[Kea-users] HA based on galera

[Klaus Steden]

In our case, our MySQL server is basically DBaaS. There was a Meetup
presentation about it (https://www.meetup.com/la-mysql/events/228737163/). If
you want, I can try to track it down -- the actual MySQL step isn't
something I had to engineer myself -- so I was able to take HA for granted.

In our use case, different scopes are handled by different individual Kea
servers (as the network is naturally partitioned), so avoiding write
conflicts is easy, but ultimately that's not the complicated part of our
configuration.

cheers,
Klaus

On Wed, Mar 14, 2018 at 10:10 AM, Jason Guy <jguy at cumulusnetworks.com>
wrote:

> Hey Klaus,
>
> Can you elaborate a bit on your design? I am in the middle of testing Kea
> 1.3 with a 3 node mysql setup, and have not seen any issues yet. I
> researched both MySQL-Cluster and MySQL Group Replication in MySQL5.7. I
> settled on first testing with group replication, since it did not require
> an extra management server arbitrating the writes.
>
> The group replication HA allows all devices to be a master, and uses GTIDs
> to prevent 2 nodes from writing the same row. So my curiosity here is if I
> am missing something in the design, and how I would easily stress test
> this, to see if the SQL group can handle the distributed writes from the
> Kea server.
>
> Thanks,
> Jason
>
> On Tue, Mar 13, 2018 at 8:17 PM, Klaus Steden <klausfiend at gmail.com>
> wrote:
>
>>
>> We have HA-MySQL behind our Kea servers, but the scopes themselves are
>> sharded to avoid this kind of write contention. In our case, we've got a
>> network partition that serves as an effective bright line between who's
>> responsible for what.
>>
>> cheers
>> Klaus
>>
>> On Tue, Mar 13, 2018 at 4:18 AM, <lex at ezelink.com> wrote:
>>
>>> Hey, pal, that's my thread ))
>>>
>>> if you installed boost and configure can't find it - use ./configure
>>> --help
>>> there should be option --with-boost-source or smth like that
>>>
>>>
>>>
>>> On 2018-03-13 15:10, Tobi Obadiah wrote:
>>>
>>>> Hello there,
>>>>
>>>> Am just getting started with Kea, on my attempt to install kea on
>>>> Ubuntu16.04 desktop  I keep getting the "Missing required header
>>>> files" errors
>>>> I have to manage to fix some of them but wonder if there is a standard
>>>> procedure to get this fix.  Though I have boost installed
>>>> "boost/shared_ptr.hpp"
>>>> is reported missing.
>>>>
>>>> I appreciate your help.
>>>>
>>>> On Tue, Mar 13, 2018 at 11:37 AM, Dominik Epple
>>>> <dominik.epple at gmail.com> wrote:
>>>>
>>>> Hi,
>>>>>
>>>>> it might help if you pick a dedicated galera write node, i.e. both
>>>>> of your kea servers write to the same galera node at each point in
>>>>> time. You can still configure that in a HA fashion by some
>>>>> active-passive loadbalancer or a failover IP or something.
>>>>>
>>>>> The other option would be that the kea code gets adjusted for
>>>>> running "galera-aware" but I am no expert on what exactly needs to
>>>>> happen there. But if it can't be fixed on application level, it
>>>>> probably can be fixed on system level by picking a dedicated write
>>>>> node.
>>>>>
>>>>> Dominik
>>>>>
>>>>> 2018-03-13 11:28 GMT+01:00 <lex at ezelink.com>:
>>>>>
>>>>> Hi guys,
>>>>>>
>>>>>> I'm running 2 kea dhcp servers and connect them to galera-mysql
>>>>>> cluster for HA.
>>>>>> Everything is ok, but sometimes i have this message in logs:
>>>>>>
>>>>>> 2018-03-13 14:14:45.036 ERROR [kea-dhcp4.alloc-engine/11864]
>>>>>> ALLOC_ENGINE_V4_ALLOC_ERROR [hwtype=1 b8:53:ac:xx:xx:xx],
>>>>>> cid=[01:b8:53:ac:xx:xx:xx], tid=0xa56d03a4: error during attempt
>>>>>> to allocate an IPv4 address: unable to execute for <SELECT
>>>>>> address, hwaddr, client_id, valid_lifetime, expire, subnet_id,
>>>>>> fqdn_fwd, fqdn_rev, hostname, state FROM lease4 WHERE client_id =
>>>>>> ? AND subnet_id = ?>, reason: Deadlock found when trying to get
>>>>>> lock; try restarting transaction (error code 1213)
>>>>>>
>>>>>> the network is relatively small - not more than 50-100 requests
>>>>>> per second
>>>>>>
>>>>>> How can I get rid of these messages? Cause after getting this
>>>>>> error, server sends NACK to client
>>>>>> _______________________________________________
>>>>>> Kea-users mailing list
>>>>>> Kea-users at lists.isc.org
>>>>>> https://lists.isc.org/mailman/listinfo/kea-users [1]
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Kea-users mailing list
>>>>> Kea-users at lists.isc.org
>>>>> https://lists.isc.org/mailman/listinfo/kea-users [1]
>>>>>
>>>>
>>>>
>>>>
>>>> Links:
>>>> ------
>>>> [1] https://lists.isc.org/mailman/listinfo/kea-users
>>>>
>>> _______________________________________________
>>> Kea-users mailing list
>>> Kea-users at lists.isc.org
>>> https://lists.isc.org/mailman/listinfo/kea-users
>>>
>>
>>
>> _______________________________________________
>> Kea-users mailing list
>> Kea-users at lists.isc.org
>> https://lists.isc.org/mailman/listinfo/kea-users
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/kea-users/attachments/20180315/924e7375/attachment.htm>