[ClusterLabs] Fencing | Dummy questions

[Klaus Wenninger]

On 03/13/2019 10:56 AM, Lopez, Francisco Javier [Global IT] wrote:
> Hello guys !
>
> I'm dealing since some time ago with this configuration:
>
> - Two node cluster.
> - Vmware boxes.
> - PostgreSql release 10.X: Master/Slave.
>
> On top of this I've set up Pacemaker/Corosync and RA/PAF.
>
> Now it's time to play with different fencing scenarios.
> I'd like to know from more experienced people about this. Based on the
> configuration I have (or similar)
> perhaps some of you can tell me, based on your experience, what kind
> of fencing sounds more appropriate,
> either node or resource level and based on what.
>
> Any quick reference would help me a lot.

Vmware being your hypervisor you can go
for fence_vmware_rest or fence_vmware_soap
as fence-agents.

That might depend on your distribution and
which fence-agent-styles are preferred/supported.

If you don't want fencing talking to your hypervisor
for whatever reason you might think of sparing a
shared disk (outside of your hypervisor or provided
by the hypervisor) and going with sbd (Sbd needs
a shared disk for useful deployments of 2-node clusters.
If you spare more virtual-nodes or deploy qdevice
that gives you 'real quorum' and sbd can be used
without a shared disk.).
As sbd is relying on a watchdog and Vmware doesn't
provide one you would have to go with softdog (software
emulation of a watchdog at least not relying on a
working userland) leaving you with somewhat reduced
reliability of the solution.

Depending on how your data is stored you might
in addition consider resource-fencing on the storage.

I'm aware that this doesn't give you an immediate solution
but I hope you'll find a few pointers to be able to
proceed.

Klaus
>
> I reviewed the threads and did not find any specific to this. I'd like
> to have this information to better
> understand what agents to install and manage.
>
> Thank you
> Javier
>
>
>
> Francisco Javier​ 	  	Lopez
>
> IT System Engineer 	 |  	Global IT
>
> O: *+34 619 728 249* <tel:+34%20619%20728%20249> 	 | 
> M: *+34 619 728 249* <tel:+34%20619%20728%20249> 	 | 
> 	*franciscojavier.lopez at solera.com*
> <mailto:franciscojavier.lopez at solera.com> 	 |  	*Solera.com*
> <https://www.solera.com/>
>
> Audatex Datos, S.A. 	 | 
> Avda. de Bruselas, 36, Salida 16, A‑1 (Diversia) 	,  	Alcobendas 	, 
> Madrid 	,  	28108 	,  	Spain
>
>  
>
>
> ------------------------------------------------------------------------
>
> " Este e-mail y sus archivos adjuntos son confidenciales y están
> dirigidos exclusivamente a la(s) persona(s) destinataria prevista. Si
> ha recibido este mensaje por error, por favor, notifique
> inmediatamente al remitente y elimine este mensaje. La empresa no
> firma contratos por e-mail y todas las negociaciones están sujetas a
> la firma de un contrato por escrito.
>
> This e-mail and any attached files are confidential and intended for
> the named addressee(s) only. If you have received this message in
> error, please notify the sender and delete the email immediately. The
> company does not conclude contracts by email and all negotiations are
> subject to written contract. "
>
>
> _______________________________________________
> Users mailing list: Users at clusterlabs.org
> https://lists.clusterlabs.org/mailman/listinfo/users
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20190313/40e8d5cc/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image309044.png
Type: image/png
Size: 8543 bytes
Desc: not available
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20190313/40e8d5cc/attachment-0002.png>