[DRBD-user] Number of nodes supported for DRBD

[Gábor Hernádi]

On Mon, Feb 25, 2019 at 1:09 PM Roman Szabados <Roman at szabados.sk> wrote:

> Dear All,
>
> Updated question: Is it possible to configure the "N" parameter supplied
> by linstor for --max-peers "N"?
>
> As far as i understand it now, linstor uses  "--max-peers 7" as an fixed
> argument, which substantialy limits the DRBD capability to handle up to 31
> peers.
> Thank you.
>
> Best regards
>
> Roman Szabados
>
> On 2/22/19 2:45 PM, Roman Szabados wrote:
>
> Dear All,
>
> In our use case, we need to have up to 16 peers sharing the same resource
> and snapshot, which is later used on all hypervisors to build the actual
> specific VM from the snapshot. But after trying to add more than the 8th
> node, i get this error message:
>
> root at dagobah:~# linstor r c alderaan felucia kessel vm-base-1804-system
> --storage-pool ssd-pool
> ERROR:
> Description:
>     *Resource on node bespin has insufficient peer slots to add another
> peer*
> Details:
>     Node(s): 'alderaan', 'felucia', 'kessel', Resource:
> 'vm-base-1804-system'
> Show reports:
>     linstor error-reports show 5C28F669-00000-000115
> root at dagobah:~#
>
> I do assume, that this is a limitation of Linstor. Can this limit be
> configured to accept more than 8 nodes per ressource in Linstr?
> Thank you.
>
> Roman Szabados
>
> On 2/21/19 10:39 AM, Roland Kammerer wrote:
>
> On Wed, Feb 20, 2019 at 02:35:16PM +0530, Malhar vora wrote:
>
> Hi Experts,
>
> Does anyone know how many nodes can be supported for DRBD ?
>
> Can DRBD support count like 1000+ nodes ?
>
> What Robert wote, plus:
>
> IMO the important thing here is "at the same time or not" and "1000
> times redundant"? The second one is easy, it does not make sense. For
> the first one: Usually you make resource available 2 or 3 times
> redundant (every bit is also on 1 or 2 different servers). And then you
> have like a ton of hypervisors.  But only one connects to the storrage
> at a given point in time. Or 2 because of live migration. These
> hypervisors would be diskless clients, that connect to the storage via
> network only. And if they don't need it anymore, another diskless client
> accesses the data. Most of our plugins that integrate DRBD/LINSTOR with
> other platforms like k8s, openstack, opennebula,... do exactly that:
> They create storage N-times redundant and create/delete diskless
> assignments on demand. So yes, conceptually you then can have 1000
> nodes.
>
> Regards, rck
> _______________________________________________
> drbd-user mailing listdrbd-user at lists.linbit.comhttp://lists.linbit.com/mailman/listinfo/drbd-user
>
>
> _______________________________________________
> drbd-user mailing listdrbd-user at lists.linbit.comhttp://lists.linbit.com/mailman/listinfo/drbd-user
>
> _______________________________________________
> drbd-user mailing list
> drbd-user at lists.linbit.com
> http://lists.linbit.com/mailman/listinfo/drbd-user
>


Hello,

for existing resources you cannot change this option as it would require a
recreation of the metadata.
For new resources, you can set a property on the corresponding
resource-definition:

linstor rd sp <resource-name> PeerSlotsNewResource 31

Regards,
Gabor
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20190225/dfd01468/attachment.htm>