[ovirt-users] Re: Upgrade from ovirt-node-ng 4.4.1 to 4.4.2 fails with "Local storage domains were found on the same filesystem as / ! Please migrate the data to a new LV before upgrading, or you will

[Nir Levy]

On Tue, Oct 13, 2020 at 5:51 PM Nir Soffer  wrote:

> On Tue, Oct 13, 2020 at 4:20 PM Yedidyah Bar David 
> wrote:
> >
> > On Tue, Oct 13, 2020 at 1:11 PM Yedidyah Bar David 
> wrote:
> > >
> > > On Thu, Oct 8, 2020 at 1:01 PM gantonjo-ovirt--- via Users
> > >  wrote:
> > > >
> > > > So, we have a cluster of 3 servers running oVirt Node 4.4.1. Now we
> are attempting to upgrade it to latest version, 4.4.2, but it fails as
> shown below. Problem is that the Storage domains listed are all located on
> an external iSCSI SAN. The Storage Domains were created in another cluster
> we had (oVirt Node 4.3 based) and detached from the old cluster and
> imported successfully into the new cluster through the oVirt Management
> interface. As I understand, oVirt itself has created the mount points under
> /rhev/data-center/mnt/blockSD/ for each of the iSCSI domains, and as such
> they are not really storaged domains on the / filesystem.
> > > >
> > > > I do believe the solution to the mentioned BugZilla bug has caused a
> new bug, but I may be wrong. I cannot see what we have done wrong when
> importing these storage domains to the cluster (well, actually, some were
> freshly created in this cluster, thus fully managed by oVirt 4.4 manager
> interface).
> > >
> > > This is likely caused by the fix for:
> > > https://bugzilla.redhat.com/show_bug.cgi?id=1850378 .
> > >
> > > Adding Nir.
> > >
> > > >
> > > > What can we do to proceed in upgrading the hosts to latest oVirt
> Node?
> > >
> > > Right now, without another fix? Make sure that the following command:
> > >
> > > find / -xdev -path "*/dom_md/metadata" -not -empty
> > >
> > > Returns an empty output.
> > >
> > > You might need to move the host to maintenance and then manually
> > > umount your SDs, or something like that.
> > >
> > > Please open a bug so that we can refine this command further.
> >
> > Nir (Levy) - perhaps we should change this command to something like:
> >
> > find / -xdev -path "*/dom_md/metadata" -not -empty -not -type l
> >
> > >
> > > Thanks and best regards,
> > >
> > > >
> > > > Dependencies resolved.
> > > >
> =
> > > >  Package
>Architecture
>   Version
> Repository
>   Size
> > > >
> =
> > > > Upgrading:
> > > >  ovirt-node-ng-image-update
>   noarch
> 4.4.2-1.el8
>   ovirt-4.4
> 782 M
> > > >  replacing  ovirt-node-ng-image-update-placeholder.noarch
> 4.4.1.5-1.el8
> > > >
> > > > Transaction Summary
> > > >
> =
> > > > Upgrade  1 Package
> > > >
> > > > Total download size: 782 M
> > > > Is this ok [y/N]: y
> > > > Downloading Packages:
> > > > ovirt-node-ng-image-update-4.4.2-1.el8.noarch.rpm
>
>
> 8.6 MB/s |
> 782 MB 01:31
> > > >
> -
> > > > Total
>
>
> 8.6 MB/s |
> 782 MB 01:31
> > > > Running transaction check
> > >

[ovirt-users] Re: Can't add freshly installed node.. host has no default route

[Nir Levy]

Hi Giorgio,

Ovirt-node is based on being a closed system with certain predefined
packages,
so the system updates itself to a newer version with an updated bundle of
packages.

additional packages can be installed if you enable the repositories
residing at:
/etc/yum.repos.d in this case /etc/yum.repos.d/ovirt-4.3.repo
this should resolve what you are encountering.





