[Users] Issues starting hosted engine VM
Andrew Lau
andrew at andrewklau.com
Mon Jan 20 11:19:02 UTC 2014
Hi,
That bug seems to be private :(
I'm interested also to hear about this feature, as with 3.3.2 I had my
gluster vms go into paused state quite a few times and they actually
couldn't be resumed at all, they needed to be forced off and back on.
On Mon, Jan 20, 2014 at 10:13 PM, Dafna Ron <dron at redhat.com> wrote:
> interesting... :) so this is now configurable...
> what happens if qemu fails to start the vm (this happens sometimes -
> mostly on file type storage). do we have a re-try or a specific error
> telling the use that the activation failed and manual intervention is
> required?
>
>
>
> On 01/20/2014 11:02 AM, Leonid Natapov wrote:
>
>> All vms. Check this PRD: https://bugzilla.redhat.com/
>> show_bug.cgi?id=723055
>>
>>
>> ----- Original Message -----
>> From: "Dafna Ron" <dron at redhat.com>
>> To: "Leonid Natapov" <lnatapov at redhat.com>
>> Cc: "Andrew Lau" <andrew at andrewklau.com>, "Yedidyah Bar David" <
>> didi at redhat.com>, "users" <users at ovirt.org>
>> Sent: Monday, January 20, 2014 12:44:46 PM
>> Subject: Re: [Users] Issues starting hosted engine VM
>>
>> On 01/20/2014 10:38 AM, Leonid Natapov wrote:
>>
>>> 1.hosted-engine --vm-start should start engine vm. There was no problem
>>> with it when I tested it.
>>> 2.hosted-engine --vm-start-paused was added for the case when something
>>> is wrong with engine vm and it can't start and requires user intervention.
>>> For example in case of kernel panic.
>>> User can start engine vm in paused mode ,connect to it and try to fix
>>> the problem by booting in single user mode ,etc.
>>> 3.When the connectivity to shared storage is lost engine vm becomes
>>> paused. VM should be automatically unpaused after connectivity resumes (we
>>> introduced this feature in 3.3) but in case of NFS it could take quite
>>> time.so may be we should add something like --vm-resume in order to resume
>>>
>> Are we talking only on the hosted engine vm or all other vm's? if I have
>> other vm's they will also stop, will they be auto started as well?
>>
>>> the engine vm manually.
>>>
>>> Thanks,
>>> Leonid.
>>>
>>>
>>> ----- Original Message -----
>>> From: "Andrew Lau" <andrew at andrewklau.com>
>>> To: dron at redhat.com
>>> Cc: "Leonid Natapov" <lnatapov at redhat.com>, "Yedidyah Bar David" <
>>> didi at redhat.com>, "users" <users at ovirt.org>
>>> Sent: Monday, January 20, 2014 12:28:15 PM
>>> Subject: Re: [Users] Issues starting hosted engine VM
>>>
>>> It was paused due to the connection loss to the NFS server, I would
>>> assume
>>> once the connection is restored it could attempt to restore it? But I can
>>> try dig up the vdsm logs if you want, they would only be a few hours old
>>>
>>> I think having an option like --vm-resume would at least hide the reason
>>> of
>>> having to dig into virsh and messing with authentication at the very
>>> least.
>>>
>>> On Mon, Jan 20, 2014 at 9:23 PM, Dafna Ron <dron at redhat.com> wrote:
>>>
>>> the question is what was the vm paused on... this can be found in the
>>>> qemu
>>>> vm log.
>>>> if the vm is paused it will not be auto started - so I am not sure what
>>>> you expect to change? virsh requires authentication regardless to hosted
>>>> engine :)
>>>> Leonid, did you do any testing there?
>>>>
>>>>
>>>> On 01/20/2014 10:13 AM, Andrew Lau wrote:
>>>>
>>>> I have opened this BZ 1055461 anyway just in case
>>>>>
>>>>>
>>>>> On Mon, Jan 20, 2014 at 8:33 PM, Andrew Lau <andrew at andrewklau.com
>>>>> <mailto:
>>>>> andrew at andrewklau.com>> wrote:
>>>>>
>>>>> I was more interested in how the score process would be
>>>>> calculated, the vm-status option considered the VM in a bad
>>>>> state.
>>>>>
>>>>> I left it for a few minutes and nothing seemed to have changed, I
>>>>> think it relates to hosted engine as virsh requires
>>>>> authentication. Should I still open a bz?
>>>>>
>>>>> Cheers,
>>>>> Andrew.
>>>>>
>>>>> On Jan 20, 2014 7:48 PM, "Dafna Ron" <dron at redhat.com
>>>>> <mailto:dron at redhat.com>> wrote:
>>>>>
>>>>> I am not sure this is a hosted engine question as much as a
>>>>> qemu question.
>>>>> qemu-kvm will not support auto start of vm's after EIO
>>>>> because
>>>>> of remote possibility of corruption.
