[Kea-users] High Availability

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

[Kea-users] High Availability

Marcos Renato da Silva Junior

Hello,

I understood that Kea still does not have failover or any other type of high availability (I have version 1.1.0 and use memfile).

Case 1: If I use only static lease (reservations) and keep two servers with the same configuration file, would it work as high availability ?.

Case 2: If I use a pool on each server, would it be a way of high availability? :
server1 = {"pool": "192.168.0.50 - 192.168.0.100"}
server2 = {"pool": "192.168.0.101 - 192.168.0.200"}

Thanks,

-- 
Marcos Renato da Silva Junior

_______________________________________________
Kea-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/kea-users
Reply | Threaded
Open this post in threaded view
|

Re: [Kea-users] High Availability

Sutherland, Robert

We, too, hope to attain high availability by

  • using Pacemaker/Corosync to monitor Kea (fronted with a floating IP address since we deal exclusively with dhcrelay)
  • using replicated postgres databases for leases

 

Note the complete lack of HA with respect to configuration files. We are hoping that the configuration can be stored in the database in a future release.

 

Rob

 

From: Kea-users <[hidden email]> On Behalf Of Marcos Renato da Silva Junior
Sent: Thursday, May 10, 2018 8:56 AM
To: [hidden email]
Subject: [Kea-users] High Availability

 

Hello,

I understood that Kea still does not have failover or any other type of high availability (I have version 1.1.0 and use memfile).

Case 1: If I use only static lease (reservations) and keep two servers with the same configuration file, would it work as high availability ?.

Case 2: If I use a pool on each server, would it be a way of high availability? :
server1 = {"pool": "192.168.0.50 - 192.168.0.100"}
server2 = {"pool": "192.168.0.101 - 192.168.0.200"}

Thanks,

-- 
Marcos Renato da Silva Junior
This email message and any attachments are for the sole use of the intended recipient(s). Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message and any attachments.

_______________________________________________
Kea-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/kea-users
Reply | Threaded
Open this post in threaded view
|

Re: [Kea-users] High Availability

Munroe Sollog
We currently use kea in an HA (slightly unsupported setup) and I believe 1.4 will include the rest of the missing pieces for HA (due in a few weeks).  We have 3 servers in a galera mysql cluster and two dhcp server.  We are double VRRP'ing between them so both IPs always stay up.  Using 1.3, while some metrics have been screwed up, the core functionality has been rock solid for ~6 months.

On Thu, May 10, 2018 at 9:24 AM Sutherland, Robert <[hidden email]> wrote:

We, too, hope to attain high availability by

  • using Pacemaker/Corosync to monitor Kea (fronted with a floating IP address since we deal exclusively with dhcrelay)
  • using replicated postgres databases for leases

 

Note the complete lack of HA with respect to configuration files. We are hoping that the configuration can be stored in the database in a future release.

 

Rob

 

From: Kea-users <[hidden email]> On Behalf Of Marcos Renato da Silva Junior
Sent: Thursday, May 10, 2018 8:56 AM
To: [hidden email]
Subject: [Kea-users] High Availability

 

Hello,

I understood that Kea still does not have failover or any other type of high availability (I have version 1.1.0 and use memfile).

Case 1: If I use only static lease (reservations) and keep two servers with the same configuration file, would it work as high availability ?.

Case 2: If I use a pool on each server, would it be a way of high availability? :
server1 = {"pool": "192.168.0.50 - 192.168.0.100"}
server2 = {"pool": "192.168.0.101 - 192.168.0.200"}

Thanks,

-- 
Marcos Renato da Silva Junior
This email message and any attachments are for the sole use of the intended recipient(s). Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message and any attachments.
_______________________________________________
Kea-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/kea-users


--
Munroe Sollog
Senior Network Engineer

_______________________________________________
Kea-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/kea-users
Reply | Threaded
Open this post in threaded view
|

Re: [Kea-users] High Availability

Tomek Mrugalski
On 10.05.2018 14:55, Marcos Renato da Silva Junior wrote:
> I understood that Kea still does not have failover or any other type
> of high availability (I have version 1.1.0 and use memfile).
1.1.0 is EOLed since Dec 2017. That means that if you encounter a bug,
you're on your own. ISC will not be fixing anything in versions that are
too old. If I were in your position, I'd seriously consider moving to
1.4.0 once it becomes available. It will have a HA support that works
with memfile.

