Re: No such file or directory: '/home/mturk/core.pub'

[Mike Freemon <mfreemon@xxxxxxxxxxxx>]

Hi Zeo,

I have it working with "from novaclient.v1_1 import client" in it, so 
I'm not sure about this:

> I add this line to startup_ndslabs.py:
> 'import novaclient'
> and comment out
> #from novaclient.v1_1 import client

Yes, you need environment variables set for OS_AUTH_URL, OS_USERNAME, 
etc., which would normally be set by sourcing NDS-openrc.sh.  However, 
you could instead specify them as command line arguments if you want. 
They do not come from production.env.

As Kacper mentioned, you do need network connectivity to the OpenStack 
management network, which is on a private subnet on the vlad-mgmt node. 
 All of your network timeout type errors are probably related to this 
if you're not running the script on vlad-mgmt.

Ticket NDS-20 is all about improving the situation with these 
networking-related hassles on the current VLAD cluster.
https://jira.ncsa.illinois.edu/browse/NDS-20

If you want, update to my branch ("mfreemon-dev") of nds-labs from 
bitbucket and try that.  I put an explicit authentication call to 
OpenStack to help isolate network and authentication related errors from 
other types of errors.  If startup_ndslabs.py fails on the 
authenticate() call, the problem is probably related to networking or 
user authentication.  If the scripts passes that point, and fails on the 
nt.servers.create() call, then the problem is something else.

- Mike


On 6/21/2015 7:09 PM, Kacper Kowalik wrote:
> On 06/19/2015 02:58 PM, Lengyue Chen wrote:
> > Or would it possible wrong with the net-id? Maybe the network is not
> > working now?
> >
> > On Fri, Jun 19, 2015 at 2:35 PM, Lengyue Chen <lchen95@xxxxxxxxxxxx>
> > wrote:
> >
> > > Hi all
> > >
> > > I have solved the last issue with client object. And here is the
> > > new one again... It timed out in this function call:
> > >
> > > instance = nt.servers.create( "coreos_%s" % args.cluster_name,
> > > args.image_id, args.flavor_id, min_count=n, max_count=n,
> > > security_groups=["default", "coreos"],
> > > userdata=open('cloud-config_%s.yaml' % public, 'r'),
> > > key_name=args.ssh_key_name, nics=[{"net-id": args.net_id}] )
> > >
> > > What I can think of is the key does not match but I guess there
> > > should be a corresponding error message indicating the wrong key. I
> > > tried to extract the ndsisl.pub from both my own
> > > service_test_key_zeo.pem and nds.pem which Kacper gave me but they
> > > did not work.
> > >
> > >
> > > Here are the full error message and all my env variables. Thanks.
> > >
> > >
> > > Lengyues-MacBook-Air:~ LY$ python startup_ndslabs.py ...
>
> Hi Lengyue,
> I apologize if I wasn't sufficiently clear before, but you *won't* be
> able to run this script anywhere else besides vlad-mngt machine.
> Openstack that's installed there exposes REST API (which novaclient
> internally uses) through a set of local ip addresses.
> Cheers,
> Kacper