Cloudera documentation identifies the steps for MySQL replication. I cannot find any information of how you set up high availabilty should the DB crash. Yes the data is replicated to another server but we are not pointing to that server.
I was told that Cloudera only supprorts the Db being Active/Passive. So how do you reference the "passive" if needed?
There are nine configurations for database hostname.
1. Cloudera Manager com.cloudera.cmf.db.host
2. Hue Database Hostname
3. Activity Monitor Database Hostname
4. Navigator Audit Server Database Hostname
5. Navigator Metadata Server Database Hostname
6. Reports Manager Database Hostname
7. Oozie Server Database Host
8. Hive Metastore Database Host
9. Sentry Server Database Host
We have tried to use various load balancers (single IP) to multiple DB servers, but have found this inconsistant. Sentry fails to start. We get DB timeouts as if connection are stale and not resubmitted when the DBs failover.
Is it expected that we manually have to change nine configuration points and do service restarts? Is this the concept of high availability?