Open Stack

++

On 05/13/2015 11:57 AM, Swann Croiset wrote:
> Hi Fuelers,
>
> It would be valuable to configure the healthcheck middleware [0] for all
> services deployed by fuel, available since Kilo.
>
> Several (obvious) benefits:
> - Provide a common API for healthcheck across OpenStack services
> - HAproxy  performs more accurate HTTP checks for its backend status
> - Operators can disable a service (maintenance mode, test, ..) by
> creating a simple file w/o stopping the service.
> - monitoring systems can leverage this API to provide insight
> information of service status with a lighten checks w/o authentication
> (in opposition to heavily check for specific resources) <-- here my use
> case
>
> Does it sound reasonable to target this for 7.0? any clue?  drawback?
> I can handle the writing of the spec to bootstrap this effort and maybe
> more.
>
> Implementation details to estimate the workload:
> * Configure WSGI pipelines for each project "api-paste.ini"
> * For security reason as mentioned in the original spec [1], the
> 'healthcheck' URI mustn't be open to the WWW:
>     * restricted access to local/management network or by auth
>     * or random/configurable URI at deployment time
> * HAproxy "httpchk" option will be "GET /<healthcheck path>" instead of
> the default "OPTION /"
>
> Parallel (future) work on oslo or whatever:
> * extend with specific checks (DB unavailable, RabbitMQ stuck, ..) to
> provide a detailed status of subsystem dependencies.
>
> [0]
> http://docs.openstack.org/developer/oslo.middleware/healthcheck_plugins.html
> [1]
> http://git.openstack.org/cgit/openstack/oslo-specs/plain/specs/kilo/oslo-middleware-healthcheck.rst
>
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>