[Rdo-list] [Mitaka] Automatic restart for services

Matthew Mosesohn mmosesohn at mirantis.com
Thu Nov 12 16:13:55 UTC 2015


Typically if you manage services with pacemaker, you would put the affected
service into unmanaged state before you run `yum update`, so that RPM can
do its thing. Afterwards, just mark the services managed.

On Thu, Nov 12, 2015 at 7:06 PM, Ihar Hrachyshka <ihrachys at redhat.com>
wrote:

> Fabio M. Di Nitto <fdinitto at redhat.com> wrote:
>
>
>>
>> On 11/12/2015 5:01 PM, Ihar Hrachyshka wrote:
>>
>>> Fabio M. Di Nitto <fdinitto at redhat.com> wrote:
>>>
>>> On 11/12/2015 4:55 PM, Javier Pena wrote:
>>>>
>>>>> ----- Original Message -----
>>>>>
>>>>>> On 11/12/2015 05:42 AM, Haïkel wrote:
>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> this has been discussed for a while, so I'd like to set automatic
>>>>>>> restart for OpenStack services as a goal for Mitaka - milestone 1.
>>>>>>> These are easy fixes, and by setting it to milestone 1, we'll have
>>>>>>> more than enough time to test this and optimize.
>>>>>>>
>>>>>>> I already created a trello card to track that effort, but I'd like to
>>>>>>> get your feedback first.
>>>>>>> https://trello.com/c/HfXMLSTD/106-set-automatic-restart-for-service
>>>>>>>
>>>>>>
>>>>>> can you describe what you mean here by automatic restart for services?
>>>>>>
>>>>>> I'm worried what the implications would be for things that are running
>>>>>> under Pacemaker control in an HA environment, so I think we need more
>>>>>> details here
>>>>>>
>>>>>
>>>>> FWIW, this option is already enabled in some services, such as
>>>>> nova-api
>>>>> (
>>>>> https://github.com/openstack-packages/nova/blob/rpm-master/openstack-nova-api.service
>>>>> )
>>>>> where Restart=always. I've checked a Juno deployment, and that option
>>>>> was already set back then.
>>>>>
>>>>
>>>> It shouldn´t affect pacemaker deployment, because we do override some of
>>>> those values via dbus config.
>>>>
>>>
>>> To clarify things, doesn’t it affect just nova-api because you have some
>>> special treatment for the service? or it would not affect you if applied
>>> to other units? [In that case we could go forward and set automatic
>>> restart policy there, while you could stick to your hooks to influence
>>> service restart behaviour.]
>>>
>>
>> all services in pacemaker that are managed/monitored via systemd gets a
>> set of overrides to avoid those kind of issues.
>>
>
> Then we should be safe to proceed with the feature.
>
> Thanks,
>
> Ihar
>
> _______________________________________________
> Rdo-list mailing list
> Rdo-list at redhat.com
> https://www.redhat.com/mailman/listinfo/rdo-list
>
> To unsubscribe: rdo-list-unsubscribe at redhat.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rdoproject.org/pipermail/dev/attachments/20151112/ee788210/attachment.html>


More information about the dev mailing list