[ovirt-users] New host HE 4.0 -> [ ERROR ] Failed to execute stage 'Environment customization': Specified gateway is not pingable

Sun Jul 24 09:19:46 UTC 2016

Hi,

It was indeed NetApp, and as you say I can remember something like it
but I'm not 200% sure.

For some host that I wasn't able to remove I needed hosted-engine
--reinitialize-lockspace indeed, figured that out.

dd over the same went well, didn't see anyone complaining about that either.

Thanks Didi for upodating this post.

2016-07-24 8:12 GMT+02:00 Yedidyah Bar David <didi at redhat.com>:
> On Thu, Jul 21, 2016 at 9:56 PM, Matt . <yamakasi.014 at gmail.com> wrote:
>> OK, it seems that a snapshot was having the files but linux didn't saw
>> a change between the images as the size was still the same.
>
> What kind of snapshot?
>
> AFAIR from old days with netapp, snapshots on netapp expose the same
> inode. diff will see this and will not further look at the file, even if
> everything else is different - size, timestamps etc. I guess it's still
> the same today.
>
> To answer your question, you might be looking for 'hosted-engine
> --reinitialize-lockspace'. But if you ask me, if the content was zeroed,
> you have some serious issues there - either on the storage or on the
> hosts.
>
>>
>> I'm going to dd back :)
>
> Good luck.
>
> Note that dd and/or the shell might too refuse to copy over the "same"
> file (same inode).
>
> Best,
>
>>
>> 2016-07-21 20:04 GMT+02:00 Matt . <yamakasi.014 at gmail.com>:
>>> OK, it seems that my config volume for the hosted engine is empty, I
>>> cannot tar anything out of it and I'm quite sure it has been empty for
>>> ever as I got some strange warning every time.
>>>
>>> Is there a way to recover that from files ? Backups/snapshots are empty so...
>>>
>>> 2016-07-21 13:43 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>> On Thu, Jul 21, 2016 at 12:38 PM, Matt . <yamakasi.014 at gmail.com> wrote:
>>>>> The deploy went OK from the GUI but it's still not "active" as Hosted
>>>>> engine Host. The install also wen very fast so it looked like it
>>>>> didn't do anything witl the deploy setting in the GUI @ the hosted
>>>>> engine tab.
>>>>
>>>> The difference with a standard host is just that it should create
>>>> /etc/ovirt-hosted-engine/hosted-engine.conf and start ovirt-ha-agent
>>>> and ovirt-ha-broker so it shouldn't be that longer.
>>>>
>>>> At the end, due to the issue with syncing the local maintenance mode
>>>> from the engine, you have to connect to the host and run
>>>> 'hosted-engine -set-maintenance --mode=none' from there.
>>>>
>>>> Can you please post the result of
>>>> hosted-engine --vm-status and the status of ovirt-ha-agent?
>>>>
>>>>> Or should I re-deploy again on the commandline after adding the host
>>>>> through the GUI ?
>>>>>
>>>>> About the initial GW, I didn't see that in the GUI.
>>>>>
>>>>> 2016-07-21 12:31 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>>>> On Thu, Jul 21, 2016 at 12:28 PM, Matt . <yamakasi.014 at gmail.com> wrote:
>>>>>>> No problem I checked it out and applied it, no solution it seems.
>>>>>>
>>>>>> Do you mean that it's still referring to your initial gateway or that
>>>>>> the deploy failed?
>>>>>>
>>>>>>> 2016-07-21 12:14 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>>>>>> On Thu, Jul 21, 2016 at 12:09 PM, Matt . <yamakasi.014 at gmail.com> wrote:
>>>>>>>>> OK, I did a deploy from the engine and the Hosted is "Not Active" and
>>>>>>>>
>>>>>>>> Yes, sorry, I forgot to mention that.
>>>>>>>> We have an open bug about re-activating an hosted-engine host from the
>>>>>>>> engine: https://bugzilla.redhat.com/show_bug.cgi?id=1353600
>>>>>>>> Connecting to the host and exiting the maintenance mode from there is
>>>>>>>> the temporary solution.
>>>>>>>>
>>>>>>>>> Engine HA is empty on 4.0.1
>>>>>>>>>
>>>>>>>>> 2016-07-21 11:58 GMT+02:00 Matt . <yamakasi.014 at gmail.com>:
>>>>>>>>>> True, but you can up an IP on any device in that case because you know
>>>>>>>>>> what you are doing.
>>>>>>>>>>
>>>>>>>>>> I think that this can be finetuned, or at least multiple IPs would be
>>>>>>>>>> nice as I now need to add one IP from my vrrp setup, so if there is a
>>>>>>>>>> failover.... or I need to take down the second switch I'm also done.
>>>>>>>>>>
>>>>>>>>>> 2016-07-21 10:23 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>>>>>>>>> On Thu, Jul 21, 2016 at 10:03 AM, Matt . <yamakasi.014 at gmail.com> wrote:
>>>>>>>>>>>> OK, I will check this laster on, as I was having issues with the
>>>>>>>>>>>> deploy from the GUI before as nothing really happened, I thought it
>>>>>>>>>>>> was a bug on the first 4.0 release ?
>>>>>>>>>>>>
>>>>>>>>>>>> I think this should be investigated, the whole deployment as it would
