> OK, so either patch PAF yourself (not recommended) or choose something
> else. Note that two other ways are working with Pacemaker:
> * the pgsql resource agent (see FAQ of PAF)
> * a shared disk architecture (no pgsql replication)
> * the blade itself, but it would fence all the container running on it
> * the access to the SAN from the failing container
> I don't know if fencing agent exists for a container itself. Note that I'm not
> familiar with the container world, I lack a lot of knowledge on this
> else. Note that two other ways are working with Pacemaker:
> * the pgsql resource agent (see FAQ of PAF)
> * a shared disk architecture (no pgsql replication)
Probably, i will be interested by the solution "patroni-etcd-haproxy" you suggested
> I suppose you could find fencing agents for:
> * the blade itself, but it would fence all the container running on it
> * the access to the SAN from the failing container
> I don't know if fencing agent exists for a container itself. Note that I'm not
> familiar with the container world, I lack a lot of knowledge on this
>
technology.
I am not familiar with this too. I heard that for the first time few days ago by reading how to improve HA ... :-)
Many thanks for your opinions/advices
Le mer. 5 sept. 2018 à 15:44, Jehan-Guillaume (ioguix) de Rorthais <ioguix@xxxxxxx> a écrit :
On Wed, 5 Sep 2018 15:06:21 +0200
Thomas Poty <thomas.poty@xxxxxxxxx> wrote:
> > In fact, PAF does not support slots. So it is not a good candidate if
> > slot are a requirement.
> Effectively slots are a requirement we prefer to keep
OK, so either patch PAF yourself (not recommended) or choose something
else. Note that two other ways are working with Pacemaker:
* the pgsql resource agent (see FAQ of PAF)
* a shared disk architecture (no pgsql replication)
> > > a proxy HAproxy and
> > > for fencincg, i am a bit disappointed, i don't know what to do/use
> > Depend on your hardware or your virtualization technology.
> Our production cluster (master and slave) runs on LXC container. Each LXC
> container runs on a HPE Blade Server. The storage is on a SAN 3PAR array.
> Any advice ?
I suppose you could find fencing agents for:
* the blade itself, but it would fence all the container running on it
* the access to the SAN from the failing container
I don't know if fencing agent exists for a container itself. Note that I'm not
familiar with the container world, I lack a lot of knowledge on this
technology.
++
