[ovirt-users] Re: R: Re: Failed to deploy ovirt engine with "hosted-engine --deploy"

[RabidCicada]

Hey there,
I posted solutions here earlier:
https://mail.google.com/mail/u/0/?tab=wm#search/in%3Asent+ovirt+users/LXphbRLrghxkrJpsrHvKlSpMLsGVkKkHknsLMHHKnQV

Thanks for the gist reference giuseppe.  I haven't updated the approach in
a while...but should absolutely explain the reasoning and I think should
still just work:).

On Sat, Sep 8, 2018 at 11:31 AM Giuseppe Ragusa 
wrote:

> Hi all,
> I confirm that:
>
> 1) the machine type pc-i440fx-rhel7.2.0 is enough to solve the problem
>
> 2) the OVEHOSTED_VM/emulatedMachine variable in answers file is NOT enough
> (the engine vm hangs again as soon as it gets rebooted, after the
> successful initial setup), but the VDSM hook in
> https://gist.github.com/RabidCicada/40655db1582ca5d07c9bbf2c429cdd01
> solves the problem (for further vms too, arguably)
>
> Many thanks.
>
> Best regards,
> Giuseppe
> 
> Da: Giuseppe Ragusa 
> Inviato: venerdì 7 settembre 2018 16:04
> A: Simone Tiraboschi; bon...@gmail.com
> Cc: users
> Oggetto: R: [ovirt-users] Re: Failed to deploy ovirt engine with
> "hosted-engine --deploy"
>
> Hi Simone,
> sorry for the late comment (I just found this thread while researching on
> nested oVirt-on-VMware issues).
>
> It seems that the problem is generally known and has been solved upstream
> in kernel 4.16:
>
> https://bugs.launchpad.net/qemu/+bug/1636217
>
> Any ETA on backports in RHEL7? ;-)
> (Note: should I open a Bugzilla on kernel package for this?)
> I seems that Hyper-V would experience the same troubles and I think that
> making nested-inside-Windows-host work should be deemed fairly important
> for demo/trial/lab uses... :-)
>
> In the meanwhile, workarounds have been published for the new
> Ansible-based setup flow:
>
>
> https://lists.ovirt.org/archives/list/users@ovirt.org/thread/UDVRS5I64WLFJJF7YXKXZNTQOZRJ6DOJ/?sort=date
>
> on the other side, I can confirm that with the legacy setup flow (with
> option --noansible) adding the following to the answer file avoids the
> problem (did not test whether newer types like pc-i440fx-rhel7.2.0 solve it
> too):
>
> OVEHOSTED_VM/emulatedMschine=str:rhel6.0.0
>
> Many thanks.
>
> Best regards,
> Giuseppe
>
> 
> Da: Simone Tiraboschi 
> Inviato: lunedì 30 luglio 2018 09:20
> A: bon...@gmail.com
> Cc: users
> Oggetto: [ovirt-users] Re: Failed to deploy ovirt engine with
> "hosted-engine --deploy"
>
>
>
> On Mon, Jul 30, 2018 at 6:04 AM Bong Shau Fui  bon...@gmail.com> wrote:
> Hi Simone:
>Yes, it's in a nested environment.  L0 is vmware esxi 5.5.
>
> I know for sue that a nested kvm env over vmware esxi is still
> problematic; kvm over kvm works fine instead.
>
>
>
> regards,
> Bong SF
> ___
> Users mailing list -- users@ovirt.org<mailto:users@ovirt.org>
> To unsubscribe send an email to users-le...@ovirt.org users-le...@ovirt.org>
> Privacy Statement: https://www.ovirt.org/site/privacy-policy/
> oVirt Code of Conduct:
> https://www.ovirt.org/community/about/community-guidelines/
> List Archives:
> https://lists.ovirt.org/archives/list/users@ovirt.org/message/YQZHWNK2CPSWGWIOB3WOJJ2QZK4HGLAD/
> mailto:users-le...@ovirt.org>mailto:users@ovirt.org> <mailto:bon...@gmail.com>
> 
> ___
> Users mailing list -- users@ovirt.org
> To unsubscribe send an email to users-le...@ovirt.org
> Privacy Statement: https://www.ovirt.org/site/privacy-policy/
> oVirt Code of Conduct:
> https://www.ovirt.org/community/about/community-guidelines/
> List Archives:
> https://lists.ovirt.org/archives/list/users@ovirt.org/message/W4GOU6N43OGPPDTKQ2TUOI6O2VI2IUOS/
>
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/UQKFLO26QQ6AXCTAOFKPRU6S3WV67T2P/

[ovirt-users] Re: OVA export/import

[RabidCicada]

Third attempt to send to list also

> Yaniv,
> Most of this is obe...but I'll answer the questions.  It's been resolved
> as in my other email to this chain.
>
>
> On Thu, Jun 14, 2018, 4:56 PM Yaniv Kaul  wrote:
>
>>
>>
>> On Wed, Jun 13, 2018, 4:05 PM RabidCicada  wrote:
>>
>>> All,
>>> I recently tried to used the OVA export/import functionality.  It
>>> seems I misunderstood the intentions.  I expected OVA export/import to
>>> be reciprocal functionalities.  I expected what we export as OVA to be
>>> importable as OVA in ovirt.
>>>
>>> What I have found is the following:
>>>
>>> * It seems OVA export does export an OVA, though the format is not
>>> spec compliant
>>>
>>
>> Which spec?
>>
> I believe every version of the dmtd ovf specification.
>
>>
>>   * We use 'disk/' in the HostResource xml field instead of '/disk/'
>>> for example in the ovf.
>>>
>>
>> Who's 'we'?
>>
> Ovirt software/group.  I was including myself in a friendly manner soas to
> not sound accusatory.  I know it's a group effort :).
>
>
>> * It seems OVA import is intended specifically for VMWare OVA's?
>>>
>>
>> That's a common case, but we can look at other OVAs. The problem is that
>> the spec is quite loose.
>> Y.
>>
>
> Agreed.  I know that's why it's hard to be inter-compatible.
>
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/PYU5FT7MFQNNRF6FN5WYHQFORLF32YWZ/

