You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@cloudstack.apache.org by Richard Lawley <ri...@richardlawley.com> on 2019/06/17 12:48:46 UTC
Redundant VR Guest IPs
Hi,
When using RVRs, the guest VMs still see the VR IP instead of the VIP
for a number of things:
* DHCP Server
* Static NAT source IP for hairpin NAT
Just wondering what the reason is for this, as it causes a number of issues:
* Password reset doesn't work if active VR has changed since VM boot
* Issues with DHCP renewals if VR has changed since VM boot (will discuss later)
* Some static NAT connections confusingly (for the end user) come from
the VR IP rather than the VIP
It seems to me that it would be better if anything referencing the VR
IP actually used the VIP, but I'm assuming someone has made a decision
not to do this at some point - just wondered what that reason is, and
whether it's still valid.
Regards,
Richard
Re: Redundant VR Guest IPs
Posted by Rohit Yadav <ro...@shapeblue.com>.
Hi Wei,
Great, please send a PR on master.
Regards.
Regards,
Rohit Yadav
________________________________
From: Wei ZHOU <us...@gmail.com>
Sent: Tuesday, June 18, 2019 12:54:49 PM
To: dev@cloudstack.apache.org
Subject: Re: Redundant VR Guest IPs
Hi Rohit,
We have some changes to use gateway IP for
userdata/metadata/dhcp/dns/password server in cloudstack 4.7.1
I can port these changes to 4.11 or 4.13 if possible.
-Wei
Rohit Yadav <ro...@shapeblue.com> 于2019年6月17日周一 下午3:02写道:
> Hi Richard,
>
>
> You've probably found an implementational bug. I think the VIP should be
> used for both source NAT, password, user-data and dhcp/dns services.
>
>
> Regards,
>
> Rohit Yadav
>
> Software Architect, ShapeBlue
>
> https://www.shapeblue.com
>
> ________________________________
> From: Richard Lawley <ri...@richardlawley.com>
> Sent: Monday, June 17, 2019 6:18:46 PM
> To: dev@cloudstack.apache.org
> Subject: Redundant VR Guest IPs
>
> Hi,
>
> When using RVRs, the guest VMs still see the VR IP instead of the VIP
> for a number of things:
> * DHCP Server
> * Static NAT source IP for hairpin NAT
>
> Just wondering what the reason is for this, as it causes a number of
> issues:
> * Password reset doesn't work if active VR has changed since VM boot
> * Issues with DHCP renewals if VR has changed since VM boot (will discuss
> later)
> * Some static NAT connections confusingly (for the end user) come from
> the VR IP rather than the VIP
>
> It seems to me that it would be better if anything referencing the VR
> IP actually used the VIP, but I'm assuming someone has made a decision
> not to do this at some point - just wondered what that reason is, and
> whether it's still valid.
>
> Regards,
>
> Richard
>
> rohit.yadav@shapeblue.com
> www.shapeblue.com<http://www.shapeblue.com>
> Amadeus House, Floral Street, London WC2E 9DPUK
> @shapeblue
>
>
>
>
rohit.yadav@shapeblue.com
www.shapeblue.com
Amadeus House, Floral Street, London WC2E 9DPUK
@shapeblue
Re: Redundant VR Guest IPs
Posted by Wei ZHOU <us...@gmail.com>.
Hi Rohit,
We have some changes to use gateway IP for
userdata/metadata/dhcp/dns/password server in cloudstack 4.7.1
I can port these changes to 4.11 or 4.13 if possible.
-Wei
Rohit Yadav <ro...@shapeblue.com> 于2019年6月17日周一 下午3:02写道:
> Hi Richard,
>
>
> You've probably found an implementational bug. I think the VIP should be
> used for both source NAT, password, user-data and dhcp/dns services.
>
>
> Regards,
>
> Rohit Yadav
>
> Software Architect, ShapeBlue
>
> https://www.shapeblue.com
>
> ________________________________
> From: Richard Lawley <ri...@richardlawley.com>
> Sent: Monday, June 17, 2019 6:18:46 PM
> To: dev@cloudstack.apache.org
> Subject: Redundant VR Guest IPs
>
> Hi,
>
> When using RVRs, the guest VMs still see the VR IP instead of the VIP
> for a number of things:
> * DHCP Server
> * Static NAT source IP for hairpin NAT
>
> Just wondering what the reason is for this, as it causes a number of
> issues:
> * Password reset doesn't work if active VR has changed since VM boot
> * Issues with DHCP renewals if VR has changed since VM boot (will discuss
> later)
> * Some static NAT connections confusingly (for the end user) come from
> the VR IP rather than the VIP
>
> It seems to me that it would be better if anything referencing the VR
> IP actually used the VIP, but I'm assuming someone has made a decision
> not to do this at some point - just wondered what that reason is, and
> whether it's still valid.
>
> Regards,
>
> Richard
>
> rohit.yadav@shapeblue.com
> www.shapeblue.com
> Amadeus House, Floral Street, London WC2E 9DPUK
> @shapeblue
>
>
>
>
Re: Redundant VR Guest IPs
Posted by Rohit Yadav <ro...@shapeblue.com>.
Hi Richard,
You've probably found an implementational bug. I think the VIP should be used for both source NAT, password, user-data and dhcp/dns services.
Regards,
Rohit Yadav
Software Architect, ShapeBlue
https://www.shapeblue.com
________________________________
From: Richard Lawley <ri...@richardlawley.com>
Sent: Monday, June 17, 2019 6:18:46 PM
To: dev@cloudstack.apache.org
Subject: Redundant VR Guest IPs
Hi,
When using RVRs, the guest VMs still see the VR IP instead of the VIP
for a number of things:
* DHCP Server
* Static NAT source IP for hairpin NAT
Just wondering what the reason is for this, as it causes a number of issues:
* Password reset doesn't work if active VR has changed since VM boot
* Issues with DHCP renewals if VR has changed since VM boot (will discuss later)
* Some static NAT connections confusingly (for the end user) come from
the VR IP rather than the VIP
It seems to me that it would be better if anything referencing the VR
IP actually used the VIP, but I'm assuming someone has made a decision
not to do this at some point - just wondered what that reason is, and
whether it's still valid.
Regards,
Richard
rohit.yadav@shapeblue.com
www.shapeblue.com
Amadeus House, Floral Street, London WC2E 9DPUK
@shapeblue