No change in the last two weeks, just sent code for review.

https://reviews.apache.org/r/24611/


Regards,

--
Daniel Simões
Time Evolução Infra - Globo.com
E-mail: daniel.sim...@corp.globo.com
Tel.: +55 21 2483-6977


2014-07-30 11:04 GMT-03:00 Daniel Vega Simões <daniel.sim...@corp.globo.com>
:

> Hi guys,
>
> Since nobody was opposed to the approach suggested in the last e-mail, we
> went on and implemented the global option that authorizes the plugin to
> override record entries when they already exist in the DNS server.
>
> We also updated the name to GloboDNS and the functional spec to reflect
> both changes. You can take a look at the table that summarizes the actions
> for the GloboDNS plugin on the wiki
> <https://cwiki.apache.org/confluence/display/CLOUDSTACK/Bind+and+PowerDNS+integration+by+Globo+DNSAPI>.
> Unit tests were implemented to make sure all of these cases behave as
> expected.
>
> Please let me know how we can move forward to integrate this in the
> Cloudstack repository.
>
>
> Cheers,
>
>  --
> Daniel Simões
> Time Evolução Infra - Globo.com
> E-mail: daniel.sim...@corp.globo.com
> Tel.: +55 21 2483-6977
>
>
> 2014-07-14 16:41 GMT-03:00 Daniel Vega Simões <
> daniel.sim...@corp.globo.com>:
>
> Hi guys,
>>
>> @David
>> As Silvano mentioned earlier, our cloud-related DNS have unique names
>> (based on network info), so conflicts do not happen. As such, these domains
>> are exclusive to our cloud infrastructure and we let Cloudstack have full
>> authority over them. For that reason, whenever a network is deleted, the
>> domain associated to it is removed, along with every record in it.
>>
>>
>> @Chiradeep
>> If a domain already exists when a network is created, the plugin takes
>> over that domain, creating and removing records as needed. It does not
>> handle any records created externally. The only exception is when the
>> domain is removed, as explained just above.
>>
>> Secondary IPs are currently not handled by the plugin, since we
>> understand that once that IP is reserved by Cloudstack, routing and name
>> resolution should be done manually by the operator.
>>
>> As for the name, GloboDNS sounds about right :) We'll make the
>> appropriate changes to avoid confusion.
>>
>>
>>
>> This is what makes sense for us in this first release. Implementing every
>> option available to the operator might make the plugin configuration very
>> confusing.
>>
>> A simpler approach would be to create one single global/zone option that
>> allows Cloudstack to have full authority over domains/records or not. In
>> the case where Cloudstack does not have authority, an error could be thrown
>> if the domain/record already exists and domains are never removed by the
>> plugin itself, so an operator would need to do it manually (or by means of
>> another tool).
>>
>>
>> What do you think?
>>
>>
>>  --
>> Daniel Simões
>> Time Evolução Infra - Globo.com
>> E-mail: daniel.sim...@corp.globo.com
>> Tel.: +55 21 2483-6977
>>
>>
>> 2014-07-14 14:20 GMT-03:00 Chiradeep Vittal <chiradeep.vit...@citrix.com>
>> :
>>
>> I think the question is relevant to network creation as well. If I
>>> provide a domain that already exists, what is the result?
>>>
>>> A couple of other comments:
>>>  - are we going to handle the case of secondary IP addresses?
>>>  - DNSAPI sounds generic, but it actually refers to one specific API
>>> architected by Globo. To avoid confusion, would it make sense to rename it
>>> GloboDNSAPI? Alternately, give the DNSAPI project a less generic name
>>> (e.g., vincular)
>>>
>>> From: David Nalley <da...@gnsa.us<mailto:da...@gnsa.us>>
>>> Reply-To: "dev@cloudstack.apache.org<mailto:dev@cloudstack.apache.org>"
>>> <dev@cloudstack.apache.org<mailto:dev@cloudstack.apache.org>>
>>> Date: Friday, July 11, 2014 at 10:06 AM
>>> To: "dev@cloudstack.apache.org<mailto:dev@cloudstack.apache.org>" <
>>> dev@cloudstack.apache.org<mailto:dev@cloudstack.apache.org>>
>>> Subject: Re: [DISCUSS] [PROPOSAL] Implementation of DNS Provider for
>>> Bind (for 4.5)
>>>
>>> I tend to agree with Erik, flexibility solves the problem for more
