New Question
0

Storage SMB driver not working.

asked 2016-11-07 11:39:24 +0300

Niclas gravatar image

Hi everyone,

I have a controller based on Ubuntu 14.04 and one windows 2016 with hyper-v which is compute and storage. I am trying to get the SMB driver to work but I get to a point where there are no errors but still do not work. I have installed the driver to my hyper-v node and created a share. I have verified that the share is accessible from my controller node by mounting the share on the controller. if I do a cinder service-list on my controller I can see that I have a cinder volume on my hyper-v and it is up and running together with a cinder-scheduler on the controller. Now to my errors. If I type “cinder create 1” on the controller or try to create a volume from horizon the volume gets stuck at creating. There is not log at all on the controllers cinder-scheduler.log nor on the hyper-v cinder log. Cinder-api.log on controller has some logs that all is successful (no errors or warnings). And there is a new entry in the cinder mysql database. I have tried to restart the services many times. If I restart the smb driver it notices that there is a volume stuck at creating so it sets it to error. If I change that do deleting smb driver will delete it at next restart. If I in horizon create a new Volume type with correct volume backend name it behaves the exact same way. However, if I create a false one and creates a volume to it i get errors directly and I get logs in the cinder-scheduler.log on the controller.

What am I missing to get this working?

edit retag flag offensive close merge delete

Comments

Forgot to mention that i use Mitaka release of controller and storage driver.

Niclas gravatar imageNiclas ( 2016-11-08 11:57:57 +0300 )edit

2 answers

Sort by » oldest newest most voted
0

answered 2016-11-08 16:31:01 +0300

lpetrut gravatar image

Hi Niclas,

So, it seems that the volume is properly scheduled to that host, it's just that the request is hanging. In most of the cases, this is caused by time synchronization issues among hosts.

I'd suggest ensuring that your nodes are time synced and that the AMQP options are properly set.

edit flag offensive delete link more
0

answered 2016-11-09 10:26:10 +0300

Niclas gravatar image

Hi, Thank you for your answer.

Both controller and hyper-v node is synced to same ntp time source and they have the same time. I suppose both neutron and nova would stop aswell if the time were not synced? neutron and nova works fine. If i wireshark on the rabbitMQ traffic on hyperv i can see there is connection but when i create a volume there is no indication that a new volume should be created. but if i tcpdump on the controller i see alot of thing going on to the rabbitMQ.

edit flag offensive delete link more

Comments

Have you installed Cinder using our MSI installer? Just to make sure, have you used the Mitaka MSI?

lpetrut gravatar imagelpetrut ( 2016-11-09 19:13:33 +0300 )edit

Yes, I used the mitaka installer found here https://cloudbase.it/openstack-windows-storage/

Niclas gravatar imageNiclas ( 2016-11-09 19:29:42 +0300 )edit

Could you please set the cinder-volume service to run in debug mode on the Windows node? I'm wondering if it gets the volume creation request. How is RabbitMQ configured? Are you using vhosts, ha queues or anything like that? Could you please paste your Cinder config files from both nodes?

lpetrut gravatar imagelpetrut ( 2016-11-10 14:19:54 +0300 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower

Stats

Asked: 2016-11-07 11:39:24 +0300

Seen: 396 times

Last updated: Nov 09 '16