[Users] Adding oVirt Node Fails

Tue May 8 08:11:36 UTC 2012

On 08/05/12 06:06, Nicholas Kesick wrote:
> Results of testing, round 3 (trying to get [Errno 16] Device or resource busy '/etc/pki/vdsm/certs/cacert.pem' error.)
> Variable: no symlink to /etc/system-release
>  
> I started with a reinstall of ovirt-engine and ovirt-node-2.3.0 (formatted drive, reinstalled), enabled network (via logical network breth0 through device eth0), and set the Ovirt Engine Admin Password on the node.
> Created a new host in ovirt-manager, host went to "Pending Approval".
> On the manager, clicked the host, clicked approve, and clicked ok (on power management warning)
> And.... we passed the cert install!  And everything worked up until I hit the cpuFlags issue.
> 
> So... it wasn't the sym-link.  Borked install? I guess the only think I did different these times from before was I used a Fedora 16 disk to format the drive between each attempt, instead of using the ovirt-node disk and performing a reinstall, as well as rebuilding ovirt-manager.
> 
> Not sure where this leaves this topic. Seems I can properly add a node now without a cert or bridge renaming issue.
> - Nick

Nick,
Thanks for taking the time to do this.
Can you just verify you used the same ovirt-node ISO for all your tests?

> ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------!
 ---
> From: cybertimber2000 at hotmail.com
> To: dfediuck at redhat.com; mburns at redhat.com
> Date: Mon, 7 May 2012 18:53:39 -0400
> CC: users at ovirt.org
> Subject: Re: [Users] Adding oVirt Node Fails
> 
> Results of testing, round 2 (trying to get [Errno 16] Device or resource busy '/etc/pki/vdsm/certs/cacert.pem' error.)
>  
> Again, I started with a reinstall of ovirt-engine and ovirt-node-2.3.0 (formatted drive with F16 disk, reinstalled), enabled network (logical = breth0, device = eth0). 
> This time, I set the management server (by IP address and default port), with the [*] Connect to oVirt Engine and Validate Certificate set, and entered the oVirt Engine Admin Password. 
> Ovirt prompts to confirm fingerprint, accepted. (quickly a message about pki and "persisted" appears on screen).
> Created a new host in ovirt-manager, host went to "Pending Approval".
> On the node shell, performed "ln -s /etc/system-release /etc/rhev-hypervisor-release" ((note: not sure this make an effect yet, I think it's the validating certificate part, but did so that only ONE thing changed between first and second test.))
> On the manager, clicked the host, clicked approve, and clicked ok (on power management warning)
> Ended at missing cpuFlags.
>  
> So we are at the same point as before, which means my idea on what caused it (validate certificate) didn't cause it.  Now for testing without the sym-link that Doron suggested.
>  
> - Nick
>  
> ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------!
 ---
> From: cybertimber2000 at hotmail.com
> To: dfediuck at redhat.com; mburns at redhat.com
> Date: Mon, 7 May 2012 18:32:34 -0400
> CC: users at ovirt.org
> Subject: Re: [Users] Adding oVirt Node Fails
> 
> Ok so here is the results of testing:
> I started with a reinstall of ovirt-engine and ovirt-node-2.3.0 (formatted drive, reinstalled), enabled network (via logical network breth0 through device eth0), and set the Ovirt Engine Admin Password on the node. Created a new host in ovirt-manager, host went to "Pending Approval".
> On the node shell, performed "ln -s /etc/system-release /etc/rhev-hypervisor-release"
> On the manager, clicked the host, clicked approve, and clicked ok (on power management warning)
> Ended at missing cpuFlags.
>  
