[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Broker and proxy redundancy



I'm trying to set up a fully redundant cluster, with at least 2 broker server.
I've seen an advice on your forum where we should create 2 broker and
then a dns entry pointing to both.
But with DNS we will have a round robin with no failover, if one of
both node will fail, half dns queries will fail.

Which kind of data are stored on the broker server? Can we use an
ha-proxy in front of both to archieve load balancing and redundancy?
data between brokers should be syncronized? (rsync, drbd, glusterfs,
or similar)

I don't know if broker is just a REST API endpoint that will execute
command directly on nodes, or if it save something on a local database
that should be replicated on other servers.

- In case of a scaled app, the git repository, is scaled as well?
Users can connect to any gears with git in case the primary gear is
failed?
- Is possibile to scale the haproxy gear?


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]