[ovirt-users] Re: Ovirt-Hosted-Engine-Install-Purge Ansible Script Snippet

[RabidCicada]

I went ahead and tried to add this script to the appropriate section
of the ovirt.org website.  Opened a pull request here:
https://github.com/oVirt/ovirt-site/pull/1650

On Thu, Jun 14, 2018 at 10:15 AM, RabidCicada  wrote:
> All,
> I put together a full purge ansible script for CentOS.  I figured I'd
> share it here.  I ran into many issues with spurious failures of
> installs when I would try to reinstall ovirt on CentOS after having
> "removed" it.
>
> I was developing an automated install of our own software along with
> ovirt but along the way I had to reinstall on the same box numerous
> times.  Initially I'd get spurious failures from leftover cruft that
> would get in the way.  I eventually found
> this:https://www.ovirt.org/documentation/how-to/hosted-engine/#recoving-from-failed-install
> and then crafted my own ansible purge script.
>
> It uses ovirts own engine cleanup script first if it exists, then
> fully cleans every other thing that can affect it.
>
> Tested on CentOS, probably will work on other distros without too much
> trouble.Hopefully someone saves themselves some grief
>
> - name: Clean Old Install
>
>   #This attempts to remove all old cruft from previous install attempts
>
>   #The reason we include the ovirt packages is so that they can be 
> reinstall
>
>   #At potentially newer versions along with dependency packages
>
>   block:
>
> - name: Detect existing cleanup script
>
>   shell: which ovirt-hosted-engine-cleanup | cat
>
>   register: ohes_cleanup
>
> - name: Debug ohes_cleanup.stdout
>
>   debug:
>
> var: ohes_cleanup.stdout
>
> - name: Run Ovirt's Hosted Engine Cleanup Script
>
>   shell: ovirt-hosted-engine-cleanup -q
>
>   when: ohes_cleanup.stdout != ""
>
> - name: Clean old packages
>
>   package:
>
> name: "{{item}}"
>
> state: absent
>
>   with_items:
>
> - "*vdsm*"
>
> - "*ovirt*"
>
> - "*libvirt*"
>
> - "*cockpit*"
>
> - name: Remove old configs etc
>
>   shell: "rm -rf /etc/{{item}}"
>
>   args:
>
> warn: False
>
>   with_items:
>
> - "/etc/*ovirt*"
>
> - "/etc/*vdsm*"
>
> - "/etc/libvirt/qemu/HostedEngine*"
>
> - "/etc/*libvirt*"
>
> - "/etc/guacamole"
>
> - "/etc/pki/vdsm"
>
> - "/etc/pki/libvirt"
>
> - "/etc/pki/CA"
>
> - "/etc/pki/keystore"
>
> - "/etc/ovirt-hosted-engine"
>
> - "/var/lib/libvirt/"
>
> - "/var/lib/vdsm/"
>
> - "/var/lib/ovirt-hosted-engine-*"
>
> - "/var/log/ovirt-hosted-engine-setup/"
>
> - "/var/cache/libvirt/"
>
> - "/etc/libvirt/nwfilter/vdsm-no-mac-spoofing.xml"
>
> - "/var/cache/tomcat/temp/*"
>
> - "/var/cache/tomcat/work/Catalina/localhost/*"
>
> - "/usr/share/tomcat/webapps/guacamole"
>
> - "/usr/share/tomcat/webapps/guacamole.war"
>
> - "/etc/guacamole/extensions/guacamole*"
>
> - name: Clean old repo files
>
>   shell: "rm -rf /etc/yum.repos.d/{{item}}"
>
>   args:
>
> warn: False
>
>   with_items:
>
> - "ovirt*"
>
> - "virt*"
>
> - name: clean interface configs
>
>   shell: "rm -rf /etc/sysconfig/network-scripts/ifcfg-ovirtmgmt"
>
>   args:
>
> warn: False
>
> - name: clean network stuff
>
>   shell: "{{item}}"
>
>   args:
>
> warn: False
>
>   with_items:
>
> - "brctl delbr ovirtmgmt | cat"
>
> - "ip link del ovirtmgmt | cat"
>
> - "ip link del dummy0 | cat"
>
> - "ip link del virbr0 | cat"
>
> - "ip link del virbr0-nic | cat"
>
> - 'ip link del \;vdsmdummy\; | cat'
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/7BWNGA77HKVJJUJE3S5JUQHXK2XMBJSB/

[ovirt-users] Ovirt-Hosted-Engine-Install-Purge Ansible Script Snippet

[RabidCicada]

All,
I put together a full purge ansible script for CentOS.  I figured I'd
share it here.  I ran into many issues with spurious failures of
installs when I would try to reinstall ovirt on CentOS after having
"removed" it.

I was developing an automated install of our own software along with
ovirt but along the way I had to reinstall on the same box numerous
times.  Initially I'd get spurious failures from leftover cruft that
would get in the way.  I eventually found
this:https://www.ovirt.org/documentation/how-to/hosted-engine/#recoving-from-failed-install
and then crafted my own ansible purge script.

It uses ovirts own engine cleanup script first if it exists, then
fully cleans every other thing that can affect it.

