I ran a test with a colleague this week connecting two cs vpcs. this works.

On Thu, May 22, 2014 at 6:05 AM, Sanjeev Neelarapu
<sanjeev.neelar...@citrix.com> wrote:
> In site-to-site vpn both sides need not to be under cloudstack control. Only 
> one site can be under cs control.
>
> -----Original Message-----
> From: Erik Weber [mailto:terbol...@gmail.com]
> Sent: Thursday, May 22, 2014 4:23 AM
> To: dev
> Subject: Re: VPC Site to Site VPN CIDR RFC1918
>
> The documentation says something else, excerpt:
> " The difference from Remote VPN is that Site-to-site VPNs connects entire 
> networks to each other, for example, connecting a branch office network to a 
> company headquarters network. In a site-to-site VPN, hosts do not have VPN 
> client software; they send and receive normal TCP/IP traffic through a VPN 
> gateway.".
>
> Assuming that both sides is under cloudstack control is just odd and makes no 
> sense.
>
> I'll file a ticket whenever I find the time, but still appreciate input :-)
>
> Erik Weber
> 22. mai 2014 00:16 skrev "Daan Hoogland" <daan.hoogl...@gmail.com> følgende:
>
>> I guess you shouldn't use the site to site vpn but just a vpn. I am
>> not sure how to configure that but you should just create an active
>> (client side) vpn to the external network. I see no reason why it
>> should't work. the site to site assumes you have both sides in
>> cloudstack and thus with rfc1918 networks.
>>
>> On Wed, May 21, 2014 at 6:10 PM, Erik Weber <terbol...@gmail.com> wrote:
>> > Site to site vpn.
>> >
>> > I'm not in control of the 50.0.1 network, but the client is.
>> >
>> > Basically the use case is that they want to secure the traffic to
>> > their cloud vms, and are fortunate enough to not have to use rfc1918
>> > on their network.
>> >
>> > I guess we could work around it by terminating the vpn on our
>> > equipment
>> and
>> > access it as a private gateway instead, but I'd prefer to use the
>> > technology that we so braverly tell our clients about.
>> >
>> > Erik
>> > 21. mai 2014 17:05 skrev "Daan Hoogland" <daan.hoogl...@gmail.com>
>> følgende:
>> >
>> >> Are you creating a site to site vpn or connecting to an external
>> network?
>> >>
>> >> On Wed, May 21, 2014 at 5:02 PM, Daan Hoogland
>> >> <daan.hoogl...@gmail.com
>> >
>> >> wrote:
>> >> > Erik, If it doesn't work it is probably been blocked on purpose
>> >> > but I don't see why it is. I don't know your use case either and
>> >> > it seems an unlikely one. But if the 50.0.1 net is out of your
>> >> > control you maybe should be able to configure this. So I would
>> >> > say it is a bug/lack of feature. I'll look into the code for the place 
>> >> > the error is generated.
>> >> >
>> >> > in short: file a ticket
>> >> >
>> >> > On Wed, May 21, 2014 at 2:34 PM, Erik Weber <terbol...@gmail.com>
>> wrote:
>> >> >> I understand that, but what my client wants is to connect public
>> >> >> ips instead of rfc1918 on one of the sides.
>> >> >>
>> >> >> e.g. one network has 10.0.1.0/24 and ip 1.2.3.4 the other has
>> >> >> 50.0.1.0/24 and ip 50.0.0.1
>> >> >>
>> >> >> but cloudstack currently does not let you do that, because it
>> >> >> expects
>> >> cidrs
>> >> >> to be rfc1918. see log excerpt:
>> >> >>