> SO: We are at the same point that I was at with the plain Fedora16 + VDSM install, which is good. No issue with failing to install the cert! 
>  
> But I think I know what has been causing the OSError: [Errno 16] Device or resource busy '/etc/pki/vdsm/certs/cacert.pem' error.
> Working through that now, will post the step by step when I have it.
>  
> - Nick
>  
>> To: mburns at redhat.com; abaron at redhat.com; cybertimber2000 at hotmail.com
>> From: dfediuck at redhat.com
>> CC: users at ovirt.org
>> Date: Mon, 7 May 2012 17:28:20 -0400
>> Subject: Re: [Users] Adding oVirt Node Fails
>>
>> Nicholas,
>> The code I saw is testing for this file to verify it's a nods platform,
>> and if it's missing the behaviour changes. So the code will try to
>> unlink the file, which will give
>> you this result in a mounted file.
>>
>> Sent from my Android phone. Please ignore typos.
>>
>> -----Original Message-----
>> From: Nicholas Kesick [cybertimber2000 at hotmail.com]
>> Received: Tuesday, 08 May 2012, 0:08
>> To: mburns at redhat.com, abaron at redhat.com
>> CC: oVirt Mailing List [users at ovirt.org]
>> Subject: Re: [Users] Adding oVirt Node Fails
>>
>>
>> Ayal, did you see something in the code that made you want me to try the sym-link to /etc/rhev-hypervisor-release?Mike, is there any harm in linking it? - Nick
>> > Subject: Re: [Users] Adding oVirt Node Fails
>> > From: mburns at redhat.com
>> > To: abaron at redhat.com
>> > CC: cybertimber2000 at hotmail.com; users at ovirt.org
>> > Date: Sun, 6 May 2012 21:26:08 -0400
>> >
>> > On Sat, 2012-05-05 at 16:39 -0400, Ayal Baron wrote:
>> > >
>> > > ----- Original Message -----
>> > > >
>> > > >
>> > > > > Date: Sat, 5 May 2012 12:29:20 +0300
>> > > > > From: dfediuck at redhat.com
>> > > > > To: cybertimber2000 at hotmail.com
>> > > > > CC: mburns at redhat.com; users at ovirt.org
>> > > > > Subject: Re: [Users] Adding oVirt Node Fails
>> > > > >
>> > > > > On 04/05/12 02:49, Nicholas Kesick wrote:
>> > > > > > Fri, 27 Apr 2012 05:22:24 DEBUG instCert: try to delete old
>> > > > > > certificates
>> > > > > > Fri, 27 Apr 2012 05:22:24 DEBUG instCert: failed.
>> > > > > > Traceback (most recent call last):
>> > > > > > File "/usr/share/vdsm-reg/deployUtil.py", line 1133, in instCert
>> > > > > > File "/usr/share/vdsm-reg/deployUtil.py", line 1098, in
>> > > > > > pkiCleanup
>> > > > > > OSError: [Errno 16] Device or resource busy:
>> > > > > > '/etc/pki/vdsm/certs/cacert.pem'
>> > > > >
>> > > > >
>> > > > > Nicholas,
>> > > > > I think I got it;
>> > > > > Please check if this file exists: '/etc/rhev-hypervisor-release'.
>> > > > > If not, try to create a sym-link in this name to
>> > > > > /etc/system-release,
>> > > > > and restart vdsm-reg service.
>> >
>> > No, /etc/rhev-hypervisor-release should not exist. It doesn't have to.
>> > We went through significant effort before the 3.0 release making sure
>> > that vdsm used /etc/ovirt-node-image-release for determining if the
>> > machine is ovirt-node based or not.
>> >
>> > There should not be any need for anything that says "rhev" in
>> > ovirt-node.
>> >
>> > Mike
>> >
>> > > > >
>> > > > > Also please provide the rpm information for vdsm-reg and vdsm as
>> > > > > well.
>> > > > > --
>> > > > >
>> > > > > /d
>> > > > >
>> > > > > "Common sense is not so common." --Voltaire, Dictionnaire
>> > > > > Philosophique (1764)
>> > > >
>> > > > Just to make sure I do it right, can you specify which you wanted
>> > > > done on the node, and which you wanted on the manager? I have the
>> > > > feeling you wanted it all done on the node.
>> > >
>> > > you are correct, all on node.
>> > >

-- 

/d

"Email returned to sender -- insufficient voltage."