahB3d

2014-08-25 20:25:07 -0500 received badge  Famous Question (source)
2014-07-23 08:01:17 -0500 received badge  Enthusiast
2014-04-09 14:11:49 -0500 received badge  Notable Question (source)
2014-04-03 07:20:18 -0500 received badge  Popular Question (source)
2014-03-24 14:19:35 -0500 answered a question How does the nova_api service determine the suffix to add to each instance name to generate a hostname?

This is drawn from the dhcp_domain setting in nova.conf. After changing the setting, one must: 1) killall dnsmasq 2) Restart nova-network 3) Restart nova-api.

I had not yet done the third step.

2014-03-24 14:17:55 -0500 commented answer How to set domain name in quantum

Never mind. dhcp_domain worked once I restarted nova-api. Thanks!

2014-03-24 14:03:59 -0500 commented answer How to set domain name in quantum

Changing dhcp_domain didn't fix it for me.

https://ask.openstack.org/en/question/25778/how-does-the-nova_api-service-determine-the-suffix-to-add-to-each-instance-name-to-generate-a-hostname/ (https://ask.openstack.org/en/question...)

2014-03-24 11:36:21 -0500 commented question How does the nova_api service determine the suffix to add to each instance name to generate a hostname?

BTW, in case it matters, we're running openstack-grizzly on RHEL 6.4.

2014-03-24 11:35:46 -0500 asked a question How does the nova_api service determine the suffix to add to each instance name to generate a hostname?

By default, the nova-api service generates a key in meta_data.json "hostname" which is set to "<instance_name>.novalocal". I need to change novalocal to something else, but I cannot find where this setting is stored. I set dhcp_domain in nova.conf, and this added my suffix to the dns search list sent to instances by dnsmasq, but it didn't affect the hostnames. Any idea where/how this can be set?

2013-06-20 08:26:09 -0500 received badge  Supporter (source)