Hi Lalit,

I don't really understand what you mean by an MCF "node".  Can you
clarify?  If you are sharing one database instance, so that one set of
database tables is being managed by two entirely separate MCF instances,
that will only work if both MCF "nodes" use the same zookeeper instances,
and are effectively part of the same cluster.  Otherwise they will
interfere with one another.  You *can* use the same Postgresql server but
with two distinct database instances within it; that works fine.

But that is not the reason you would get 404 errors.  As for session
stickiness, all I can add at this point is that the MCF UI uses cookies and
JSP sessions, so you should plan accordingly.

Karl


On Fri, Aug 8, 2014 at 4:26 AM, lalit jangra <[email protected]>
wrote:

> Hi,
>
> I have configured a set of two nodes- node1 & node2 with MCF 1.5.1 with 6
> instances of Zookeeper and 3 ZK instances on each node with Postgresql DB,
> with both MCF nodes pointing to same DB instance.I can access both nodes
> individually using node1 & node 2 IP and i can see cluster is working fine
> with any change on node1 is reflected immediately on node2 with ZK in place.
>
>
> Now i added a load balancer on top of it which has its own IP and this
> load balancer takes care internally of both MCF nodes. Next i am trying to
> access this cluster with load balancer IP but i am not able to access any
> of MCF node and getting 404 error.
>
>
> Is it something i am missing here such as with Zookeeper or MCF? Or do i
> need to configure my cluster differently to make it working?
>
> Finally do we need sticky sessions here to make it work?
>
> Regards,
> Lalit.
>

Reply via email to