Tested on CentOS, probably will work on other distros without too much
trouble.Hopefully someone saves themselves some grief

- name: Clean Old Install

  #This attempts to remove all old cruft from previous install attempts

  #The reason we include the ovirt packages is so that they can be reinstall

  #At potentially newer versions along with dependency packages

  block:

- name: Detect existing cleanup script

  shell: which ovirt-hosted-engine-cleanup | cat

  register: ohes_cleanup

- name: Debug ohes_cleanup.stdout

  debug:

var: ohes_cleanup.stdout

- name: Run Ovirt's Hosted Engine Cleanup Script

  shell: ovirt-hosted-engine-cleanup -q

  when: ohes_cleanup.stdout != ""

- name: Clean old packages

  package:

name: "{{item}}"

state: absent

  with_items:

- "*vdsm*"

- "*ovirt*"

- "*libvirt*"

- "*cockpit*"

- name: Remove old configs etc

  shell: "rm -rf /etc/{{item}}"

  args:

warn: False

  with_items:

- "/etc/*ovirt*"

- "/etc/*vdsm*"

- "/etc/libvirt/qemu/HostedEngine*"

- "/etc/*libvirt*"

- "/etc/guacamole"

- "/etc/pki/vdsm"

- "/etc/pki/libvirt"

- "/etc/pki/CA"

- "/etc/pki/keystore"

- "/etc/ovirt-hosted-engine"

- "/var/lib/libvirt/"

- "/var/lib/vdsm/"

- "/var/lib/ovirt-hosted-engine-*"

- "/var/log/ovirt-hosted-engine-setup/"

- "/var/cache/libvirt/"

- "/etc/libvirt/nwfilter/vdsm-no-mac-spoofing.xml"

- "/var/cache/tomcat/temp/*"

- "/var/cache/tomcat/work/Catalina/localhost/*"

- "/usr/share/tomcat/webapps/guacamole"

- "/usr/share/tomcat/webapps/guacamole.war"

- "/etc/guacamole/extensions/guacamole*"

- name: Clean old repo files

  shell: "rm -rf /etc/yum.repos.d/{{item}}"

  args:

warn: False

  with_items:

- "ovirt*"

- "virt*"

- name: clean interface configs

  shell: "rm -rf /etc/sysconfig/network-scripts/ifcfg-ovirtmgmt"

  args:

warn: False

- name: clean network stuff

  shell: "{{item}}"

  args:

warn: False

  with_items:

- "brctl delbr ovirtmgmt | cat"

- "ip link del ovirtmgmt | cat"

- "ip link del dummy0 | cat"

- "ip link del virbr0 | cat"

- "ip link del virbr0-nic | cat"

- 'ip link del \;vdsmdummy\; | cat'
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/RHPHB4NUS6ETZZ7OP73F2LUC6U2SG3L6/

[ovirt-users] Re: OVA export/import

[RabidCicada]

All,
It turns out the problem was enabling ssh/firewall rules on the
expected ssh port for ovirt-engine.  Error reporting in 4.2.4 is
better and will fail earlier in the process according to the devs.  If
you experience symptoms similar to mine try making sure no firewall
rules or sshd configurations are preventing ansible from talking to
'localhost' on the expected port.

I have an install where we try to lock down the node after
hosted-engine --deploy and we change the default ssh port.  Most
engine things had functioned properly until this so I was unaware that
it had caused a problem.

You can set OVEHOSTED_NETWORK/sshdPort in your answers file for
hosted-engine --deploy to correctly set the port to an exotic port
from the getgo.

Thanks for the help devs, and I'm glad we have better error messages,
or at least earlier failing, coming in 4.2.4.

~Kyle


On Wed, Jun 13, 2018 at 6:01 PM, Arik Hadas  wrote:
>
>
> On Wed, Jun 13, 2018 at 8:36 PM, RabidCicada  wrote:
>>
>> Arik,
>> I've opened a bug here with all my information:
>> https://bugzilla.redhat.com/show_bug.cgi?id=1590948
>
>
> Thanks
>
>>
>>
>>
>> ~Kyle
>>
>> On Wed, Jun 13, 2018 at 10:04 AM, Arik Hadas  wrote:
>> >
>> >
>> > On Wed, Jun 13, 2018 at 4:58 PM, RabidCicada 
>> > wrote:
>> >>
>> >> Thanks for the quick reply,
>> >>
>> >> I am using 4.2.3 and Ovirt cannot import it's own generated OVA.  That
>> >> is why my whole investigation started.  Is it known to work for anyone
>> >> else?  Based on your statement it seems like it should be capable of
>> >> importing it's own;  But my install, perhaps JUST my install and
>> >> no-one elses :), cannot do so.  Can anyone verify import of ovirt
>> >> generated OVA's works on ovirt 4.2.3?
>> >
>> >
>> > I would suggest to file a bug and attach the engine and ansible logs so
>> > we
>> > can investigate why it fails for you.
>> > It generally works in 4.2.3 but there were some fixes we've made in this
>> > area - you may have encountered some of those issues or discovered a new
>> > one
>> > :)
>> >
>> >>
>> >>
>> >> ~Kyle
>> >>
>> >> On Wed, Jun 13, 2018 at 9:54 AM, Arik Hadas  wrote:
>> >> >
>> >> >
>> >> > On Wed, Jun 13, 2018 at 4:04 PM, RabidCicada 
>> >> > wrote:
>> >> >>
>> >> >> All,
>> >> >> I recently tried to used the OVA export/import functionality.  It
>> >> >> seems I misunderstood the intentions.  I expected OVA export/import
>> >> >> to
>> >> >> be reciprocal functionalities.  I expected what we export as OVA to
>> >> >> be
>> >> >> importable as OVA in ovirt.
>> >> >>
>> >> >> What I have found is the following:
>> >> >>
>> >> >> * It seems OVA export does export an OVA, though the format is not
>> >> >> spec compliant
>> >> >>   * We use 'disk/' in the HostResource xml field instead of '/disk/'
>> >> >> for example in the ovf.
>> >> >> * It seems OVA import is intended specifically for VMWare OVA's?
>> >> >> * I thought I read about export domains being deprecated, and
>> >> >> incorrectly assumed the recently added OVA export/import was to make
>> >> >> things better for one off import/export.
>> >> >>
>> >> >>
>> >> >> Can anyone clarify the following:
>> >> >> * Is OVA import really just VMWare OVA's?
>> >> >
>> >> >
>> >> > No, you can provide OVAs that were generated by oVirt as an input as
>> >> > well.
>> >> > See examples [1] and [2] for doing this via the python-sdk. Doing
>> >> > that
>> >> > from
>> >> > the webadmin is trivial is exactly like importing OVA that was
>> >> > generated
>> >> > by
>> >> > VMware.
>> >> >
>> >> > [1]
>> >> >
>> >> >
>> >> > https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/import_vm_from_ova.py
>> >> > [2]
>> >> >
>> >> >
>> >> > https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/upload_ova_as_vm.py
>> >> >
>> >> >>
>> >> >>