>>> people, while solution 1 is likely the easiest to implement. I am not
>>> sure that it makes sense for most people though - and would really
>>> only work for greenfield deployments or clouds that had all of the DNS
>>> entries relating to instances in the cloud.
>>>
>>> First question; how are you intending on using it? Which of those
>>> solutions works for you?
>>>
>>> --David
>>>
>>>
>>>
>>> On Thu, Jul 3, 2014 at 3:49 PM, Erik Weber <terbol...@gmail.com<mailto:
>>> terbol...@gmail.com>> wrote:
>>> To push that choice over to the operator you could add it as a
>>> global/zone/network option.
>>>
>>> As an operator i would prefer to have my own logic to handle cleanup, but
>>> this varies for everyone hence the option :-)
>>>
>>> Erik
>>> 3. juli 2014 21:45 skrev "Silvano Nogueira Buback" <
>>> silv...@corp.globo.com<mailto:silv...@corp.globo.com>>
>>> følgende:
>>>
>>> Hi guys,
>>>
>>>      I think you are busy because 4.4 release tasks, but I'm worried
>>> about
>>> the time to 4.5 feature freeze. I put the documentation of feature in
>>> wiki
>>> as requested and I hoped people read there and make some comments here.
>>>
>>> To help, I will put design issues that are in document, one by one, and
>>> we
>>> can discuss in this thread. After each discussion I will change the
>>> document.
>>>
>>>     I have one question about removing DNS domain when network has been
>>> deleted. In my current implementation I remove DNS domain when network is
>>> removed. But if the DNS domain is shared with another network or maybe
>>> is a
>>> dns domain used outside ACS this can be a problem. What I can do with DNS
>>> domain when network is removed:
>>>
>>>     1. Keep the current implementation. Always deleted DNS domain when
>>>     network is removed (works well if the ACS is the only manager for the
>>> DNS
>>>     (one network domain per network).
>>>     2. Remove DNS domain only if the domain was created by ACS. This can
>>> be
>>>     a problem if someone put records after ACS creation.
>>>     3. Remove DNS domain only if there is no more records there. Maybe
>>> DNS
>>>     domain can stay forever there because an inconsistency that keep only
>>> one
>>>     record.
>>>
>>>
>>> Which one is the best?
>>>
>>> []'s,
>>>
>>> Silvano Buback
>>>
>>>
>>>
>>> On Thu, Jun 26, 2014 at 11:34 AM, Silvano Nogueira Buback <
>>> silv...@corp.globo.com<mailto:silv...@corp.globo.com>> wrote:
>>>
>>> > Thank you David.
>>> >
>>> > I put design documents on wiki:
>>> >
>>>
>>> https://cwiki.apache.org/confluence/display/CLOUDSTACK/Bind+and+PowerDNS+integration+by+Globo+DNSAPI
>>> .
>>> > I create an issue
>>> https://issues.apache.org/jira/browse/CLOUDSTACK-6998
>>> > too.
>>> >
>>> > I look forward to hearing your feedbacks.
>>> >
>>> > []'s,
>>> >
>>> > Silvano Buback
>>> >
>>> >
>>> > On Wed, Jun 25, 2014 at 5:50 PM, David Nalley <da...@gnsa.us<mailto:
>>> da...@gnsa.us>> wrote:
>>> >
>>> >> On Wed, Jun 25, 2014 at 4:38 PM, Silvano Nogueira Buback
>>> >> <silv...@corp.globo.com<mailto:silv...@corp.globo.com>> wrote:
>>> >> > Hi guys,
>>> >> >
>>> >> >    I finish the first version of design document:
>>> >> >
>>> >>
>>>
>>> https://docs.google.com/document/d/1kbPQJrBC87ZtR-t7LwHFDzAmT436ShtjwKE84FVfByM/pub
>>> >> > .
>>> >> >
>>> >> >    Someone could give me access to put design documents in wiki?
>>> Bellow
>>> >> the
>>> >> > username of people work with Cloudstack in Globo.com and need
>>> access.
>>> >> >
>>> >> > snbuback silv...@corp.globo.com<mailto:silv...@corp.globo.com>
>>> >> > daniel.simoes daniel.sim...@corp.globo.com<mailto:
>>> daniel.sim...@corp.globo.com>
>>> >> > lokama - lok...@gmail.com<mailto:lok...@gmail.com>
>>> >> >
>>> >> > Regards,
>>> >> >
>>> >> > Silvano Buback
>>> >> >
>>> >> >
>>> >> >
>>> >> > On Thu, Jun 19, 2014 at 11:29 AM, Silvano Buback <
>>> snbub...@gmail.com<mailto:snbub...@gmail.com>>
>>> >> wrote:
>>> >> >
>>> >> >> Of course, I forgotten my account info:
>>> >> >> snbuback / silv...@corp.globo.com<mailto:silv...@corp.globo.com>
>>> >> >>
>>> >>
>>> >>
>>> >> Done.
>>> >>
>>> >> --David
>>> >>
>>> >
>>> >
>>>
>>>
>>>
>>
>

Reply via email to