cloudstack-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chip Childers <chip.child...@sungard.com>
Subject Re: Multiple IP's to one instance
Date Wed, 28 Nov 2012 17:31:34 GMT
On Wed, Nov 28, 2012 at 12:23 PM, Hari Kannan <hari.kannan@citrix.com> wrote:
> Thanks, Chip - that helps!!! Another follow-up question, though..
>
> Lets outline the process:
>
> - Someone interested in implementing feature X proposes on the dev list - lets say it
has the following info: background on why (s)he thinks it is important (*what*/ *why*) as
well as advertise a design page on the wiki and / or a functional spec that elaborates on
the discussed feature idea (*how*) -
> - This would generate discussion on a) how sound the arguments for feature introductions
are etc. (i.e. discuss on *what* / *why*) b) how sound the design is or provide alternate
suggestions/enhancements etc.
> - upon consensus of the validity of both the *what/why* as well as *how*, developer goes
ahead with the implementation and presents for code review
> - upon approval of code (perhaps after a few iterations), a committer checks the code
in
>
> Does that sound like a decent outline?

That's about right...  of course variation is OK too.  The main points
are (a) that we do our development in a collaborative and open way and
(b) that consensus gathering is critical to producing the best result.

> If so, in this example case (multiple IPs), we are more or less at the *what/why* stage
- although I realize the discussion needs to happen on the dev list - so, using this as an
example case, what must a developer do now? Does (s)he still go ahead and make design specs
as it is not clear if we have a consensus one way or the other on the *what/why* stage...

At this point, I would suggest summarizing a few of the feature
options that may solve the use case.  For example, there are
discussions about "reserving IP's for manual assignment within
instances", "figure out if DHCP can do this correctly", "use existing
LB and Elastic IP capabilities to achieve a similar result", etc...

>
> Regards,
>
> Hari Kannan
>
> -----Original Message-----
> From: Chip Childers [mailto:chip.childers@sungard.com]
> Sent: Wednesday, November 28, 2012 9:02 AM
> To: <cloudstack-users@incubator.apache.org>
> Subject: Re: Multiple IP's to one instance
>
> Great questions!
>
> So first, actual development decisions need to happen on the cloudstack-dev list, not
users (although we love seeing user discussions about feature requests!).
>
> The way it works on the dev list, is if you (or anyone) wants to propose building a feature,
you would email the specifics of what you want to add.  Generally, changes are agreed to by
gathering consensus on both *what* and *how*.  The developer that is building the feature
would then (hopefully) do a design page on the wiki and / or a functional spec that elaborates
on the discussed feature idea.
> Getting consensus on that deign / functional spec then means moving forward to implementation.
>
> The dev list is also the place to get help with figuring out how to add the feature (both
logistically and technically).
>
> Does that help?
>
> -chip
>
> On Wed, Nov 28, 2012 at 11:51 AM, Hari Kannan <hari.kannan@citrix.com> wrote:
>> As a newbie to Apache, I have a dumb procedural question -
>>
>> I see there are folks who are in favor of implementing this as well as folks who
either believe there are workarounds and/or believe there is no NEED for this. However, if
someone really wants to implement this, can/must anyone object? I can imagine if the implementation
is bad, we can reject it, but are there any other reasons to NOT implement a feature?
>>
>> PS1: While this feature is an example, I want to under the general
>> process/guideline for future reference
>>
>> Hari Kannan
>>
>> -----Original Message-----
>> From: Boylan, James [mailto:JAMES.BOYLAN@orbitz.com]
>> Sent: Tuesday, November 27, 2012 3:18 PM
>> To: cloudstack-users@incubator.apache.org
>> Subject: RE: Multiple IP's to one instance
>>
>> I agree with Tamas. Cloudstack has functionality that would allow implementations
using NAT/LoadBalancer configurations that would not require multiple IPs. Plus the NetScaler
Service, if it is a true Netscaler implementation, supports loading the certs into it thus
removing the multiple IP address requirement.
>>
>> -- James
>> ________________________________________
>> From: Tamas Monos [tamasm@veber.co.uk]
>> Sent: Tuesday, November 27, 2012 4:20 AM
>> To: cloudstack-users@incubator.apache.org
>> Subject: RE: Multiple IP's to one instance
>>
>> Hi,
>>
>> Also I have tested earlier 3.0.2 and if you create multiple networks for the same
account then the VM automatically gets that many IPs and NICs as many network the VM has.
>> Then you can use static nat if you need to support multiple websites.
>>
>> Or another solution to this is to have many static nat IPs on the same network and
port forward each IPs port 80 and 443 onto a different port on the VM and configure apache
to listen on different ports for different virtual hosts.
>> There is always a workaround.
>>
>> Regards
>>
>> Tamas Monos                                               DDI         +44(0)2034687012
>> Chief Technical                                             Office    +44(0)2034687000
>> Veber: The Hosting Specialists               Fax         +44(0)871 522 7057
>> http://www.veber.co.uk
>>
>> Follow us on Twitter: www.twitter.com/veberhost Follow us on Facebook:
>> www.facebook.com/veberhost
>>
>> -----Original Message-----
>> From: Kevin Kluge [mailto:Kevin.Kluge@citrix.com]
>> Sent: 26 November 2012 19:03
>> To: cloudstack-users@incubator.apache.org
>> Subject: RE: Multiple IP's to one instance
>>
>>>
>>> I am conflicted on this - I know the limitations with adding additional interfaces.
>>> I'd almost argue for requiring number of interfaces/IPs to be set at
>>> machine instantiation and still handing out addresses to all
>>> interfaces via DHCP, but not all OSes will handle that cleanly.
>>
>> As you probably know CS hands out IPs via DHCP to multiple NICs today, and it is
challenging to work reliably since different guest OS's have different behavior with respect
to setting/re-setting the default route in DHCP replies.  Today CloudStack has some knowledge
of guest OS to make this work but we've seen bugs with that where the list of guest OS behavior
is not right.  Those are fixable.  But there are other cases (e.g. boot from ISO, a user changes
the DHCP client post-create, or the user upgrades the OS which changes the DHCP client) where
I don't think there is a reasonable fix.  So I think we need at a minimum an option to disable
DHCP reply on NICs that are not the default route.  We should even consider a behavior change
to make that the only option.
>>
>> -kevin
>>
>

Mime
View raw message