[ovirt-users] Re: OVA export/import

[RabidCicada]

Arik,
I've opened a bug here with all my information:
https://bugzilla.redhat.com/show_bug.cgi?id=1590948

~Kyle

On Wed, Jun 13, 2018 at 10:04 AM, Arik Hadas  wrote:
>
>
> On Wed, Jun 13, 2018 at 4:58 PM, RabidCicada  wrote:
>>
>> Thanks for the quick reply,
>>
>> I am using 4.2.3 and Ovirt cannot import it's own generated OVA.  That
>> is why my whole investigation started.  Is it known to work for anyone
>> else?  Based on your statement it seems like it should be capable of
>> importing it's own;  But my install, perhaps JUST my install and
>> no-one elses :), cannot do so.  Can anyone verify import of ovirt
>> generated OVA's works on ovirt 4.2.3?
>
>
> I would suggest to file a bug and attach the engine and ansible logs so we
> can investigate why it fails for you.
> It generally works in 4.2.3 but there were some fixes we've made in this
> area - you may have encountered some of those issues or discovered a new one
> :)
>
>>
>>
>> ~Kyle
>>
>> On Wed, Jun 13, 2018 at 9:54 AM, Arik Hadas  wrote:
>> >
>> >
>> > On Wed, Jun 13, 2018 at 4:04 PM, RabidCicada 
>> > wrote:
>> >>
>> >> All,
>> >> I recently tried to used the OVA export/import functionality.  It
>> >> seems I misunderstood the intentions.  I expected OVA export/import to
>> >> be reciprocal functionalities.  I expected what we export as OVA to be
>> >> importable as OVA in ovirt.
>> >>
>> >> What I have found is the following:
>> >>
>> >> * It seems OVA export does export an OVA, though the format is not
>> >> spec compliant
>> >>   * We use 'disk/' in the HostResource xml field instead of '/disk/'
>> >> for example in the ovf.
>> >> * It seems OVA import is intended specifically for VMWare OVA's?
>> >> * I thought I read about export domains being deprecated, and
>> >> incorrectly assumed the recently added OVA export/import was to make
>> >> things better for one off import/export.
>> >>
>> >>
>> >> Can anyone clarify the following:
>> >> * Is OVA import really just VMWare OVA's?
>> >
>> >
>> > No, you can provide OVAs that were generated by oVirt as an input as
>> > well.
>> > See examples [1] and [2] for doing this via the python-sdk. Doing that
>> > from
>> > the webadmin is trivial is exactly like importing OVA that was generated
>> > by
>> > VMware.
>> >
>> > [1]
>> >
>> > https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/import_vm_from_ova.py
>> > [2]
>> >
>> > https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/upload_ova_as_vm.py
>> >
>> >>
>> >> * Are we really not spec compliant with the OVA/OVF format as my
>> >> example above shows?  OR do I misunderstand something?
>> >
>> >
>> > Well, we now use the same code for parsing OVFs that were generated by
>> > oVirt
>> > and ones that were generated by VMware. So that's an evidence for being
>> > compliant with the specification to some degree.
>> >
>> > Indeed, there may be things that we don't use when reading the OVF, like
>> > the
>> > field you mentioned, and therefore there may be some gaps here and
>> > there.
>> >
>> > Is this the only gap you found so far? if so then I would say we are in
>> > a
>> > really good place since fixing that should be trivial, see [3]. You are
>> > more
>> > than welcome to send a patch!
>> >
>> > [3]
>> >
>> > https://github.com/oVirt/ovirt-engine/blob/master/backend/manager/modules/utils/src/main/java/org/ovirt/engine/core/utils/ovf/OvfOvaWriter.java#L179
>> >
>> >
>> >>
>> >> * After spending my time yesterday, I recall now that the deprecation
>> >> of export domain is in favor of detaching and re-attaching ANY data
>> >> domain.
>> >
>> >
>> > Depending what you're looking for.
>> > If you want to *move* your entities (VMs/templates) from one deployment
>> > to
>> > another then sure, detach the domain (that will cause those entities to
>> > be
>> > deleted from the source environment), attach the domain to the new
>> > deployment and register the entities you want. That is relatively quick
>> > operation since no copy is needed.
>> >
>> > Export domains/OVAs a

[ovirt-users] Re: OVA export/import

[RabidCicada]

I am using the Web GUI.  Not the python sdk.