On Wed, May 13, 2020 at 2:18 PM Giorgio Biacchi  wrote:

> Hi Lev,
> I just used the iso you provided to reinstall the same host and now I
> see vdsm-hook-nestedvt is pre installed, but this is only a workaround.
>
> The hook is always present, no matter what I put in
> /usr/share/ovirt-host-deploy/plugins/ovirt-host-deploy/vdsmhooks/packages.d/
>
> on the engine host.
>
> If I add, for example, vdsm-hook-macspoof in the same directory on the
> engine host the installation fails again:
>
> 2020-05-13 10:39:32,590+ ERROR
> otopi.plugins.otopi.packagers.yumpackager yumpackager.error:85 Yum
> Cannot queue package vdsm-hook-macspoof: Package vdsm-hook-macspoof
> cannot be found
> 2020-05-13 10:39:32,590+ DEBUG otopi.context
> context._executeMethod:145 method exception
> Traceback (most recent call last):
>File "/tmp/ovirt-CQNPURostK/pythonlib/otopi/context.py", line 132, in
> _executeMethod
>  method['method']()
>File
> "/tmp/ovirt-CQNPURostK/otopi-plugins/ovirt-host-deploy/vdsmhooks/hooks.py",
>
> line 109, in _packages
>  self.packager.installUpdate(f.read().splitlines())
>File
> "/tmp/ovirt-CQNPURostK/otopi-plugins/otopi/packagers/yumpackager.py",
> line 305, in installUpdate
>  ignoreErrors=ignoreErrors
>File "/tmp/ovirt-CQNPURostK/pythonlib/otopi/miniyum.py", line 884, in
> installUpdate
>  **kwargs
>File "/tmp/ovirt-CQNPURostK/pythonlib/otopi/miniyum.py", line 500, in
> _queue
>  package=package,
> RuntimeError: Package vdsm-hook-macspoof cannot be found
>
> On https://resources.ovirt.org/pub/ovirt-4.3/rpm/el7/noarch/ I see many
> packetized hooks and I thought that adding what I need in
> /usr/share/ovirt-host-deploy/plugins/ovirt-host-deploy/vdsmhooks/packages.d/
>
> was the correct way to install them. Am I wrong??
>
> Regards
>
> Il 12/05/2020 19:30, Lev Veyde ha scritto:
> > Hi Giorgio,
> >
> > Do you have a staging test (non production) environment?
> > I built a test ovirt-node-ng image that includes this package, and if
> > you want you can download it from here:
> >
> https://jenkins.ovirt.org/job/ovirt-node-ng-image_standard-check-patch/176/artifact/check-patch.el7.x86_64/
> >
> > If you do, please let us know if it resolved the issue for you,
> >
> > Thanks in advance,
> >
> > On Tue, May 12, 2020 at 6:57 PM Giorgio Biacchi  > > wrote:
> >
> > Il 12/05/2020 17:07, Dominik Holler ha scritto:
> >  >
> >  >
> >  > On Tue, May 12, 2020 at 4:25 PM Giorgio Biacchi
> > mailto:gior...@di.unimi.it>
> >  > >> wrote:
> >  >
> >  > On 5/12/20 12:28 PM, Dominik Holler wrote:
> >  >  >
> >  >  >
> >  >  > On Tue, May 12, 2020 at 8:49 AM Giorgio Biacchi
> >  > mailto:gior...@di.unimi.it>
> > >
> >  >  > 
> >  >  >  >
> >  >  > On 5/11/20 5:53 PM, Dominik Holler wrote:
> >  >  > >
> >  >  > >
> >  >  > > On Mon, May 11, 2020 at 12:31 PM Giorgio Biacchi
> >  >  > mailto:gior...@di.unimi.it>
> > >
> >  > 
> > >>
> >  >  > >  >   > >
> >  > 
> >  wrote:
> >  >  > >
> >  >  > > Hi list,
> >  >  > > I've spent a couple of days trying to understand
> why
> >  > this was
> >  >  > > happening...
> >  >  > >
> >  >  > > For the installation I have a well tested
> > installation
> >  > server
> >  >  > with a
> >  >  > > custom kickstart file to setup ssh keys and
> custom
> >  > hooks for
> >  >  > infiniband
> >  >  > > and I'm installing Ovirt Node 4.3.9 via pxe,
> this is
> >  > particularly
> >  >  > > useful
> >  >  > > when I have to install a bunch of blades at
> > once.. In
> >  > the past
> >  >  > I had no
> >  > 

[ovirt-users] Re: Can't add freshly installed node.. host has no default route

[Nir Levy]

Hi strahil,

The current ovirt node image built by Lev includes
vdsm-hook-nestedvt-4.30.46-1.el7.noarch
so yes it should be available when adding the node.