On 10.05.2018 15:24, Sutherland, Robert wrote:
> We hoping that the configuration can be stored in the database in a
> future release.
Correct. That's currently under consideration for 1.5. If you are
interested in this functionality and are willing to help fund its
development and/or test it, please contact me off-line. We do have an
early description of the solution here:

http://kea.isc.org/wiki/CentralizedConfigNetconf

For various reasons we had to mix two aspects together: configuration
storage in DB and netconf. If you don't care about netconf, pay
attention only to the DB sections.

On 10.05.2018 15:28, Munroe Sollog wrote:
> We currently use kea in an HA (slightly unsupported setup) and I
> believe 1.4 will include the rest of the missing pieces for HA (due
> in a few weeks).  We have 3 servers in a galera mysql cluster and
> two dhcp server.  We are double VRRP'ing between them so both IPs
> always stay up.  Using 1.3, while some metrics have been screwed up,
> the core functionality has been rock solid for ~6 months.
Great to hear about the rock solid part.

The metric fix is coming up in 1.4-beta.

Tomek
_______________________________________________
Kea-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/kea-users
Reply | Threaded
Open this post in threaded view
|

Re: [Kea-users] High Availability

Joelson Vendramin
Hi,

We have a "home-made" HA with two servers running KEA 1.2.0 for about one year without any troubles.

However, in our scenario:
- We don't use any DBs, just memfiles;
- We have no dynamic pools, just reservations based on "circuit-id" (option 82) for IPv4 and "interface-id" (option 18) for IPv6.

For this use, both servers can have almost the same kea.conf file and there's no need to keep lease CSV files in sync.

--
Joelson Vendramin


Em quinta-feira, 10 de maio de 2018 10:45:58 BRT, Tomek Mrugalski <[hidden email]> escreveu:


On 10.05.2018 14:55, Marcos Renato da Silva Junior wrote:
> I understood that Kea still does not have failover or any other type
> of high availability (I have version 1.1.0 and use memfile).
1.1.0 is EOLed since Dec 2017. That means that if you encounter a bug,
you're on your own. ISC will not be fixing anything in versions that are
too old. If I were in your position, I'd seriously consider moving to
1.4.0 once it becomes available. It will have a HA support that works
with memfile.

On 10.05.2018 15:24, Sutherland, Robert wrote:
> We hoping that the configuration can be stored in the database in a
> future release.
Correct. That's currently under consideration for 1.5. If you are
interested in this functionality and are willing to help fund its
development and/or test it, please contact me off-line. We do have an
early description of the solution here:

http://kea.isc.org/wiki/CentralizedConfigNetconf

For various reasons we had to mix two aspects together: configuration
storage in DB and netconf. If you don't care about netconf, pay
attention only to the DB sections.

On 10.05.2018 15:28, Munroe Sollog wrote:
> We currently use kea in an HA (slightly unsupported setup) and I
> believe 1.4 will include the rest of the missing pieces for HA (due
> in a few weeks).  We have 3 servers in a galera mysql cluster and
> two dhcp server.  We are double VRRP'ing between them so both IPs
> always stay up.  Using 1.3, while some metrics have been screwed up,
> the core functionality has been rock solid for ~6 months.
Great to hear about the rock solid part.

The metric fix is coming up in 1.4-beta.

Tomek

_______________________________________________
Kea-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/kea-users

_______________________________________________
Kea-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/kea-users