On Wed, Jun 13, 2018 at 9:58 AM, RabidCicada  wrote:
> Thanks for the quick reply,
>
> I am using 4.2.3 and Ovirt cannot import it's own generated OVA.  That
> is why my whole investigation started.  Is it known to work for anyone
> else?  Based on your statement it seems like it should be capable of
> importing it's own;  But my install, perhaps JUST my install and
> no-one elses :), cannot do so.  Can anyone verify import of ovirt
> generated OVA's works on ovirt 4.2.3?
>
> ~Kyle
>
> On Wed, Jun 13, 2018 at 9:54 AM, Arik Hadas  wrote:
>>
>>
>> On Wed, Jun 13, 2018 at 4:04 PM, RabidCicada  wrote:
>>>
>>> All,
>>> I recently tried to used the OVA export/import functionality.  It
>>> seems I misunderstood the intentions.  I expected OVA export/import to
>>> be reciprocal functionalities.  I expected what we export as OVA to be
>>> importable as OVA in ovirt.
>>>
>>> What I have found is the following:
>>>
>>> * It seems OVA export does export an OVA, though the format is not
>>> spec compliant
>>>   * We use 'disk/' in the HostResource xml field instead of '/disk/'
>>> for example in the ovf.
>>> * It seems OVA import is intended specifically for VMWare OVA's?
>>> * I thought I read about export domains being deprecated, and
>>> incorrectly assumed the recently added OVA export/import was to make
>>> things better for one off import/export.
>>>
>>>
>>> Can anyone clarify the following:
>>> * Is OVA import really just VMWare OVA's?
>>
>>
>> No, you can provide OVAs that were generated by oVirt as an input as well.
>> See examples [1] and [2] for doing this via the python-sdk. Doing that from
>> the webadmin is trivial is exactly like importing OVA that was generated by
>> VMware.
>>
>> [1]
>> https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/import_vm_from_ova.py
>> [2]
>> https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/upload_ova_as_vm.py
>>
>>>
>>> * Are we really not spec compliant with the OVA/OVF format as my
>>> example above shows?  OR do I misunderstand something?
>>
>>
>> Well, we now use the same code for parsing OVFs that were generated by oVirt
>> and ones that were generated by VMware. So that's an evidence for being
>> compliant with the specification to some degree.
>>
>> Indeed, there may be things that we don't use when reading the OVF, like the
>> field you mentioned, and therefore there may be some gaps here and there.
>>
>> Is this the only gap you found so far? if so then I would say we are in a
>> really good place since fixing that should be trivial, see [3]. You are more
>> than welcome to send a patch!
>>
>> [3]
>> https://github.com/oVirt/ovirt-engine/blob/master/backend/manager/modules/utils/src/main/java/org/ovirt/engine/core/utils/ovf/OvfOvaWriter.java#L179
>>
>>
>>>
>>> * After spending my time yesterday, I recall now that the deprecation
>>> of export domain is in favor of detaching and re-attaching ANY data
>>> domain.
>>
>>
>> Depending what you're looking for.
>> If you want to *move* your entities (VMs/templates) from one deployment to
>> another then sure, detach the domain (that will cause those entities to be
>> deleted from the source environment), attach the domain to the new
>> deployment and register the entities you want. That is relatively quick
>> operation since no copy is needed.
>>
>> Export domains/OVAs are targeted for different scenario - when you want to
>> *copy* you entities from one deployment to another. You export them
>> somewhere and than import them from that place.
>>
>>>
>>> * What is the purpose of our OVA export if we ourselves indeed cannot
>>> import our own OVA?  Or can you tell me how we import our own if I'm
>>> wrong.
>>
>>
>> Already answered.
>>
>>>
>>>
>>> Please feel free to overshare details :), as any context will help
>>> prevent more confusion on my part.
>>>
>>> ~Kyle
>>> ___
>>> Users mailing list -- users@ovirt.org
>>> To unsubscribe send an email to users-le...@ovirt.org
>>> Privacy Statement: https://www.ovirt.org/site/privacy-policy/
>>> oVirt Code of Conduct:
>>> https://www.ovirt.org/community/about/community-guidelines/
>>> List Archives:
>>> https://lists.ovirt.org/archives/list/users@ovirt.org/message/MYWHJF4ZJS4VSJ2RWBSYDPPOWTI6VWW5/
>>
>>
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/6V4F3Z5EPEFYSHLMNCWXUKN2JH7BORVZ/

[ovirt-users] Re: OVA export/import

[RabidCicada]

Thanks for the quick reply,

I am using 4.2.3 and Ovirt cannot import it's own generated OVA.  That
is why my whole investigation started.  Is it known to work for anyone
else?  Based on your statement it seems like it should be capable of
importing it's own;  But my install, perhaps JUST my install and
no-one elses :), cannot do so.  Can anyone verify import of ovirt
generated OVA's works on ovirt 4.2.3?

~Kyle

On Wed, Jun 13, 2018 at 9:54 AM, Arik Hadas  wrote:
>
>
> On Wed, Jun 13, 2018 at 4:04 PM, RabidCicada  wrote:
>>
>> All,
>> I recently tried to used the OVA export/import functionality.  It
>> seems I misunderstood the intentions.  I expected OVA export/import to
>> be reciprocal functionalities.  I expected what we export as OVA to be
>> importable as OVA in ovirt.
>>
>> What I have found is the following:
>>
>> * It seems OVA export does export an OVA, though the format is not
>> spec compliant
>>   * We use 'disk/' in the HostResource xml field instead of '/disk/'
>> for example in the ovf.
>> * It seems OVA import is intended specifically for VMWare OVA's?
>> * I thought I read about export domains being deprecated, and
>> incorrectly assumed the recently added OVA export/import was to make
>> things better for one off import/export.
>>
>>
>> Can anyone clarify the following:
>> * Is OVA import really just VMWare OVA's?
>
>
> No, you can provide OVAs that were generated by oVirt as an input as well.
> See examples [1] and [2] for doing this via the python-sdk. Doing that from
> the webadmin is trivial is exactly like importing OVA that was generated by
> VMware.
>
> [1]
> https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/import_vm_from_ova.py
> [2]
> https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/upload_ova_as_vm.py
>
>>
>> * Are we really not spec compliant with the OVA/OVF format as my
>> example above shows?  OR do I misunderstand something?
>
>
> Well, we now use the same code for parsing OVFs that were generated by oVirt
> and ones that were generated by VMware. So that's an evidence for being
> compliant with the specification to some degree.
>
> Indeed, there may be things that we don't use when reading the OVF, like the
> field you mentioned, and therefore there may be some gaps here and there.
>
> Is this the only gap you found so far? if so then I would say we are in a
> really good place since fixing that should be trivial, see [3]. You are more
> than welcome to send a patch!
>
> [3]
> https://github.com/oVirt/ovirt-engine/blob/master/backend/manager/modules/utils/src/main/java/org/ovirt/engine/core/utils/ovf/OvfOvaWriter.java#L179
>
>
>>
>> * After spending my time yesterday, I recall now that the deprecation
>> of export domain is in favor of detaching and re-attaching ANY data
>> domain.
>
>
> Depending what you're looking for.
> If you want to *move* your entities (VMs/templates) from one deployment to
> another then sure, detach the domain (that will cause those entities to be
> deleted from the source environment), attach the domain to the new
> deployment and register the entities you want. That is relatively quick
> operation since no copy is needed.
>
> Export domains/OVAs are targeted for different scenario - when you want to
> *copy* you entities from one deployment to another. You export them
> somewhere and than import them from that place.
>
>>
>> * What is the purpose of our OVA export if we ourselves indeed cannot
>> import our own OVA?  Or can you tell me how we import our own if I'm
>> wrong.
>
>
> Already answered.
>
>>
>>
>> Please feel free to overshare details :), as any context will help
>> prevent more confusion on my part.
>>
>> ~Kyle
>> ___
>> Users mailing list -- users@ovirt.org
>> To unsubscribe send an email to users-le...@ovirt.org
>> Privacy Statement: https://www.ovirt.org/site/privacy-policy/
>> oVirt Code of Conduct:
>> https://www.ovirt.org/community/about/community-guidelines/
>> List Archives:
>> https://lists.ovirt.org/archives/list/users@ovirt.org/message/MYWHJF4ZJS4VSJ2RWBSYDPPOWTI6VWW5/
>
>
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/FI4MX437OOMBOHXEQJNQQXQ3UW5KWBV5/