>> >> >> 2014-05-21 12:30:42,326 WARN  [c.c.u.n.NetUtils]
>> >> >> (API-Job-Executor-7:job-3072 ctx-bf3922b1) cidr 50.0.1.0/24 is
>> >> >> not
>> RFC
>> >> 1918
>> >> >> compliant
>> >> >> 2014-05-21 12:30:42,335 ERROR [c.c.a.ApiAsyncJobDispatcher]
>> >> >> (API-Job-Executor-7:job-3072) Unexpected exception while
>> >> >> executing
>> >> >>
>> org.apache.cloudstack.api.command.user.vpn.CreateVpnCustomerGatewayCmd
>> >> >> com.cloud.exception.InvalidParameterValueException: The customer
>> gateway
>> >> >> guest cidr list 50.0.1.0/24 is invalid guest cidr!
>> >> >> at
>> >> >>
>> >>
>> com.cloud.network.vpn.Site2SiteVpnManagerImpl.createCustomerGateway(Si
>> te2SiteVpnManagerImpl.java:176)
>> >> >>
>> >> >> I'm wondering if this is a bug/lacking feature, or intended.
>> >> >> As I initially said I'm not a network guy, so there might be
>> perfectly
>> >> good
>> >> >> reasons this shouldn't be allowed.
>> >> >>
>> >> >> But if it's a bug/lacking feature it would be great to know so
>> >> >> that I
>> >> could
>> >> >> file a ticket for it.
>> >> >>
>> >> >> --
>> >> >> Erik Weber
>> >> >>
>> >> >>
>> >> >> On Wed, May 21, 2014 at 2:09 PM, Daan Hoogland <
>> daan.hoogl...@gmail.com
>> >> >wrote:
>> >> >>
>> >> >>> Erik,
>> >> >>>
>> >> >>> The vpn let's you connect to all the computers in the network
>> >> >>> on the other site on their private adresses. This means that
>> >> >>> you can give
>> the
>> >> >>> cidr of the remote network in the definition on vpn connection.
>> >> >>>
>> >> >>> one network has 10.0.1.0/24 and ip 1.2.3.4 the other has
>> >> >>> 10.0.2.0/24 and ip 4.3.2.1
>> >> >>>
>> >> >>> on the first you define endpoint/gateway 4.3.2.1 with cidr
>> 10.0.1.0/24
>> >> >>> and you make it passive
>> >> >>> on the second you define the adresses of the first and stat is
>> without
>> >> >>> the passive function
>> >> >>> now you can ping a machine with address 10.0.1.123 from a
>> >> >>> machine
>> with
>> >> >>> ip 10.0.2.246
>> >> >>>
>> >> >>> Of course you can do this to an external network as well, which
>> makes
>> >> >>> far more sense.
>> >> >>>
>> >> >>> On Wed, May 21, 2014 at 12:14 PM, Erik Weber
>> >> >>> <terbol...@gmail.com>
>> >> wrote:
>> >> >>> >
>> >> >>>
>> >>
>> http://cloudstack.apache.org/docs/en-US/Apache_CloudStack/4.2.0/html/I
>> nstallation_Guide/vpn.html#site-to-site-vpnstates
>> >> >>> :
>> >> >>> >
>> >> >>> >
>> >> >>> >    - *CIDR list*: The guest CIDR list of the remote subnets.
>> Enter a
>> >> CIDR
>> >> >>> >    or a comma-separated list of CIDRs. Ensure that a guest
>> >> >>> > CIDR
>> list
>> >> is
>> >> >>> not
>> >> >>> >    overlapped with the VPC’s CIDR, or another guest CIDR. The
>> >> >>> > CIDR
>> >> must
>> >> >>> be
>> >> >>> >    RFC1918-compliant.
>> >> >>> >
>> >> >>> >
>> >> >>> > I'm not a network guy, so excuse the question if it's
>> >> >>> > obvious, but
>> >> if a
>> >> >>> > customer only has public ip's on their end, why is rfc1918
>> required?
>> >> >>> >
>> >> >>> >
>> >> >>> > --
>> >> >>> > Erik Weber
>> >> >>>
>> >> >>>
>> >> >>>
>> >> >>> --
>> >> >>> Daan
>> >> >>>
>> >> >
>> >> >
>> >> >
>> >> > --
>> >> > Daan
>> >>
>> >>
>> >>
>> >> --
>> >> Daan
>> >>
>>
>>
>>
>> --
>> Daan
>>



-- 
Daan

Reply via email to