On Wed, May 13, 2020 at 12:49 AM Strahil Nikolov via Users 
wrote:

> On May 12, 2020 6:56:45 PM GMT+03:00, Giorgio Biacchi 
> wrote:
> >Il 12/05/2020 17:07, Dominik Holler ha scritto:
> >>
> >>
> >> On Tue, May 12, 2020 at 4:25 PM Giorgio Biacchi  >> > wrote:
> >>
> >> On 5/12/20 12:28 PM, Dominik Holler wrote:
> >>  >
> >>  >
> >>  > On Tue, May 12, 2020 at 8:49 AM Giorgio Biacchi
> >> mailto:gior...@di.unimi.it>
> >>  > >>
> >wrote:
> >>  >
> >>  > On 5/11/20 5:53 PM, Dominik Holler wrote:
> >>  > >
> >>  > >
> >>  > > On Mon, May 11, 2020 at 12:31 PM Giorgio Biacchi
> >>  > mailto:gior...@di.unimi.it>
> >> >
> >>  > > 
> >>  >wrote:
> >>  > >
> >>  > > Hi list,
> >>  > > I've spent a couple of days trying to understand why
> >> this was
> >>  > > happening...
> >>  > >
> >>  > > For the installation I have a well tested
> >installation
> >> server
> >>  > with a
> >>  > > custom kickstart file to setup ssh keys and custom
> >> hooks for
> >>  > infiniband
> >>  > > and I'm installing Ovirt Node 4.3.9 via pxe, this is
> >> particularly
> >>  > > useful
> >>  > > when I have to install a bunch of blades at once..
> >In
> >> the past
> >>  > I had no
> >>  > > issues and all was working like a charm until now
> >when some
> >>  > hardware
> >>  > > failed and I had to replace it.
> >>  > >
> >>  > > As expected I have no issues in the node
> >installation
> >>  > process.. the
> >>  > > troubles begins when I try to add the node,
> >> installation fails
> >>  > and in
> >>  > > the UI I have an exclamation mark with the message
> >> "Host has
> >>  > no default
> >>  > > route." but I can ping and do ssh to the host from
> >the
> >>  > manager.. the
> >>  > > problem is somewhere else in the communication
> >between the
> >>  > engine and
> >>  > > vdsmd preventing the engine to refresh the host
> >> capabilities.
> >>  > >
> >>  > > So from the engine I tried:
> >>  > >
> >>  > > [root@manager ~]# openssl s_client -connect
> >> 172.20.22.78:54321 
> >>  > 
> >>  > > 
> >>  > > CONNECTED(0003)
> >>  > > ---
> >>  > > Certificate chain
> >>  > >   0 s:/CN=cn128.lagrange.di.unimi.it/O=VDSM
> >> 
> >>  > 
> >>  > > 
> >Certificate
> >>  > > i:/CN=VDSM Certificate Authority
> >>  > >   1 s:/CN=VDSM Certificate Authority
> >>  > > i:/CN=VDSM Certificate Authority
> >>  > > ---
> >>  > >
> >>  > > The host has still the self signed vdsm
> >certificate..
> >> and on the
> >>  > > host in
> >>  > > vdsm.log I find:
> >>  > >
> >>  > > 2020-05-11 09:52:25,433+ ERROR (Reactor thread)
> >>  > > [ProtocolDetector.SSLHandshakeDispatcher] ssl
> >> handshake: SSLError,
> >>  > > address: :::159.149.129.220 (sslutils:264)
> >>  > >
> >>  > > So I tried to enroll the certificate from the UI and
> >> from the
> >>  > events
> >>  > > tab
> >>  > > I sow the enrolling was successful but:
> >>  > >
> >>  > > [root@manager ~]# openssl s_client -connect
> >> 172.20.22.78:54321 
> >>  > 
> >>  > > 
> >>  > >
> >>  > > 140084336994192:error:140790E5:SSL
> >routines:ssl23_write:ssl
> >>  > handshake
> >>  > > failure:s23_lib.c:177:
> >>  > > CONNECTED(0003)
> >>  > > ---
> >>  > > no peer certificate available
> >>  > > ---
> >>  > >
> >>  > > there's still some issue with the certificates.. so
> >on the
> >>  > host again:
> >>  > >
> >>  > > [root@cn128 vdsm]# find /etc/pki/vdsm/ -type f -cmin
> >-10|
> >>  

[ovirt-users] Re: update to 4.2.8 fails

[Nir Levy]

repo:
https://github.com/oVirt/ovirt-site.git

patch attached for review.