[ovirt-users] Ovirt Nested in VMWare (Working approach)

[RabidCicada]

I worked through this previously and posted in IRC yesterday about
working around running ovirt nested under VMWare.  I figured i'd post
it to the mail group so that it would be available in it's entirety
here for future reference:

The following gists explain the 3 major adjustments that have to be
made currently to get nested virt under vmware working:

https://gist.github.com/RabidCicada/993349b3a7eaf35637122f0122fdedb3
https://gist.github.com/RabidCicada/bc01eb0b13195faa26520c0fb666ec34
https://gist.github.com/RabidCicada/37ff6f1ed2afd8dd4edefac735536c69
https://gist.github.com/RabidCicada/40655db1582ca5d07c9bbf2c429cdd01

Hopefully it can save others the huge amount of time it took me to
figure it all out.

Specifically it's a problem with the --machine type at 3 major parts
of the vm creation for the hosted-engine.
1) You have to fix the virt-install to use --machine pc-i440fx-rhel7.2.0
2) You have to fix the guestfish spinup of the partially configured vm
to use --machine pc-i440fx-rhel7.2.0, where they edit the nic config
among other things
3) You have to tell vdsm on the node to use --machine pc-i440fx-rhel7.2.0

I don't know the underlying reason why a --machine type >
pc-i440fx-rhel7.2.0 breaks the system but it ends up resulting in a
bios freeze.

Again, Hope this helps someone.  Suggested only to use for a test/dev system.
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/UDVRS5I64WLFJJF7YXKXZNTQOZRJ6DOJ/

[ovirt-users] OVA export/import

[RabidCicada]

All,
I recently tried to used the OVA export/import functionality.  It
seems I misunderstood the intentions.  I expected OVA export/import to
be reciprocal functionalities.  I expected what we export as OVA to be
importable as OVA in ovirt.

What I have found is the following:

* It seems OVA export does export an OVA, though the format is not
spec compliant
  * We use 'disk/' in the HostResource xml field instead of '/disk/'
for example in the ovf.
* It seems OVA import is intended specifically for VMWare OVA's?
* I thought I read about export domains being deprecated, and
incorrectly assumed the recently added OVA export/import was to make
things better for one off import/export.


Can anyone clarify the following:
* Is OVA import really just VMWare OVA's?
* Are we really not spec compliant with the OVA/OVF format as my
example above shows?  OR do I misunderstand something?
* After spending my time yesterday, I recall now that the deprecation
of export domain is in favor of detaching and re-attaching ANY data
domain.
* What is the purpose of our OVA export if we ourselves indeed cannot
import our own OVA?  Or can you tell me how we import our own if I'm
wrong.

Please feel free to overshare details :), as any context will help
prevent more confusion on my part.

~Kyle
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/site/privacy-policy/
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/MYWHJF4ZJS4VSJ2RWBSYDPPOWTI6VWW5/

[ovirt-users] Re: ovirt host upgrade 4.2.2 -> 4.2.3

[RabidCicada]

I've had the exact same error.  My solution was to disable obsoletes logic
in yum.  Obsoletes =0 in yum.conf.

I haven't investigated much but it's caused by udisks2 obsoleting
storaged.

On Sun, May 13, 2018, 2:08 PM Artem Tambovskiy 
wrote:

> Hello,
>
> I'm upgrading my cluster from 4.2.2 to 4.2.3, HE upgrade went well, but
> having some issues with hosts upgrade: for some reason yum complaining
> about conflicts during transaction check:
>
>
> Transaction check error:
>   file /usr/share/cockpit/networkmanager/manifest.json from install of
> cockpit-system-160-3.el7.centos.noarch conflicts with file from package
> cockpit-networkmanager-160-1.el7.centos.noarch
>
> Any ideas about the reason for this?
>
> Regards,
> Artem
> ___
> Users mailing list -- users@ovirt.org
> To unsubscribe send an email to users-le...@ovirt.org
>
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org

Re: [ovirt-users] hosted-engine Debug Help

[RabidCicada]

Indeed that is the problem.  I am sshing as root

The only possible explanations I have for this working in the ovirt
autmated testing setup are:

   - Either the ovirt automated tests must be using a host inventory file
   with become: true?
   - The inventory file ssh credentials happen to be to vdsm user? (And
   become is just not doing anything)
   - Some other janky thing that allows the access


It also happens for the immediately following task `Initialize metadata
volume` and many more following that task.

