hi Louis.
Your response was very helpful.

thanks for all your help

best regards.
Sérgio

Louis Santillan <[email protected]> escreveu no dia quarta, 27/02/2019
à(s) 19:30:

> Sergio,
>
> Some customers have a previously optimized etcd cluster(s).  We can have
> OpenShift point at that cluster, if so desired.  That is the External Etcd
> scenario.  This can reduce significant load from the Master nodes.  The
> other scenario is when OpenShift installs an etcd cluster on the Master
> nodes.  I rarely hear about the former scenario anymore.
>
>
> As for Masters, the idea is that etcd quorum [0] (in the latter scenario
> above) is optimally achieved when we have `(n+1)/2` agreement.
> Essentially, what that means that we start with 3 members (2 needed to
> achieve quorum) and we should increase the cluster by 2 members to continue
> along that curve.  However, since etcd needs to sync it's DB across the
> Masters, we great greatly increase our network traffic requirements needed
> for syncing the DB between the masters for relatively little availability
> gain.
>
> [0] https://coreos.com/etcd/docs/latest/faq.html
>
> ___________________________________________________
>
> LOUIS P. SANTILLAN
>
> Architect, OPENSHIFT & DEVOPS
>
> Red Hat Consulting, <https://www.redhat.com/> Container and PaaS Practice
>
> [email protected]   M: 3236334854
> <https://red.ht/sig>
> TRIED. TESTED. TRUSTED. <https://redhat.com/trusted>
>
>
>
>
> On Wed, Feb 27, 2019 at 4:55 AM Sérgio Cascão <[email protected]> wrote:
>
>> hi Trevor, thanks for your response.
>>
>> in your link they talk about HA, and these  case  you have one etcd for
>> each master, if the one master goes down, you have always the etcd
>> available.
>>
>>
>>> It therefore provides an HA setup where losing a control plane instance
>>> or an etcd member has less impact and does not affect the cluster
>>> redundancy as much as the stacked HA topology.
>>
>>
>> But my question is more about performance, for example, i know that
>> inplementation you have more latency in acess to master. if you can show me
>> more vantages and advantages I would be grateful.
>>
>> best regards
>> Sergio
>>
>> W. Trevor King <[email protected]> escreveu no dia terça, 26/02/2019 à(s)
>> 22:30:
>>
>>> On Tue, Feb 26, 2019 at 2:26 PM Sérgio Cascão wrote:
>>> > i like know what the advantages between put etcd separated from
>>> masters?
>>>
>>> Some more docs around these choices in [1].
>>>
>>> Cheers,
>>> Trevor
>>>
>>> [1]: https://kubernetes.io/docs/setup/independent/ha-topology/
>>>
>> _______________________________________________
>> users mailing list
>> [email protected]
>> http://lists.openshift.redhat.com/openshiftmm/listinfo/users
>>
>
_______________________________________________
users mailing list
[email protected]
http://lists.openshift.redhat.com/openshiftmm/listinfo/users

Reply via email to