On Fri, Feb 22, 2019 at 5:04 PM Sandro Bonazzola 
wrote:

> Nir Levy can you please add this to upgrade guide? Please add Yuval to
> reviewers for the PR.
>
> Il dom 17 feb 2019, 10:49 Yuval Turgeman  ha scritto:
>
>> It's mentioned in the release notes [1], probably worth to mention it in
>> the general upgrade guide
>>
>> [1]
>> https://ovirt.org/release/4.3.0/#install--upgrade-from-previous-versions
>>
>>
>> On Fri, Feb 15, 2019 at 1:05 PM Greg Sheremeta 
>> wrote:
>>
>>>
>>> On Thu, Feb 14, 2019 at 11:06 PM Vincent Royer 
>>> wrote:
>>>
>>>> Greg,
>>>>
>>>> Can I contribute?
>>>>
>>>
>>> I'm so glad you asked -- YES, please do! :)
>>>
>>> The docs, along with the rest of the ovirt.org site, are in GitHub, and
>>> very easily editable. To edit a single page -- in the page footer, click
>>> "Edit this page on GitHub". That will create a simple Pull Request for
>>> review.
>>>
>>> You can also clone and fork like any other project, if you are more
>>> comfortable with that. If you want to run the site locally,
>>> https://github.com/oVirt/ovirt-site/blob/master/CONTRIBUTING.md
>>>
>>> Let me know if you have any questions!
>>>
>>>
>>>>
>>>>
>>>>
>>>> On Thu, Feb 14, 2019 at 2:05 PM Greg Sheremeta 
>>>> wrote:
>>>>
>>>>>
>>>>> On Thu, Feb 14, 2019 at 2:16 PM Vincent Royer 
>>>>> wrote:
>>>>>
>>>>>> Greg,
>>>>>>
>>>>>> The first thing on the list in your link is to check what repos are
>>>>>> enabled with yum repolist, but makes no mention of what repos
>>>>>> *should *be enabled on node-ng, nor what to do about it if you have
>>>>>> the wrong ones. I've never had an Ovirt update go the way it was 
>>>>>> "supposed"
>>>>>> to go, despite having a bunch of documentation at hand.  Usually I end up
>>>>>> blowing away the hosts and starting with a fresh ISO.
>>>>>>
>>>>>> The page you linked makes no mention of the command Edward mentioned
>>>>>> that got things working for me:
>>>>>>
>>>>>> yum update ovirt-node-ng-image-update
>>>>>>
>>>>>> All the instructions I can find just say to run yum update, but that
>>>>>> resulted in a bunch of dependency errors for me, on a normal 4.2.6 node
>>>>>> install that I haven't touched since installation.  Why?  If I'm 
>>>>>> following
>>>>>> the instructions, shouldn't it work?
>>>>>>
>>>>>>
>>>>>> running this command in the upgrade guide:
>>>>>>
>>>>>> [image: image.png]
>>>>>>
>>>>>> Gives me "This system is not registered with an entitlement server".
>>>>>>  Is that an outdated instruction?  Does it apply to the particular update
>>>>>> I'm trying to apply?   No way to tell...
>>>>>>
>>>>>
>>>>> It only applies to RHEL. If you are not on RHEL, you wouldn't run that.
>>>>> So it definitely needs improvement.
>>>>>
>>>>>
>>>>>>
>>>>>> What would really help is a clear separation between commands
>>>>>> intended for centos/RHEL and commands intended for Node.  As an outsider,
>>>>>> it's very difficult to know.   Every chapter, where there is any 
>>>>>> difference
>>>>>> in procedure, the documents should be split with RHEL on one side and 
>>>>>> NODE
>>>>>> on the other.
>>>>>>
>>>>>
>>>>> +1.
>>>>>
>>>>>
>>>>>>
>>>>>> The documentation would also benefit from tags like dates and
>>>>>> versions that they apply to.  "Valid for Ovirt 4.2.6 to 4.2.8 as of Feb 
>>>>>> 2,
>>>>>> 2019".Then the documents should be tested and the dates/versions
>>>>>> adjusted, or the docs adjusted, as needed.
>>>>>>
>>>>>
>>>>> Agree.
>>>>>
>>&g