I will correct and then open a pull request to fix with your input that you
believe it's a problem.

~Kyle


On Tue, Apr 3, 2018 at 3:08 PM, RabidCicada <rabidcic...@gmail.com> wrote:

> Alright.  By differential comparing to successfull postgres-command
> sudo-becomes.  It looks like the salient difference is that `Copy
> configuration archive to storage` is missing become:true.  Testing now.
>
> The similar postgres commands have become:true while this one does not.
>
> ~Kyle
>
> On Tue, Apr 3, 2018 at 1:19 PM, Simone Tiraboschi <stira...@redhat.com>
> wrote:
>
>>
>>
>> On Tue, Apr 3, 2018 at 5:21 PM, RabidCicada <rabidcic...@gmail.com>
>> wrote:
>>
>>> I've attached a full debug packet below.  I include the log file from
>>> ovirt-hosted-engine-setup.  I include relevant cmd line info.  I include
>>> info from the command line where epdb has a breakpoint in playbook.py from
>>> ansible itself.  I also include info from commands I ran after it failed.
>>> I also include attached the ferried over script in /root/.ansible/tmp that
>>> is run.
>>>
>>>
>>> *Output on cmd line:*
>>> [ INFO  ] TASK [Copy configuration archive to storage]
>>> [ ERROR ] fatal: [localhost]: FAILED! => {
>>>  "changed": true,
>>>  "cmd": [
>>>  "dd",
>>>  "bs=20480",
>>>  "count=1",
>>>  "oflag=direct",
>>>  "if=/var/tmp/localvmbCDQIR/5e
>>> f881f5-c992-48d2-b969-a0b6156bdf7c",
>>>  "of=/rhev/data-center/mnt/nod
>>> e.local:_srv_data/81292f3f-11d3-4e38-9afa-62e133aa8017/image
>>> s/c5510e77-1ee0-479c-b6cf-24c179313a45/5ef881f5-c992-48d2-b9
>>> 69-a0b6156bdf7c"
>>>  ],
>>>  "delta": "0:00:00.004336",
>>>  "end": "2018-04-03 15:01:55.581823",
>>>  "invocation": {
>>>  "module_args": {
>>>  "_raw_params": "dd bs=20480 count=1 oflag=direct
>>> if=\"/var/tmp/localvmbCDQIR/5ef881f5-c992-48d2-b969-a0b6156bdf7c\"
>>> of=\"/rhev/data-center/mnt/node.local:_srv_data/81292f3f-11d
>>> 3-4e38-9afa-62e133aa8017/images/c5510e77-1ee0-479c-b6cf-24c1
>>> 79313a45/5ef881f5-c992-48d2-b969-a0b6156bdf7c\"",
>>>  "_uses_shell": false,
>>>  "chdir": null,
>>>  "creates": null,
>>>  "executable": null,
>>>  "removes": null,
>>>  "stdin": null,
>>>  "warn": true
>>>  }
>>>  },
>>>  "msg": "non-zero return code",
>>>  "rc": 1,
>>>  "start": "2018-04-03 15:01:55.577487",
>>>  "stderr": "dd: failed to open ‘/rhev/data-center/mnt/node.lo
>>> cal:_srv_data/81292f3f-11d3-4e38-9afa-62e133aa8017/images/c5
>>> 510e77-1ee0-479c-b6cf-24c179313a45/5ef881f5-c992-48d2-b969-a0b6156bdf7c’:
>>> Permission denied",
>>>  "stderr_lines": [
>>>  "dd: failed to open ‘/rhev/data-center/mnt/node.lo
>>> cal:_srv_data/81292f3f-11d3-4e38-9afa-62e133aa8017/images/c5
>>> 510e77-1ee0-479c-b6cf-24c179313a45/5ef881f5-c992-48d2-b969-a0b6156bdf7c’:
>>> Permission denied"
>>>  ],
>>>  "stdout": "",
>>>  "stdout_lines": []
>>>  }
>>> [ ERROR ] Failed to execute stage 'Closing up': Failed executing
>>> ansible-playbook
>>>
>>
>> In the playbook we have on that task:
>> become_user: vd

Re: [ovirt-users] hosted-engine Debug Help

[RabidCicada]

Alright.  By differential comparing to successfull postgres-command
sudo-becomes.  It looks like the salient difference is that `Copy
configuration archive to storage` is missing become:true.  Testing now.

The similar postgres commands have become:true while this one does not.

~Kyle

On Tue, Apr 3, 2018 at 1:19 PM, Simone Tiraboschi <stira...@redhat.com>
wrote:

>
>
> On Tue, Apr 3, 2018 at 5:21 PM, RabidCicada <rabidcic...@gmail.com> wrote:
>
>> I've attached a full debug packet below.  I include the log file from
>> ovirt-hosted-engine-setup.  I include relevant cmd line info.  I include
>> info from the command line where epdb has a breakpoint in playbook.py from
>> ansible itself.  I also include info from commands I ran after it failed.
>> I also include attached the ferried over script in /root/.ansible/tmp that
>> is run.
>>
>>
>> *Output on cmd line:*
>> [ INFO  ] TASK [Copy configuration archive to storage]
>> [ ERROR ] fatal: [localhost]: FAILED! => {
>>  "changed": true,
>>  "cmd": [
>>  "dd",
>>  "bs=20480",
>>  "count=1",
>>  "oflag=direct",
>>  "if=/var/tmp/localvmbCDQIR/5e
>> f881f5-c992-48d2-b969-a0b6156bdf7c",
>>  "of=/rhev/data-center/mnt/nod
>> e.local:_srv_data/81292f3f-11d3-4e38-9afa-62e133aa8017/image
>> s/c5510e77-1ee0-479c-b6cf-24c179313a45/5ef881f5-c992-48d2-
>> b969-a0b6156bdf7c"
>>  ],
>>  "delta": "0:00:00.004336",
>>  "end": "2018-04-03 15:01:55.581823",
>>  "invocation": {
>>  "module_args": {
>>  "_raw_params": "dd bs=20480 count=1 oflag=direct
>> if=\"/var/tmp/localvmbCDQIR/5ef881f5-c992-48d2-b969-a0b6156bdf7c\"
>> of=\"/rhev/data-center/mnt/node.local:_srv_data/81292f3f-11d
>> 3-4e38-9afa-62e133aa8017/images/c5510e77-1ee0-479c-b6cf-24c1
>> 79313a45/5ef881f5-c992-48d2-b969-a0b6156bdf7c\"",
>>  "_uses_shell": false,
>>  "chdir": null,
>>  "creates": null,
>>  "executable": null,
>>  "removes": null,
>>  "stdin": null,
>>  "warn": true
>>  }
>>  },
>>  "msg": "non-zero return code",
>>  "rc": 1,
>>  "start": "2018-04-03 15:01:55.577487",
>>  "stderr": "dd: failed to open ‘/rhev/data-center/mnt/node.lo
>> cal:_srv_data/81292f3f-11d3-4e38-9afa-62e133aa8017/images/c5
>> 510e77-1ee0-479c-b6cf-24c179313a45/5ef881f5-c992-48d2-b969-a0b6156bdf7c’:
>> Permission denied",
>>  "stderr_lines": [
>>  "dd: failed to open ‘/rhev/data-center/mnt/node.lo
>> cal:_srv_data/81292f3f-11d3-4e38-9afa-62e133aa8017/images/c5
>> 510e77-1ee0-479c-b6cf-24c179313a45/5ef881f5-c992-48d2-b969-a0b6156bdf7c’:
>> Permission denied"
>>  ],
>>  "stdout": "",
>>  "stdout_lines": []
>>  }
>> [ ERROR ] Failed to execute stage 'Closing up': Failed executing
>> ansible-playbook
>>
>
> In the playbook we have on that task:
> become_user: vdsm
> become_method: sudo
>
> but I fear it got somehow ignored.
> I'll investigate it.
>
>
>>
>> *Output from ansible epdb tracepoint:*
>>
>> Using module file /usr/lib/python2.7/site-packag
>> es/ansible/modules/commands/command.py
>>  ESTABLISH LOCAL CONNECTION FOR USER: root
>>  EXEC /bin/sh -c 'echo ~ && sleep 0'
>>  EXEC /bin/sh -c '( umask 77 && mkdir -p "` echo
>> /root/.ansible/tmp/ansible-tmp-1522767715.36-81496549401055 `" && echo
>> ansible-tmp-1522767715.36-81496549401055="` echo
>> /root/.ansible/tmp/ansible-tmp-1522767715.36-81496549401055 `" ) &&
>> sleep 0'
>>  PUT /tmp/tmpGMGdjh TO /root/.ansible/tmp/ansible-tmp
>> -1522767715.36-81496549401055/command.py
>>  EXEC /bin/sh -c 'chmod u+x 
>> /root/.ansible/tmp/ansible-tmp-1522767715.36-81496549401055/
>> /root/.ansible/tmp/ansible-tmp-1522767715.36-81496549401055/command.py
>> && sl

Re: [ovirt-users] Persistent dummy interface

[RabidCicada]

I do this on CentOS with /etc/modules-lomodules-loadad.d.
You can stick a file in there that is basically just a bash script that
loads your driver and then does the ip link command all the same time.

On Tue, Apr 3, 2018, 4:20 AM  wrote:

> Hi,
> What is the right way to get a persistent dummy interface ?
> I need an internal network for lab purpoposes, so I created  a dummy link
> "ip link add dummy_1 type dummy" and I was able to assign my logical
> network to it.
> But how to make it persistent ? Of course, after a reboot, the dummy
> device has gone and the server becomes non operational because of this
> non-compliant network. I have to recereate it by hand in order to get
> things work.
> Regards
>
> --
> FreeMail powered by mail.fr
> ___
> Users mailing list
> Users@ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>
___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users

[ovirt-users] hosted-engine Debug Help

[RabidCicada]

Heyo everyone.  I'm trying to debug hosted-engine --deploy.  It is failing
in `Copy configuration archive to storage` in `create_target_vm.yml`
from `hosted-engine
--deploy`.  My general and most important query here is how to get good
debug output from ansible through hosted-engine.  I'm running hosted-engine
through an ssh session.

I can't figure out how to get good debug output from ansible within that
workflow.  I see it's running through otopi, I tried setting typical
`debugger: on_failed` hooks etc and tried many incantations on the command line
and config files to get ansible to help me out.  The debugger: directive
and other debugger related ansible config file stuff wouldn't result in any
debugger popping up.  I also can't seem to pass normal - flags to
hosted-engine either and get it to ansible.  Ultimately I tried to use a
`pause` directive and it complained that it was in a non-interactive
shell.  I figured it might be the result of my ssh session so I enabled tty
allocation with -t -t.  It did not resolve the issue.

I eventually wrote-my-own/stole a callback_plugin that checks an
environmental variable and enables `display.verbosity = int(v)` since I
can't seem to pass typical - stuff to ansible through `hosted-engine
--deploy`.  It give me the best info that I have so far.   But it wont give
me enough to debug issues around Gathering Facts or what looks like a
sudo/permission problem in `Copy configuration archive to storage` in
`create_target_vm.yml`.  I took and used the exact command that they use
manually and it works when I run it manually (But I can't get debug output
to show me the exact sudo command being executed), hence my interest in
passing - or equivalent to ansible through `hosted-engine`.  I
intentionally disabled the VM_directory cleanup so that I could execute the
same stuff.

Soafter all that...what is a good way to get deep debug info from
hosted-engine ansible stuff?

Or does anyone have intuition for the possible sudo problem?
~Kyle
___
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users