>>>>>>>>>>>> be good to have the possibility to list some GW's in case you use
>>>>>>>>>>>> failover.
>>>>>>>>>>>>
>>>>>>>>>>>> I was even thinking of setting the engine IP there, should be working
>>>>>>>>>>>> as good as also.
>>>>>>>>>>>
>>>>>>>>>>> Not really since the agent is pinging also when the engine VM is down
>>>>>>>>>>> due to maintenance.
>>>>>>>>>>>
>>>>>>>>>>>> 2016-07-21 9:53 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>>>>>>>>>>> On Thu, Jul 21, 2016 at 9:47 AM, Matt . <yamakasi.014 at gmail.com> wrote:
>>>>>>>>>>>>>> OK, I already thought that but where is this saved as I don't have any
>>>>>>>>>>>>>> rights as root to the ha_agent folder on the HE storage ?
>>>>>>>>>>>>>
>>>>>>>>>>>>> It's not a folder: it's a tar file directly saved over a VDSM volume.
>>>>>>>>>>>>> Editing it's a bit tricky.
>>>>>>>>>>>>>
>>>>>>>>>>>>>> In worst case scenario, how would i be able to deploy a new hosted
>>>>>>>>>>>>>> engine from a hosted engine ? This sounds undoable to me and also
>>>>>>>>>>>>>> unwanted as the HE is OK so far.
>>>>>>>>>>>>>
>>>>>>>>>>>>> With a 4.0 engine, when you choose to deploy a new host from the
>>>>>>>>>>>>> webadmin or from the rest API you can also choose to deploy it as an
>>>>>>>>>>>>> hosted-engine host if the engine has been deployed with hosted-engine.
>>>>>>>>>>>>> In this case, if I'm not wrong, you can also specify a different gateway value.
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>> 2016-07-21 9:41 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>>>>>>>>>>>>> On Thu, Jul 21, 2016 at 9:27 AM, Matt . <yamakasi.014 at gmail.com> wrote:
>>>>>>>>>>>>>>>> Even when I change that on each host before I do a new deploy it still
>>>>>>>>>>>>>>>> uses the old GW address.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> If you deploy additional hosts with the CLI utility, each new host
>>>>>>>>>>>>>>> will be deployed consuming the answerfile generated by the first host
>>>>>>>>>>>>>>> and saved on the shared storage.
>>>>>>>>>>>>>>> You have to manually tweak there if you to avoid.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Another smarted option is to deploy additional HE host from the
>>>>>>>>>>>>>>> webadmin interface.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> 2016-07-21 9:24 GMT+02:00 Simone Tiraboschi <stirabos at redhat.com>:
>>>>>>>>>>>>>>>>> On Wed, Jul 20, 2016 at 11:57 PM, Matt . <yamakasi.014 at gmail.com> wrote:
>>>>>>>>>>>>>>>>>> OK, I tried to edit the answerfiles on the other hosts to use another
>>>>>>>>>>>>>>>>>> IP but it doesn't ping and still uses the old GW address.
>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> Any idea ?
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> The gateway address is saved under
>>>>>>>>>>>>>>>>> /etc/ovirt-hosted-engine/hosted-engine.conf
>>>>>>>>>>>>>>>>> on each host.
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> 2016-07-20 22:27 GMT+02:00 Matt . <yamakasi.014 at gmail.com>:
>>>>>>>>>>>>>>>>>>> Maybe it's a good idea to be able to add multiple addresses it should
>>>>>>>>>>>>>>>>>>> ping in order of response.
>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>> 2016-07-20 22:13 GMT+02:00 Sandro Bonazzola <sbonazzo at redhat.com>:
>>>>>>>>>>>>>>>>>>>> Il 20/Lug/2016 10:06 PM, "Matt ." <yamakasi.014 at gmail.com> ha scritto:
>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>> Hi,
>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>> I'm trying to add a host to a HE installation but my gateway is not
>>>>>>>>>>>>>>>>>>>>> pingable. It's a floating IP which is usable but not pingable in this
>>>>>>>>>>>>>>>>>>>>> case.
>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>> During the first install of the HE I could give any IP to ping and I
>>>>>>>>>>>>>>>>>>>>> used a source IP for the floating GW address.
>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>> What can I do to get around this as the network is OK, even when the
>>>>>>>>>>>>>>>>>>>>> GW is not pingable.
>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>> I think that gw not pingable will be interpreted as network failure by the
>>>>>>>>>>>>>>>>>>>> agent and will lower the score of the hosts.
>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>> Any idea is welcome!
>>>>>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>>>>> Thanks!
>>>>>>>>>>>>>>>>>>>>> _______________________________________________
>>>>>>>>>>>>>>>>>>>>> Users mailing list
>>>>>>>>>>>>>>>>>>>>> Users at ovirt.org
>>>>>>>>>>>>>>>>>>>>> http://lists.ovirt.org/mailman/listinfo/users
>> _______________________________________________
>> Users mailing list
>> Users at ovirt.org
>> http://lists.ovirt.org/mailman/listinfo/users
>
>
>
> --
> Didi