>>>>>
>>>>> On 01/20/2014 05:46 AM, Andrew Lau wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> Quick question, in the scenario eg. the NFS server
>>>>> becomes
>>>>> unreachable and the hosted-engine goes into a paused
>>>>> state. Will other hosts attempt to bring it back up?
>>>>> Should there be a command eg. hosted-engine --vm-resume
>>>>> ?
>>>>>
>>>>> When this happened, I manually forced it to resume using
>>>>> virsh
>>>>>
>>>>>
>>>>> On Sun, Jan 19, 2014 at 7:21 PM, Yedidyah Bar David
>>>>> <didi at redhat.com <mailto:didi at redhat.com>
>>>>> <mailto:didi at redhat.com <mailto:didi at redhat.com>>>
>>>>> wrote:
>>>>>
>>>>> Thanks a lot for your efforts and the report!
>>>>> -- Didi
>>>>>
>>>>> ------------------------------
>>>>> ------------------------------------------
>>>>>
>>>>> *From: *"Andrew Lau" <andrew at andrewklau.com
>>>>> <mailto:andrew at andrewklau.com>
>>>>> <mailto:andrew at andrewklau.com
>>>>>
>>>>> <mailto:andrew at andrewklau.com>>>
>>>>> *To: *"users" <users at ovirt.org
>>>>> <mailto:users at ovirt.org> <mailto:users at ovirt.org
>>>>>
>>>>> <mailto:users at ovirt.org>>>
>>>>> *Sent: *Saturday, January 18, 2014 3:20:22 PM
>>>>> *Subject: *Re: [Users] Issues starting hosted
>>>>> engine VM
>>>>>
>>>>>
>>>>> I believe I found the issue and have reported it
>>>>> here
>>>>> https://bugzilla.redhat.com/show_bug.cgi?id=1055059
>>>>>
>>>>> On Sat, Jan 18, 2014 at 11:33 PM, Andrew Lau
>>>>> <andrew at andrewklau.com
>>>>> <mailto:andrew at andrewklau.com>
>>>>> <mailto:andrew at andrewklau.com
>>>>>
>>>>> <mailto:andrew at andrewklau.com>>> wrote:
>>>>>
>>>>> The interesting thing - trying it with the
>>>>> paused option
>>>>> vdsm seems to create the VM
>>>>>
>>>>> hosted-engine --vm-start-paused
>>>>>
>>>>> vdsm.log http://www.fpaste.org/69604/
>>>>> 13900482/
>>>>>
>>>>> But I'm not sure how to then proceed to
>>>>> "resume" it.
>>>>>
>>>>> On Sat, Jan 18, 2014 at 10:23 PM, Andrew Lau
>>>>> <andrew at andrewklau.com
>>>>> <mailto:andrew at andrewklau.com>
>>>>> <mailto:andrew at andrewklau.com
>>>>>
>>>>> <mailto:andrew at andrewklau.com>>> wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> With the great help from sbonazzo, I
>>>>> managed to step
>>>>> past the initial bug with the
>>>>> hosted-engine-setup but
>>>>> appear to have run into another show
>>>>> stopper.
>>>>>
>>>>> I ran through the install process
>>>>> successfully up to
>>>>> the stage where it completed and the
>>>>> engine VM was to
>>>>> be shutdown. (The engine has already been
>>>>> installed on
>>>>> the VM and the host has been connected to
>>>>> the engine).
>>>>>
>>>>> The issue starts here that the host finds
>>>>> itself not
>>>>> able to start the VM up again.
>>>>>
>>>>> VDSM Logs:
>>>>> http://www.fpaste.org/69592/00427141/
>>>>> ovirt-hosted-engine-ha agent.log
>>>>> http://www.fpaste.org/69595/43609139/
>>>>>
>>>>> It seems to keep failing to start the
>>>>> VM..
>>>>> when I
>>>>> restart the agent I can see the score
>>>>> drop
>>>>> to 0 after
>>>>> 3 boot attempts. The interesting thing
>>>>> seems to be in
>>>>> the VDSM Logs "'Virtual machine does not
>>>>> exist',
>>>>> 'code': 1}}"
>>>>>
>>>>> I'm not sure where else to look.
>>>>> Suggestions?
>>>>>
>>>>> Cheers,
>>>>>
>>>>> Andrew
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> Users at ovirt.org <mailto:Users at ovirt.org>
>>>>> <mailto:Users at ovirt.org <mailto:Users at ovirt.org>>
>>>>>
>>>>> http://lists.ovirt.org/mailman/listinfo/users
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> -- Didi
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> Users at ovirt.org <mailto:Users at ovirt.org>
>>>>> http://lists.ovirt.org/mailman/listinfo/users
>>>>>
>>>>>
>>>>>
>>>>> -- Dafna Ron
>>>>>
>>>>>
>>>>>
>>>>> --
>>>> Dafna Ron
>>>>
>>>>
>>
>
> --
> Dafna Ron
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20140120/16632d00/attachment-0001.html>
More information about the Users
mailing list