fuel-dev team mailing list archive

[Сергей Юдин <tsipa740@xxxxxxxxx>]

Max, i've checked right now - it is possible to define pool of replicas like
mongodb://ceilometer:PASSWD@172.16.0.161:27017,172.16.0.162:27017,
172.16.0.163:27017/ceilometer?replicaset=ceilometer
you just need to fix the template a bit


2014-04-02 11:07 GMT+04:00 Maksim Mazur <mmaxur@xxxxxxxxxxxx>:

> Hi!
>
>
> It looks like ceilometer can use replicaset instead of primary node.
> I will test it today
>
> Best Regards,
> Max
>
>
>
> On Wed, Apr 2, 2014 at 7:54 AM, Andrew Woodward <xarses@xxxxxxxxx> wrote:
>
>> blueprint appears to be a full clone of a template and not modified.
>>
>> Based on the last review I had of the Mongo changes; the major design
>> issues I have with this are:
>>
>> 1) that we are requiring MongoDB (which is AGPL) and loosing the ability
>> to have Ceilometer on MySQL which will be an issue for some people. If this
>> is expected, then it needs to be documented.
>> 2) there was no failover connection mechanism between primary-mongo and
>> mongo slaves (idk if "replicast connection" takes care of this or not)
>>
>> Of minor issue: I'd like to see settings like
>> mongo_node_address, current_ceilometer_db_address should be calculated by
>> nailgun and passed using fuel_settings. If possible we should avoid adding
>> business logic in the osnailyfacter classes as it makes it harder to make
>> these things more modular in the future.
>>
>>
>>
>> On Tue, Apr 1, 2014 at 5:58 AM, Vladimir Kuklin <vkuklin@xxxxxxxxxxxx>wrote:
>>
>>> Guys
>>>
>>> We had a MongoDB integration meeting today and that;s what we discussed
>>> ordered by priority.
>>>
>>> Overall list of things to be done:
>>>
>>> 1) High: Finish work on puppet manifests:
>>>              replicaset connection
>>>              fix admin user rights
>>> 2) High: Check nailgun disk allocation in case of role combination
>>> 3) High: Use the following scheme for disk allocation
>>>              floor(n-1/2) for combined roles
>>>              n-1 for mongo-only node
>>> 4) Medium: use striped LVM/mdraid for mongo nodes - (the same for ceph)
>>> 5) High: UI:
>>>              role-setting dependencies -> rebase mongo UI role request
>>>              1-disk mongo node should raise caution of non-production
>>> scenario
>>> 6) Medium: perform I/O tuning for filesystem (the same for ceph?):
>>>     a) deadline scheduler
>>>     b) common high-performance mount options (noatime, barriers=0,
>>> commit=600, journal=writeback)
>>> 7) Low: SSD-disks are preferrable
>>>
>>> These changes are to be reflected in the blueprint<https://blueprints.launchpad.net/fuel/+spec/mongodb-fuel-integration>drafted by Max Mazur.
>>>
>>> Please, provide your input or feedback if you have any.
>>>
>>>
>>> --
>>> Yours Faithfully,
>>> Vladimir Kuklin,
>>> Fuel Library Tech Lead,
>>> Mirantis, Inc.
>>> +7 (495) 640-49-04
>>> +7 (926) 702-39-68
>>> Skype kuklinvv
>>> 45bk3, Vorontsovskaya Str.
>>> Moscow, Russia,
>>> www.mirantis.com <http://www.mirantis.ru/>
>>> www.mirantis.ru
>>> vkuklin@xxxxxxxxxxxx
>>>
>>> --
>>> Mailing list: https://launchpad.net/~fuel-dev
>>> Post to     : fuel-dev@xxxxxxxxxxxxxxxxxxx
>>> Unsubscribe : https://launchpad.net/~fuel-dev
>>> More help   : https://help.launchpad.net/ListHelp
>>>
>>>
>>
>>
>> --
>> Andrew
>> Mirantis
>> Ceph community
>>
>
>
> --
> Mailing list: https://launchpad.net/~fuel-dev
> Post to     : fuel-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~fuel-dev
> More help   : https://help.launchpad.net/ListHelp
>
>